14.11.2014 Views

An Introduction to the Ericsson Transport Network Architecture ...

An Introduction to the Ericsson Transport Network Architecture ...

An Introduction to the Ericsson Transport Network Architecture ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

ERICSSON<br />

REVIEW<br />

3<br />

1992<br />

<strong>An</strong> <strong>Introduction</strong> <strong>to</strong> <strong>the</strong> <strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong><br />

Control and Operation of SDH <strong>Network</strong> Elements<br />

AXD 4/1, a Digital Cross-Connect System


ERICSSON REVIEW<br />

Number 3 • 1992 • Volume 69<br />

Responsible publisher Bo Hedfors<br />

Edi<strong>to</strong>r Per-Olof Thyselius<br />

Edi<strong>to</strong>rial staff Martti Viitaniemi<br />

Subscription Eva Karlstein<br />

Subscription one year $30<br />

Address S-126 25 S<strong>to</strong>ckholm, Sweden<br />

Published in English with four issues per year<br />

Copyright Telefonaktiebolaget L M <strong>Ericsson</strong><br />

Contents<br />

58 • <strong>An</strong> <strong>Introduction</strong> <strong>to</strong> <strong>the</strong> <strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong><br />

62 • Control and Operation of SDH <strong>Network</strong> Elements<br />

78 • AXD 4/1, a Digital Cross-Connect System<br />

Cover<br />

<strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong> offers a<br />

complete network solution for a managable<br />

<strong>Transport</strong> <strong>Network</strong> based on <strong>the</strong> Synchronous<br />

Digital Hierarchy


<strong>An</strong> <strong>Introduction</strong> <strong>to</strong> <strong>the</strong> <strong>Ericsson</strong><br />

<strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong><br />

Stefan Danielsson<br />

The telecommunications industry has formulated a number of new standards for a<br />

new transmission hierarchy, <strong>the</strong> Synchronous Digital Hierarchy, for <strong>the</strong> purpose of<br />

reducing operating costs and improving <strong>the</strong> service offered <strong>to</strong> users. Today's fixed<br />

transmission network will evolve in<strong>to</strong> a managed and software-controlled <strong>Transport</strong><br />

<strong>Network</strong>. The <strong>Transport</strong> <strong>Network</strong> will form <strong>the</strong> infrastructure of <strong>the</strong> future telecom<br />

network and so make for successful introduction of new services, such as broadband<br />

data.<br />

The author describes <strong>the</strong> <strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong> (ETNA), a<br />

complete system solution for a manageable synchronous <strong>Transport</strong> <strong>Network</strong>.<br />

digital communication systems<br />

telecommunication networks<br />

telecommunication network management<br />

Today's digital transmission systems are<br />

based on <strong>the</strong> Plesiochronous Digital Hierarchy<br />

(PDH) standard, which was introduced<br />

step-by-step <strong>to</strong> cater for demands<br />

for higher transmission capacity in <strong>the</strong><br />

voice traffic network.<br />

This has resulted in a network that allows<br />

interconnection between different<br />

vendors' transmission equipment at certain<br />

standardised electrical interfaces, but<br />

also a network in which configuration<br />

changes are effected through hard-wiring<br />

and for which standards for interconnection<br />

at <strong>the</strong> optical level have been lacking.<br />

The complex network is difficult <strong>to</strong> manage<br />

and costly and time-consuming <strong>to</strong> adapt <strong>to</strong><br />

changing requirements for transmission<br />

capacity. It consists of a large number of<br />

multiplexers at different levels and volumes<br />

of cabling between <strong>the</strong>m. The line<br />

systems are underutilised; often, less than<br />

50 % of <strong>the</strong> capacity is used.<br />

Until now, this approach has been acceptable,<br />

because <strong>the</strong> dominating voice-only<br />

traffic is predictable <strong>to</strong> some degree. But<br />

<strong>to</strong>day, public network opera<strong>to</strong>rs around <strong>the</strong><br />

world are faced with increasing financial<br />

pressure, growing competition due <strong>to</strong> deregulated<br />

markets, and unprecedented<br />

demands from users who now view communications<br />

as a strategic business <strong>to</strong>ol.<br />

Business users, notably those requiring<br />

both data and voice communications, are<br />

seeking new levels of guaranteed quality<br />

and services. The time it takes <strong>to</strong> arrange<br />

leased-line services - several weeks,<br />

sometimes months - will no longer be <strong>to</strong>lerated.<br />

The network must provide bandwidth<br />

on demand.<br />

Synchronous Digital Hierarchy<br />

Recognising <strong>the</strong>se problems and demands,<br />

<strong>the</strong> telecommunications industry<br />

has formulated a number of new standards<br />

for a new transmission hierarchy, <strong>the</strong> Syn-<br />

Fig. 1<br />

<strong>Ericsson</strong>'s centre for transmission research and<br />

development at Kungens Kurva, S<strong>to</strong>ckholm,<br />

Sweden<br />

ERICSSON REVIEW No. 3, 1992


59<br />

STEFAN DANIELSSON<br />

<strong>Ericsson</strong> Telecom AB<br />

chronous Digital Hierarchy, aimed at easing<br />

<strong>the</strong> difficulties for opera<strong>to</strong>rs and improving<br />

<strong>the</strong> services offered <strong>to</strong> users. Included<br />

in SDH are standards for new transmission<br />

bit-rates, optical interfaces, information<br />

models and communication pro<strong>to</strong>cols<br />

for network management and proposals<br />

for network structures.<br />

Future <strong>Transport</strong> <strong>Network</strong>s<br />

SDH offers many advantages and forms<br />

<strong>the</strong> foundation for <strong>the</strong> future transport network.<br />

SDH also underlies <strong>the</strong> <strong>Ericsson</strong><br />

<strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong> (ETNA).<br />

ETNA is a single, open system concept<br />

that enables a public network opera<strong>to</strong>r <strong>to</strong><br />

optimise his use of existing resources and<br />

make a smooth migration <strong>to</strong>wards future<br />

broadband digital services.<br />

Included in ETNA are all <strong>the</strong> transmission<br />

links, switching, routing and management<br />

facilities needed <strong>to</strong> deliver wideband and<br />

broadband services - data, voice, image<br />

and video. <strong>Network</strong> management is of <strong>the</strong><br />

utmost importance. It is only through powerful<br />

network management that <strong>the</strong> cost<br />

and service benefits from SDH can be fully<br />

utilised.<br />

ETNA consists of a family of <strong>Network</strong> Elements<br />

and one common <strong>Network</strong> Management<br />

system, FMAS (Facility Management<br />

System).<br />

The <strong>Network</strong> Elements have basic functionality<br />

in common; <strong>the</strong>y terminate electric<br />

and optical signals, perform switching at<br />

various signal levels and are controlled<br />

from FMAS. Due <strong>to</strong> somewhat different applications<br />

and optimisation criteria, two<br />

product lines are defined: Digital Cross­<br />

Connect Systems (DXC) and SDH Transmission<br />

Systems (SMUX).<br />

Digital Cross-Connect Systems<br />

DXCs are transmission channel switches<br />

for semi-permanent connections. With <strong>to</strong>tally<br />

transparent switching characteristics,<br />

<strong>the</strong> DXC can terminate any PDH or SDH<br />

signal for selection and rerouting at any<br />

lower-order level. DXCs provide extensive<br />

switching capabilities for network res<strong>to</strong>ration<br />

and network configuration in central<br />

hubs with heavy concentrations of circuits.<br />

SDH Transmission Systems<br />

SDH transmission systems include a<br />

range of terminal multiplexers, intermediate<br />

regenera<strong>to</strong>rs and add/drop multiplexers<br />

based on SDH standards for transmission<br />

at 155 Mbit/s, 620 Mbit/s and<br />

2.5 Gbit/s. The systems are built from a<br />

common set of modules <strong>to</strong> reduce <strong>the</strong><br />

s<strong>to</strong>ckholding of spares and <strong>to</strong> simplify capacity<br />

upgrades. SMUXs are used in point<strong>to</strong>-point,<br />

bus or ring configurations and<br />

provide a distributed type of network configuration<br />

with line or ring protection for network<br />

res<strong>to</strong>ration.<br />

Fig. 2<br />

ETNA supports a layered-architecture approach<br />

<strong>to</strong> <strong>Transport</strong> <strong>Network</strong> configurations<br />

AXD 4/1 (DXC)<br />

AXD 1/0 (DXC)<br />

Add/Drop Multiplexer (SMUX)<br />

AXD 2500 (SMUX)<br />

AXD 620 (SMUX)<br />

ACXD 155 (SMUX)


60<br />

Fig. 3<br />

AXD 155 system integration and verification test<br />

carried out at Kungens Kurva<br />

Fig 4<br />

<strong>An</strong> example of <strong>the</strong> user interface of FMAS in<br />

Configuration Management mode<br />

Facility Management System<br />

The FMAS provides a single system for <strong>the</strong><br />

management of <strong>the</strong> complete transport<br />

network, including DXCs, SMUXs and<br />

PDH systems.<br />

The <strong>Transport</strong> <strong>Network</strong><br />

System<br />

With <strong>the</strong> introduction of ETNA, <strong>the</strong> telecom<br />

network and its operation will change drastically.<br />

Today's fixed transmission network<br />

will evolve in<strong>to</strong> a managed and softwarecontrolled<br />

network which, with its transmission<br />

resources and operation and control<br />

capabilities, is defined as <strong>the</strong> <strong>Transport</strong><br />

<strong>Network</strong>.<br />

The <strong>Transport</strong> <strong>Network</strong> will reduce <strong>the</strong><br />

costs for operation and maintenance of<br />

transmission resources, but will also form<br />

<strong>the</strong> infrastructure for <strong>the</strong> future telecom<br />

network and make for successful introduction<br />

of new services, such as broadband<br />

data.<br />

How can network opera<strong>to</strong>rs make sure that<br />

<strong>the</strong>ir transport network will reduce operating<br />

costs in <strong>the</strong> short-time perspective,<br />

while at <strong>the</strong> same time meeting <strong>the</strong> demands<br />

of <strong>the</strong> future? The answer is complete<br />

network solutions.<br />

By providing a complete family of complementary<br />

network elements, each one optimal<br />

for a specific application in <strong>the</strong> network<br />

and each one interworking with one<br />

common management system, functionality<br />

can be introduced at network level. End<strong>to</strong>-end<br />

performance moni<strong>to</strong>ring, au<strong>to</strong>matic<br />

bandwidth provision and network<br />

res<strong>to</strong>ration can be made available pending<br />

complete TMN standards.<br />

Fur<strong>the</strong>rmore, <strong>the</strong> system solution will<br />

create a platform which can be upgraded<br />

with new functionality <strong>to</strong> follow <strong>the</strong> evolvement<br />

of TMN standards and <strong>to</strong> meet demands<br />

for new features. In this way, a<br />

future-proof network solution can be<br />

achieved.<br />

ETNA is a <strong>Transport</strong> <strong>Network</strong> system solution<br />

and a platform for future network development.<br />

Included in ETNA are all <strong>Network</strong><br />

Elements and <strong>the</strong> <strong>Network</strong> Management<br />

system required <strong>to</strong> form all types of<br />

network structure and <strong>to</strong> meet <strong>the</strong> demand<br />

for software-controlled allocation and<br />

moni<strong>to</strong>ring of bandwidth in a principally<br />

self-healing network.<br />

The <strong>Network</strong> Elements are not only similar<br />

in terms of functionality but also have<br />

commonalities in design, thus contributing<br />

<strong>to</strong> <strong>the</strong> establishment of a platform for future<br />

development of new functionality and<br />

new network elements. Design commonalities<br />

have been achieved at all levels:<br />

Use of <strong>the</strong> same interface <strong>to</strong> FMAS<br />

TMN standard interfaces (Q) are used<br />

between all network elements and <strong>the</strong><br />

common network management system,<br />

FMAS.<br />

Use of <strong>the</strong> same information models<br />

All network elements are based on <strong>the</strong><br />

same information models for configuration<br />

management, fault management, performance<br />

management and security management.<br />

ERICSSON REVIEW No. 3, 1992


61<br />

References<br />

1 Bergendahl, J. and Ekelund, S.: <strong>Transport</strong><br />

network development. <strong>Ericsson</strong> Review<br />

67(1990):2, pp. 54-59.<br />

2 Breuer, H.-J. and Hellstrom, B.: Synchronous<br />

Transmission <strong>Network</strong>s. <strong>Ericsson</strong><br />

Review 67(1990):2, pp. 60-71.<br />

3 <strong>An</strong>dersson, J.-O.-.Digital Cross-Connect<br />

Systems-a System Family for <strong>the</strong> <strong>Transport</strong><br />

<strong>Network</strong>. <strong>Ericsson</strong> Review 67<br />

(1990):2, pp. 72-83.<br />

4 Tarle, H..FMAS - an Operations Support<br />

System for <strong>Transport</strong> <strong>Network</strong>s. <strong>Ericsson</strong><br />

Review 67(1990):4, pp. 163-182.<br />

Use of <strong>the</strong> same man-machine interface<br />

When using a local-craft interface at a DXC<br />

or SMUX, <strong>the</strong> opera<strong>to</strong>r will see displayed<br />

<strong>the</strong> same types of symbol and use <strong>the</strong><br />

same types of command, independent of<br />

equipment.<br />

Use of common hardware modules<br />

Common internal interfaces have been<br />

used for SMUX and DXC <strong>to</strong> ensure full<br />

compatibility of different access circuit<br />

boards within SMUXs and DXCs respectively.<br />

Such compatibility of circuit boards<br />

also exists across <strong>the</strong> two product lines.<br />

Use of <strong>the</strong> same packaging system<br />

The same equipment practice, including<br />

power distribution and alarm panels, can<br />

be used for DXCs and SMUXs.<br />

Use of <strong>the</strong> same documentation structure<br />

A standardised documentation structure<br />

will be used <strong>to</strong> facilitate administration, operation<br />

and maintenance of <strong>the</strong> products.<br />

The Open <strong>Transport</strong> <strong>Network</strong><br />

solution<br />

To provide a system that can ensure multivendor<br />

compatibility is equally important<br />

as providing a complete and future-proof<br />

network solution. <strong>An</strong> absolute prerequisite<br />

is that <strong>the</strong> benefits from ETNA are achievable<br />

in a network supplied by several<br />

vendors. This has been solved through <strong>the</strong><br />

choice of an open structure for ETNA. All<br />

existing TMN standards concerning Q-interfaces,<br />

embedded communication channels<br />

and information models will be supported<br />

from <strong>the</strong> first release of any ETNA<br />

product.<br />

Since <strong>the</strong> TMN has not yet been completed,<br />

it is necessary <strong>to</strong> include <strong>Ericsson</strong> specific<br />

additions. The TMN support is implemented<br />

in structured object-oriented<br />

software and it will thus be possible <strong>to</strong> follow<br />

<strong>the</strong> evolvement of standards through<br />

regular releases of software updates.<br />

Pending complete TMN standards, adaptations<br />

will be required for multi-vendor<br />

interoperability. Through <strong>the</strong> open and<br />

structured design of FMAS, o<strong>the</strong>r vendors'<br />

equipment can be handled by implementing<br />

<strong>the</strong> corresponding information model<br />

in one of <strong>the</strong> FMAS layers. Above this layer<br />

is <strong>the</strong> application layer, which means that<br />

<strong>the</strong> vendor-specific differences of <strong>the</strong> network<br />

element are hidden: all network elements<br />

can be handled in <strong>the</strong> same manner.<br />

<strong>Introduction</strong> <strong>to</strong> <strong>Network</strong><br />

Element descriptions<br />

Previous articles in <strong>Ericsson</strong> Review (No.2<br />

and No.4 1990) describe <strong>the</strong> development<br />

of <strong>the</strong> <strong>Transport</strong> <strong>Network</strong>, <strong>the</strong> Synchronous<br />

Digital Hierarchy, Digital Cross-Connects,<br />

and FMAS.<br />

This edition of <strong>Ericsson</strong> Review goes one<br />

step fur<strong>the</strong>r. Two articles exemplify <strong>the</strong> realisation<br />

of individual <strong>Network</strong> Elements,<br />

and <strong>the</strong>ir Control System.<br />

The design of <strong>the</strong> Digital Cross-Connect<br />

AXD 4/1 is described in one article. This<br />

network element is mainly used for configuration<br />

of <strong>the</strong> network in central nodes. The<br />

SDH Transmission System AXD 155, ideal<br />

for <strong>the</strong> construction of ring configurations<br />

in <strong>the</strong> access or local network, will be described<br />

in a coming issue of <strong>Ericsson</strong> Review.<br />

Solutions for <strong>the</strong> local Control System and<br />

<strong>the</strong> communication with FMAS is described<br />

in one article. The two <strong>Network</strong><br />

Elements perform different tasks in <strong>the</strong><br />

network, a condition which has influenced<br />

<strong>the</strong> design of <strong>the</strong> local Control System.<br />

However, what is visible <strong>to</strong> <strong>the</strong> opera<strong>to</strong>r is<br />

identical. The support for TMN standards<br />

regarding interfaces, information models<br />

and communication channels is also identical.<br />

The result is a family of network elements,<br />

optimal for <strong>the</strong>ir specific task in <strong>the</strong> network<br />

and simultaneously administrated, operated<br />

and maintained in an identical way.<br />

ERICSSON REVIEW No. 3, 1992


Control and Operation of SDH<br />

<strong>Network</strong> Elements<br />

Johan Blume, Leif Hansson, Peder Hagg and Leif Sundin<br />

own management system with proprietary<br />

Increasingly, flexible and powerful network management solutions allowing <strong>Network</strong><br />

interfaces and functionality. This situation<br />

Elements and Operations Systems <strong>to</strong> work in a multi-vendor environment are often necessitates adaptations when new<br />

becoming a key issue <strong>to</strong> network opera<strong>to</strong>rs. The Synchronous Digital Hierarchy equipment is introduced on a market,<br />

currently being standardised provides <strong>the</strong> required capabilities.<br />

which is costly in terms of time, resources<br />

The authors describe <strong>the</strong> main characteristics of SDH management and discuss and money, both for <strong>the</strong> supplier and <strong>the</strong><br />

some implementation aspects of control systems developed for <strong>Ericsson</strong>'s SDH opera<strong>to</strong>r.<br />

<strong>Network</strong> Elements.<br />

digital communication systems<br />

telecommunication networks<br />

telecommunication network management<br />

Background and Driving<br />

Forces<br />

Today's transmission network typically<br />

consists of inflexible equipment without<br />

provision for remote reconfiguration, and<br />

fixed hard-wired point-<strong>to</strong>-point connections.<br />

This means that each change of configuration<br />

- when supplying a 2 Mbit/s<br />

leased line, for example - requires hardwiring,<br />

which is time-consuming and<br />

<strong>the</strong>refore costly.<br />

The Synchronous Digital Hierarchy (SDH)<br />

eliminates <strong>the</strong>se disadvantages by providing<br />

flexible <strong>Network</strong> Elements (NE) capable<br />

of being configured remotely. This<br />

makes it possible <strong>to</strong> provide new broadband<br />

services - such as 2 Mbit/s leased<br />

lines - <strong>to</strong> cus<strong>to</strong>mers in a short time and at<br />

low cost.<br />

One of <strong>the</strong> major driving forces behind<br />

SDH is improved and standardised management<br />

interfaces and functions, allowing<br />

<strong>the</strong> SDH <strong>Network</strong> Elements and Operations<br />

Systems (OS) <strong>to</strong> interwork in a<br />

multi-vendor environment. 23<br />

Benefits of SDH and ETNA<br />

Functional overview<br />

Introducing SDH in <strong>the</strong> transport network<br />

will improve operation and maintenance,<br />

and so reduce <strong>the</strong> operational cost for <strong>the</strong><br />

Telecom opera<strong>to</strong>r. SDH also enables <strong>the</strong><br />

opera<strong>to</strong>r <strong>to</strong> control <strong>the</strong> network more efficiently,<br />

in comparison with <strong>the</strong> conditions<br />

afforded by existing transmission systems.<br />

SDH makes it possible <strong>to</strong> set up new connections<br />

from a remote site within a few<br />

seconds. This enables a Telecom opera<strong>to</strong>r<br />

<strong>to</strong> respond quickly <strong>to</strong> cus<strong>to</strong>mer demands<br />

for new or higher capacity. It also<br />

reduces <strong>the</strong> operational cost because less<br />

manpower is required.<br />

Abbreviations<br />

ACSE Association Control Service Element<br />

API Application Programmer's Interface<br />

AUI Attachment Unit Interface<br />

CLNP Conectionless <strong>Network</strong> Pro<strong>to</strong>col<br />

CMISE Common Management Information Service<br />

Element<br />

CP Central Processor<br />

CPU Central Processor Unit<br />

CS Control System<br />

CSA Control System Application<br />

CSP Control System Platform<br />

DCC Data Communications Channel<br />

DCN Data Communications <strong>Network</strong><br />

ECC Embedded Control Channel<br />

ETNA <strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong><br />

FMAS Facility Management System<br />

GNE Gateway <strong>Network</strong> Element<br />

GUI Graphical User Interface<br />

ICN Internal Communications <strong>Network</strong><br />

IM Information Model<br />

IPC Inter-Process Communication<br />

ISDN Intergrated Sevices Digital <strong>Network</strong><br />

LAPD Link Access Pro<strong>to</strong>col on D-channel<br />

LAN Local Area <strong>Network</strong><br />

<strong>An</strong>o<strong>the</strong>r characteristic of <strong>to</strong>day's transmission<br />

networks is that each vendor has his<br />

LLC<br />

MAC<br />

MAU<br />

MCF<br />

MIB<br />

MO<br />

NE<br />

OS<br />

OSI<br />

PDH<br />

PI<br />

PM<br />

ROSE<br />

SDH<br />

SDXC<br />

SEMF<br />

SMS<br />

SMUX<br />

SNI<br />

SNPA<br />

su<br />

TAU<br />

TMN<br />

UP<br />

Logical Link Control<br />

Medium Access Control<br />

Medium Attachment Unit<br />

Message Communications Function<br />

Management Information Base<br />

Managed Object<br />

<strong>Network</strong> Element<br />

Operations System<br />

Open Systems Interconnection<br />

Plesiochronous Digital Hierarchy<br />

Physical Interface<br />

Performance Moni<strong>to</strong>ring<br />

Remote Operations Service Element<br />

Synchronous Digital Hierarchy<br />

SDH Digital Cross Connect<br />

Synchronous Equipment Management<br />

Function<br />

SDH Management Subnetwork<br />

SDH Multiplexer<br />

Switching <strong>Network</strong> Interface<br />

Sub-<strong>Network</strong> Point of Attachment<br />

Support Unit<br />

Termination Access Unit<br />

Telecommunications Management <strong>Network</strong><br />

Unit Processor<br />

Self-healing networks can be configured in<br />

such a way that faults in <strong>the</strong> network - e.g.<br />

cable breaks - will not affect <strong>the</strong> traffic for<br />

more than a few milliseconds, or seconds,<br />

depending on <strong>the</strong> size of <strong>the</strong> network and<br />

<strong>the</strong> res<strong>to</strong>ration principle applied. At<br />

present it can take hours, or even days, <strong>to</strong><br />

locate <strong>the</strong> fault and take appropriate actions.<br />

Two different principles are applied <strong>to</strong> protect<br />

<strong>the</strong> network against <strong>the</strong> effects of a<br />

fault: protection switching and protection<br />

routing.<br />

Protection switching is performed by <strong>the</strong><br />

SDH NE without assistance from a central<br />

network management system, such as<br />

<strong>Ericsson</strong>'s FMAS. This means very fast<br />

res<strong>to</strong>ration but utilises network resources<br />

quite inefficiently.<br />

It is anticipated that rings will be a commonly<br />

used network <strong>to</strong>pology in SDH networks,<br />

especially in local networks. In <strong>the</strong><br />

event of a cable break, <strong>the</strong> traffic can be<br />

ERICSSON REVIEW No. 3, 1992


JOHAN BLUME<br />

LEIF HANSSON<br />

PEDER HAGG<br />

LEIF SUNDIN<br />

<strong>Ericsson</strong> Telecom AB<br />

res<strong>to</strong>red by sending it <strong>the</strong> opposite way<br />

round <strong>the</strong> ring.<br />

Protection routing requires assistance<br />

from <strong>the</strong> FMAS and takes a somewhat<br />

longer time (5-10 seconds) but can be<br />

used for any type and size of network. In<br />

this case alarm information is sent <strong>to</strong> <strong>the</strong><br />

FMAS from all affected NEs. The FMAS<br />

analyses <strong>the</strong> fault situation and calculates<br />

a new, optimised way through <strong>the</strong> network.<br />

Cross-connect commands are <strong>the</strong>n sent <strong>to</strong><br />

<strong>the</strong> NEs <strong>to</strong> set up <strong>the</strong> new connection. If<br />

desired, <strong>the</strong> opera<strong>to</strong>r can set conditions for<br />

rerouting: that <strong>the</strong> route should not pass<br />

through a particular node, for example.<br />

The Performance moni<strong>to</strong>ring parameters<br />

enable <strong>the</strong> opera<strong>to</strong>r <strong>to</strong> identify potential<br />

problems before <strong>the</strong>y cause degradation<br />

of end-user service. They also offer a <strong>to</strong>ol<br />

for verifying <strong>the</strong> quality of <strong>the</strong> connection.<br />

This is important since many cus<strong>to</strong>mers require<br />

a high guaranteed quality level,<br />

which makes it necessary <strong>to</strong> be able <strong>to</strong><br />

measure that level. The parameters used<br />

conform <strong>to</strong> CCITT Recs. G.821, G.82x and<br />

G.784, Box 1.<br />

<strong>An</strong> important issue is protection of <strong>the</strong><br />

SDH NEs from unauthorised access. This<br />

becomes important in cases where available<br />

functions may cause serious problems<br />

if used incorrectly. Each opera<strong>to</strong>r<br />

must have a unique Userid and password,<br />

issued by <strong>the</strong> system administra<strong>to</strong>r. He is<br />

also assigned one of several 'user<br />

categories'. The user category determines<br />

what functions <strong>the</strong> user is allowed <strong>to</strong> access.<br />

User categories - of which some exemples<br />

follow - can be configured as desired:<br />

- System manager: handling of user categories<br />

and database management<br />

- Read access: only read access <strong>to</strong> management<br />

information<br />

- Configuration manager: access <strong>to</strong> installation<br />

and configuration functions<br />

- Data communication manager: handling<br />

data communication facilities.<br />

BOX1<br />

PERFORMANCE PARAMETERS<br />

In <strong>the</strong> future, <strong>the</strong> demand for high-quality connections will be even greater than <strong>to</strong>day. It is <strong>the</strong>refore important<br />

<strong>to</strong> use relevant and accepted parameters when measuring and verifying <strong>the</strong> quality of connections. The<br />

quality parameter currently in use is Bit Error Ratio (BER), with alarm thresholds at (normally) 10 3 or 10*.<br />

This is not good enough for data traffic. <strong>An</strong>o<strong>the</strong>r drawback of BER is that it does not give any information as<br />

<strong>to</strong> how faults are distributed in <strong>the</strong> time domain. Normally, faults are not distributed uniformly, but "burstily".<br />

To meet <strong>the</strong>se new requirements, CCITT has defined quality parameters in Rec. G.821:<br />

ES Errored Seconds<br />

SES Severely Errored Seconds<br />

DM Degraded Minutes<br />

UAS Unavailable Second<br />

No. of errors during 1 s >0<br />

BER, measured during 1 s >10'<br />

BER, measured during 1 min. MO*<br />

10 consecutive SES gives 10 UAS<br />

These parameters are initially intended for 64 kbit/s connections. <strong>An</strong>nex D <strong>to</strong> Rec. G.821 <strong>the</strong>refore defines<br />

how <strong>to</strong> deal with higher bit rates. The G.821 parameters - after ra<strong>the</strong>r animated discussions- have not been<br />

found <strong>to</strong> be <strong>the</strong> solution <strong>to</strong> new requirements imposed on quality parameters. For example, <strong>the</strong>y are still<br />

based on BER.<br />

A draft, Rec. G.82x, defines new quality parameters for bit rates higher than 64 kbit/s. The G.82x parameters<br />

will be used within SDH when <strong>the</strong> recommendation has been approved. The G.82x parameters are<br />

based on Errored Blocks (EB) instead of BER. One EB is a block that contains one or more errored bits.<br />

The following parameters were defined in <strong>the</strong> G.82x draft of June, 1992:<br />

ES<br />

ESR<br />

Errored Seconds<br />

ES Ratio<br />

s= 1 EB during 1 s<br />

The ratio of ES <strong>to</strong> <strong>the</strong> <strong>to</strong>tal number of seconds in available time during<br />

a specified measurement interval<br />

SES<br />

Severely Errored Seconds<br />

ss Y % EB during 1 s.<br />

(Y > 30 provisionally)<br />

SESR SES Ratio<br />

BBER Background Block<br />

Error Ratio<br />

The ratio of SES <strong>to</strong> <strong>the</strong> <strong>to</strong>tal number of seconds in available time during<br />

a specified measurement interval<br />

The ratio of errored blocks <strong>to</strong> <strong>the</strong> <strong>to</strong>tal number of blocks excluding all<br />

blocks during SES and unavailable time


64<br />

These functions and a number of o<strong>the</strong>r<br />

functions are described in greater detail<br />

and from <strong>the</strong> point of view of <strong>the</strong> SDH NEs<br />

in <strong>the</strong> section 'SDH Management'.<br />

Mode of operation<br />

The SDH NE can be managed from:<br />

- FMAS, <strong>the</strong> management system for <strong>the</strong><br />

<strong>Transport</strong> <strong>Network</strong><br />

- A Local Opera<strong>to</strong>r Terminal (LOT).<br />

The FMAS is necessary for <strong>the</strong> manage-<br />

ment of large transport networks. The LOT<br />

is required during installation, and can also<br />

be used <strong>to</strong> manage a single NE or a small<br />

transport network.<br />

All SDH NEs use a standardised interface<br />

<strong>to</strong> <strong>the</strong> FMAS. The pro<strong>to</strong>cols used are TMN<br />

(Telecommunications Management <strong>Network</strong>)<br />

Q3 interfaces as defined in CCITT<br />

Recs. Q.811 and Q.812, Box 2.<br />

All SDH NEs also use a common Information<br />

Model (IM). The IM defines <strong>the</strong> syntax<br />

Fig. A<br />

Gateway <strong>Network</strong> Elements (GNE) may be connected<br />

<strong>to</strong> an OS. The GNE has an attached subnetwork<br />

of SDH NEs<br />

BOX 2<br />

Q3-lnterface<br />

The Q3 interface provides for standardised<br />

communication and exchange of management<br />

information between an NE and an Operations<br />

System. The pro<strong>to</strong>col suite and <strong>the</strong> information<br />

model must be defined when specifying a Q3<br />

interface.<br />

Gateway <strong>Network</strong> Element (GNE)<br />

A GNE is connected <strong>to</strong> an OS via a Q3 interface,<br />

Fig A. The GNE has an attached subnetwork<br />

of SDH NEs and provides remote access<br />

<strong>to</strong> <strong>the</strong>se NEs by means of Embedded Control<br />

Channels (ECC). The GNE performs Intermediate<br />

System (IS) network layer routing functions<br />

for ECC messages destined <strong>to</strong> any NE<br />

within <strong>the</strong> subnetwork.<br />

When considering implementation, <strong>the</strong>re is no<br />

difference between a GNE and any o<strong>the</strong>r SDH<br />

NE. They simply perform different roles in <strong>the</strong><br />

OSI environment.<br />

Embedded Control Channel (ECC)<br />

The ECCs provide a high-capacity data communication<br />

network between SDH NEs, utilising<br />

dedicated bytes (DCC) in <strong>the</strong> STM-N Section<br />

Overhead as <strong>the</strong> physical layer. Two types<br />

of ECC have been defined in <strong>the</strong> SDH standards:<br />

-ECCr<br />

A192 kbit/sdata communications channel accessible<br />

by all NEs, including <strong>the</strong> intermediate<br />

regenera<strong>to</strong>rs<br />

-ECCm<br />

A 576 kbit/s data communications channel accessible<br />

by all NEs, excluding <strong>the</strong> intermediate<br />

regenera<strong>to</strong>rs<br />

The ECC network is logically created by defining<br />

ECC network routes in <strong>the</strong> SDH transport<br />

network. <strong>Network</strong> Pro<strong>to</strong>col Data Units (NPDU)<br />

are <strong>the</strong>n routed according <strong>to</strong> address and routing<br />

information held locally in <strong>the</strong> NEs as routing<br />

tables, or terminated within <strong>the</strong> NE.<br />

Table 1<br />

SDH GNE Local Routing Table<br />

NPDU Destination<br />

Address<br />

"SDH NE"<br />

"OS"<br />

"SDH GNE"<br />

Next Hop<br />

(SNPA)<br />

"STM-N ECC"<br />

"Q3"<br />

"Own Agent"<br />

In <strong>the</strong> absence of standards, a set of <strong>Ericsson</strong> proprietary<br />

DCN management functions has been defined<br />

for <strong>the</strong> purpose of managing <strong>the</strong> routing tables<br />

and DCN resources.<br />

ERICSSON REVIEW No. 3, 1992


65<br />

of <strong>the</strong> messages that are sent between <strong>the</strong><br />

SDH NEs and <strong>the</strong> FMAS.<br />

SDH Management<br />

General<br />

SDH Management is based on TMN and<br />

OSI principles <strong>to</strong> allow for <strong>the</strong> building of<br />

an open network architecture. The basic<br />

concept behind TMN is <strong>to</strong> provide an organised<br />

architecture <strong>to</strong> achieve interconnection<br />

between various types of Operations<br />

System (OS) and/or telecommunications<br />

equipment for <strong>the</strong> exchange of management<br />

information, using standardised<br />

pro<strong>to</strong>cols and interfaces.<br />

From a management point of view, <strong>the</strong><br />

SDH NE can be considered from three different<br />

perspectives:<br />

- A functional perspective<br />

- <strong>An</strong> information model perspective<br />

- A data communication perspective,<br />

each of <strong>the</strong>se perspectives defining some<br />

of <strong>the</strong> aspects necessary for standardised<br />

multi-vendor operations.<br />

Functional Perspective<br />

The Functional perspective defines <strong>the</strong><br />

management services that a single<br />

SDH NE can provide <strong>to</strong> a local opera<strong>to</strong>r,<br />

or <strong>to</strong> a network management system. In<br />

<strong>the</strong> TMN context, <strong>the</strong>se functions are referred<br />

<strong>to</strong> as TMN management functions.<br />

The TMN management functions belong<br />

<strong>to</strong> different management functional areas.<br />

Those relevant for SDH NEs are: Configuration<br />

Management, Fault Management,<br />

Performance Management and Security<br />

Management. In addition, a set of Data<br />

Communication <strong>Network</strong> management<br />

functions dealing with configuration of data<br />

communication resources have been defined.<br />

Within ETNA, <strong>the</strong> SDH NE-related services<br />

are used by <strong>the</strong> FMAS <strong>to</strong> provide network-related<br />

services such as trail provisioning,<br />

protection routing, path performance<br />

moni<strong>to</strong>ring, etc. Some examples of<br />

<strong>the</strong> most important TMN management<br />

functions provided are summarised below.<br />

Configuration Management (CM)<br />

Compared with traditional transmission<br />

equipment, <strong>the</strong> SDH NEs also include a<br />

switch which provides for <strong>the</strong> set-up of<br />

broadband semi-permanent connections.<br />

The main CM task is <strong>to</strong> control this switch,<br />

but CM also controls o<strong>the</strong>r aspects of <strong>the</strong><br />

configuration of <strong>the</strong> NE:<br />

- Termination Point Provisioning<br />

The different types of termination<br />

point, i.e. physical interfaces, trail termination<br />

points and connection termination<br />

points, can be configured in different<br />

ways, e.g. assigned identities,<br />

alarm thresholds, enabling and disabling<br />

of <strong>the</strong> laser, etc. The termination<br />

points are au<strong>to</strong>matically created<br />

when <strong>the</strong> related printed board assemblies<br />

are inserted<br />

- Equipment Configuration<br />

The SDH NEs are in many ways selfconfigurable<br />

after installation or extension<br />

of <strong>the</strong> hardware, e.g. when new<br />

access ports are installed. The equipment<br />

configuration functions keep<br />

track of <strong>the</strong> equipment currently installed,<br />

e.g. printed board assemblies<br />

and software, and report <strong>to</strong> <strong>the</strong> OS if<br />

changes have been made<br />

- Cross-Connect<br />

The cross-connect functions set up<br />

connections through <strong>the</strong> switch and<br />

keep an up-<strong>to</strong>-date list of <strong>the</strong> crossconnections<br />

currently being ordered<br />

- Protection Switching<br />

The SDH NEs can be configured <strong>to</strong><br />

perform different types of au<strong>to</strong>nomous<br />

protection switching (fast res<strong>to</strong>ration)<br />

of paths or sections <strong>to</strong> dedicated<br />

standby network resources following a<br />

network failure<br />

- Synchronisation Configuration<br />

Each NE must be synchronised from a<br />

valid synchronisation source, e.g. a<br />

2 Mbit/s signal, a 2 MHz reference or<br />

an STM-N signal. The synchronisation<br />

source configuration functions define<br />

<strong>the</strong> synchronisation source <strong>to</strong> be used,<br />

and what actions should be taken<br />

when <strong>the</strong> primary source fails<br />

- NE-Recovery<br />

The SDH NEs contain a lot of data<br />

which must be administered, e.g. regularly<br />

backed up. During certain trouble<br />

conditions it may also be necessary<br />

<strong>to</strong> perform restarts at different levels.<br />

Fault Management (FM)<br />

Fault management provides functions for<br />

detection, isolation and correction of abnormal<br />

states in <strong>the</strong> network. This includes<br />

both network-related faults, resulting from<br />

cable breaks or deteriorating line systems,<br />

and abnormal conditions within <strong>the</strong> NEs<br />

<strong>the</strong>mselves. The main FM task is <strong>to</strong> report<br />

ERICSSON REVIEW No. 3, 1992


66<br />

<strong>to</strong> <strong>the</strong> OS upon detection of a serious fault<br />

in <strong>the</strong> network, but it also controls diagnostic<br />

and test routines:<br />

- Alarm Surveillance<br />

The SDH NEs have <strong>the</strong> capability <strong>to</strong><br />

send alarm reports <strong>to</strong> <strong>the</strong> OS upon detection<br />

of a failure, and <strong>to</strong> s<strong>to</strong>re <strong>the</strong><br />

alarms in an event log<br />

- Fault Localisation and Testing<br />

The SDH NEs can be ordered <strong>to</strong> perform<br />

loopbacks, error injection, self-diagnostics,<br />

etc.<br />

Performance Management (PM)<br />

Performance management deals with <strong>the</strong><br />

functions necessary for an NE <strong>to</strong> collect,<br />

s<strong>to</strong>re, threshold, and report performance<br />

data associated with its moni<strong>to</strong>red PDH<br />

and SDH trail terminations. All applicationspecific<br />

and optional parameters, as specified<br />

in CCITT Rec. G.784, are supported.<br />

- PM Data Attribute Setting<br />

Basic PM data attributes, such as<br />

threshold values, can be defined<br />

- PM Data Reporting<br />

The SDH NEs can send PM data reports<br />

<strong>to</strong> an OS, ei<strong>the</strong>r when a defined<br />

threshold is exceeded (degraded or<br />

unacceptable performance level) or in<br />

accordance with predefined schedules<br />

- PM Data Logging<br />

PM data can be s<strong>to</strong>red in logs within<br />

<strong>the</strong> NEs and fetched by <strong>the</strong> OS when<br />

demanded.<br />

Security Management (SM)<br />

Security management functions deal with<br />

user access control <strong>to</strong> protect <strong>the</strong> network<br />

against unauthorised access <strong>to</strong> resources<br />

and services.<br />

- Login/Logout<br />

A local opera<strong>to</strong>r trying <strong>to</strong> access <strong>the</strong><br />

NE is checked against a user identity<br />

and a password<br />

• User Categories<br />

A user category defines different levels<br />

of function access privileges that<br />

can be assigned <strong>to</strong> a user. The lowest<br />

privilege level is read access, and <strong>the</strong><br />

highest level is <strong>the</strong> super-user category<br />

• Users<br />

New users can be defined and assigned<br />

a user identity, password and<br />

user category. Users can also be deleted.<br />

DCN Management<br />

DCN Management provides <strong>the</strong> functions<br />

necessary <strong>to</strong> control and configure <strong>the</strong><br />

data communication resources which<br />

allow communication <strong>to</strong> take place within<br />

<strong>the</strong> SDH Management <strong>Network</strong>.<br />

- <strong>Network</strong> Node Configuration<br />

Each SDH NE can be configured as a<br />

data communication node in <strong>the</strong> OSI<br />

environment, which means that addresses,<br />

application entity titles and<br />

names used locally must be defined<br />

- <strong>Network</strong> Route Configuration<br />

<strong>Network</strong> routes in <strong>the</strong> OSI context can<br />

be defined by means of routing tables<br />

and route priorities.<br />

The Information Model Perspective<br />

<strong>An</strong>o<strong>the</strong>r, more formal way of describing <strong>the</strong><br />

operation and control functionality is in<br />

terms of an Information Model (IM). <strong>An</strong> IM<br />

is an object-oriented description, independent<br />

of <strong>the</strong> actual physical realisation of<br />

<strong>the</strong> <strong>Network</strong> Element resources and how<br />

<strong>the</strong>se are managed.<br />

The Information Model consists of a collection<br />

of object classes, e.g. equipment,<br />

software, trail termination point, SDH<br />

Fig.1<br />

Model of Duplex Cross-connection<br />

Ptr<br />

CTP<br />

Bi<br />

Pointer<br />

Connection Termination Point<br />

Bidirectional<br />

ERICSSON REVIEW No. 3, 1992


67<br />

switch fabric. The characteristics of an object<br />

class are specified in terms of<br />

- read/write attribute values in objects,<br />

e.g. values of configuration parameters<br />

and relations <strong>to</strong> o<strong>the</strong>r objects, represented<br />

as lines between objects in Fig. 1 2 3<br />

- create/delete operations of objects<br />

- actions that can be performed on <strong>the</strong> object<br />

- notifications (i.e. spontaneous messages)<br />

sent from <strong>the</strong> object.<br />

In an executing system, manageable resources<br />

are represented as instances of<br />

<strong>the</strong>se object classes. The collection of instantiated<br />

objects is referred <strong>to</strong> as <strong>the</strong> Management<br />

Information Base (MIB). 2 There<br />

are two types of object in <strong>the</strong> MIB: Managed<br />

Objects and Support Objects. A Managed<br />

Object (MO) represents a physical or<br />

logical resource in <strong>the</strong> <strong>Network</strong> Element.<br />

A Support Object (SO) represents a log or<br />

an alarm filter, for example.<br />

The link between <strong>the</strong> TMN management<br />

functions and <strong>the</strong> IM is <strong>the</strong> implementation<br />

of each management function as one or<br />

more operations, actions or notifications in<br />

<strong>the</strong> objects that build up <strong>the</strong> MIB.<br />

A <strong>Network</strong> Element has what is called a<br />

TMN Agent, which can be seen as a process<br />

(function) acting on behalf of <strong>the</strong> managing<br />

system(s), relaying messages in<br />

both directions, Fig. 2.<br />

A generic Information Model is essential<br />

for <strong>the</strong> generation of management standards<br />

concerning configuration, fault, performance<br />

and security management functions.<br />

A common network model,<br />

identifying <strong>the</strong> generic resources in <strong>the</strong><br />

network - in this case <strong>the</strong> <strong>Transport</strong> <strong>Network</strong>-<br />

and <strong>the</strong>ir associated attribute types,<br />

events, actions and behaviour, provides a<br />

basis on which <strong>to</strong> explain <strong>the</strong> interrelationships<br />

between <strong>the</strong>se resources and <strong>the</strong><br />

network management system. Without<br />

this common view, a multi-vendortelecommunication<br />

network will not be achieved.<br />

The Information Model provided by <strong>Ericsson</strong><br />

and describing <strong>the</strong> SDH NE complies<br />

with <strong>the</strong> information model developed by<br />

CCITT's SG XV SDH Model (G.774), 6 and<br />

with <strong>the</strong> SG IV Generic <strong>Network</strong> Information<br />

Model (M.3100). 5 In general, <strong>the</strong> configuration<br />

management part is derived<br />

from G.774, while <strong>the</strong> fault management<br />

and performance management parts are<br />

derived from M.3100 and Q.821. By and<br />

large, SG IV follows <strong>the</strong> CCITT X.700 series<br />

of recommendations.<br />

The SDH Information Model is a subset of<br />

<strong>the</strong> ETNA Information Model common <strong>to</strong><br />

all <strong>Ericsson</strong> transport network elements<br />

connected <strong>to</strong> <strong>the</strong> FMAS, including PDH<br />

equipment and AXD I/O NEs.<br />

The Data Communication Perspective<br />

As well as providing telecommunications<br />

services, <strong>the</strong> SDH NEs also provide powerful<br />

data communications and network<br />

layer routing functions. The TMN function<br />

block that performs <strong>the</strong>se functions is called<br />

<strong>the</strong> Message Communication Function<br />

(MCF).<br />

The MCF is based on <strong>the</strong> OSI reference<br />

model, which makes it possible for an<br />

SDH NE <strong>to</strong> work as a data communication<br />

node in an open network architecture. 7<br />

Each of <strong>Ericsson</strong>'s SDH NEs can be<br />

equipped with a Q-interface and provide<br />

for Embedded Control Channel (ECC) access,<br />

which means that each NE can be<br />

connected <strong>to</strong> any Operations System that<br />

conforms <strong>to</strong> OSI and TMN standards, without<br />

<strong>the</strong> need for additional Mediation Devices<br />

or Q-adapters.<br />

The MCF performs network layer routing<br />

functions between <strong>the</strong> Q-interface and any<br />

Fig. 2<br />

Interworking between Manager, Agent and Managed<br />

Objects<br />

ERICSSON REVIEW No. 3, 1992


68<br />

BOX 3<br />

The SDH NE MANAGEMENT ORGANISATION­<br />

AL MODEL<br />

Management of SDH NEs is based on <strong>the</strong> management<br />

organisational model as outlined in Ret<br />

5. The model consists of <strong>the</strong> following TMN function<br />

blocks and components, Fig. A: The <strong>Network</strong><br />

Element Function (NEF) including <strong>the</strong> Management<br />

Information Base (MIB), <strong>the</strong> Message Communication<br />

Function (MCF), and <strong>the</strong> Management<br />

Application Function (MAF) including <strong>the</strong> agent.<br />

In addition, an MAF functional component containing<br />

a manager for local control of <strong>the</strong> NE has<br />

been defined. The local manager is housed in <strong>the</strong><br />

local opera<strong>to</strong>r's terminal.<br />

Agent<br />

Part of <strong>the</strong> MAF which is capable of responding<br />

<strong>to</strong> network management operations issued by<br />

a manager, and of issuing notifications, e.g.<br />

event reports, on behalf of <strong>the</strong> managed objects<br />

Manager<br />

Part of <strong>the</strong> MAF which is capable of issuing requests<br />

for network management operations,<br />

e.g. request performance data, set thresholds,<br />

receive event reports, etc.<br />

Local Manager<br />

A managerwhich is housed in a local opera<strong>to</strong>r's<br />

terminal and is capable of managing a single<br />

network element<br />

Management Application Function (MAF)<br />

<strong>An</strong> application process providing TMN services.<br />

The MAF includes an agent, or a manager, or<br />

both. The MAF is <strong>the</strong> origin and termination of<br />

all TMN messages<br />

Managed Object (MO)<br />

The manager's view of a resource within <strong>the</strong><br />

telecommunications environment that may be<br />

managed via <strong>the</strong> agent. Examples of MOs residing<br />

in an SDH NE are equipment, software,<br />

trail termination point, SDH switch fabric, alarm<br />

log, etc.<br />

Message Communication Function (MCF)<br />

Provides facilities for <strong>the</strong> transport of TMN messages<br />

<strong>to</strong> and from <strong>the</strong> MAF, as well as network<br />

layer routing functions<br />

<strong>Network</strong> Element Function (NEF)<br />

The entity within an NE that supports transport<br />

network based services, e.g. multiplexing,<br />

cross-connection, regeneration, etc. The NEF<br />

is represented <strong>to</strong> <strong>the</strong> manager as a set of managed<br />

objects.<br />

ECC subnetwork, or between any of <strong>the</strong><br />

ECC subnetworks, Boxes 2 and 3.<br />

Graphical User Interface<br />

General<br />

One of <strong>the</strong> most important aspects of<br />

SDH NE management is <strong>the</strong> ease with<br />

which it can be operated.<br />

It is true that a prime driving force behind<br />

<strong>the</strong> deployment of <strong>the</strong> <strong>Transport</strong> <strong>Network</strong><br />

is <strong>to</strong> facilitate <strong>the</strong> management of network<br />

resources by way of centralised management<br />

- that is, with Q-interfaces and FMAS<br />

- but NEs will never<strong>the</strong>less be operated<br />

and maintained from <strong>the</strong> local site. Some<br />

of <strong>the</strong> reasons for local operation are:<br />

- Backup when <strong>the</strong> OS, or <strong>the</strong> communication<br />

link <strong>to</strong> OS, is down<br />

- The network opera<strong>to</strong>r may wish <strong>to</strong> adopt<br />

a more decentralised management philosophy<br />

- Certain management functions are more<br />

easily performed on site because <strong>the</strong>y<br />

require physical manipulation of <strong>the</strong><br />

equipment.<br />

For this purpose a local opera<strong>to</strong>r's terminal<br />

with a Graphical User Interface (GUI)<br />

can be connected <strong>to</strong> <strong>the</strong> NE. The GUI is a<br />

window-based and mouse-operated interface<br />

through which <strong>the</strong> opera<strong>to</strong>r has access<br />

<strong>to</strong> all <strong>the</strong> management functions.<br />

In order for an opera<strong>to</strong>r <strong>to</strong> gain access <strong>to</strong><br />

<strong>the</strong> functionality, he has <strong>to</strong> prove his legitimacy<br />

by supplying an identification code<br />

and a password. This falls under Security<br />

Management, which also includes <strong>the</strong> possibility<br />

of a Super-user assigning opera<strong>to</strong>rs<br />

<strong>to</strong> particular user categories with extensive<br />

or more limited privileges.<br />

Configuration Management covers both<br />

physical configuration in an SDH NE-typically<br />

Printed Board Assemblies such as<br />

Fig. A<br />

SDH NE Management Organisational Model<br />

ERICSSON REVIEW No. 3, 1992


69<br />

<strong>the</strong> Termination Access Unit - and a logical<br />

configuration of <strong>the</strong> capacities for<br />

switching and multiplexing. These different<br />

types of Configuration Management<br />

are supported by different graphical views,<br />

e.g. a physical view and a logical view.<br />

When <strong>the</strong>re is a fault of some kind, e.g. a<br />

<strong>Transport</strong> <strong>Network</strong> related alarm such as<br />

Loss of Frame Alignment or excessive Bit<br />

Error Ratio, or a fault related <strong>to</strong> HW or SW<br />

in <strong>the</strong> actual NE, <strong>the</strong> details and possible<br />

consequences are reported <strong>to</strong> <strong>the</strong> GUI.<br />

Fault Management, which deals with <strong>the</strong>se<br />

functions, also covers testing and diagnostics.<br />

Performance is continuously moni<strong>to</strong>red,<br />

and <strong>the</strong> GUI presents <strong>the</strong> statistics graphically<br />

and/or in tabular form. These functions<br />

fall under Performance Management.<br />

The local GUIs conform <strong>to</strong> design standards<br />

developed for user interfaces by<br />

TMOS User Interface Design Standards<br />

(TUIDS), which means that <strong>the</strong> OPEN<br />

LOOK user interface style is currently<br />

used for <strong>the</strong> SDXCs, but Motif will also be<br />

included. For <strong>the</strong> SMUXs, Microsoft Windows<br />

is used.<br />

ETNA Harmonisation<br />

The current SDX Control System contains<br />

a GUI implemented on an IPX type Sun<br />

Workstation, but a portable terminal will<br />

also be provided. The GUI for SMUX is implemented<br />

on a PC 386/486. The fact that<br />

two different types of graphical terminal<br />

are used will not prevent 'look and feel' similarity<br />

between <strong>the</strong> systems, however.<br />

The Graphical User Interfaces for SDH<br />

NEs will be harmonised with each o<strong>the</strong>r<br />

and with <strong>the</strong> GUI for FMAS. Fig. 3 shows<br />

a typical detail of <strong>the</strong> DXC user interface.<br />

Control System <strong>Architecture</strong><br />

<strong>Introduction</strong><br />

The control system has <strong>to</strong> control <strong>Network</strong><br />

Elements which vary in size from only a<br />

few connected cables <strong>to</strong> over 8,000. The<br />

functionality associated with each connected<br />

cable makes heavy demands on<br />

<strong>the</strong> control system processing capacity.<br />

<strong>An</strong>o<strong>the</strong>r important fac<strong>to</strong>r <strong>to</strong> consider is<br />

cost, which has <strong>to</strong> be kept very low for small<br />

systems.<br />

A distributed computer architecture has<br />

been chosen <strong>to</strong> meet <strong>the</strong>se requirements.<br />

Fig. 3<br />

SDXC User Interface<br />

A window from <strong>the</strong> Configuration Management<br />

functional area<br />

ERICSSON REVIEW No. 3, 1992


Fig. 4<br />

Control System <strong>Architecture</strong><br />

CP<br />

UP<br />

ICM<br />

SNC<br />

Central Processor<br />

Unit Processor<br />

Internal Communication <strong>Network</strong><br />

Switching <strong>Network</strong> Controller (only SDXC)<br />

Each unit in <strong>the</strong> system has a powerful microprocessor,<br />

and a central master computer<br />

co-ordinates all <strong>the</strong> unit processors<br />

and provides input/output.<br />

Unit processors enable <strong>the</strong> unit itself <strong>to</strong><br />

perform a large amount of processing and<br />

so reduce <strong>the</strong> load on <strong>the</strong> central processor.<br />

All processors are connected <strong>to</strong> an Internal<br />

Communication <strong>Network</strong> (ICN). Depending<br />

on <strong>the</strong> size of <strong>the</strong> <strong>Network</strong> Elements,<br />

different internal communication<br />

network structures have been implemented,<br />

Fig. 4.<br />

The central processor houses <strong>the</strong> MIB, on<br />

which <strong>the</strong> OS and <strong>the</strong> local opera<strong>to</strong>r perform<br />

management operations. It will also<br />

implement parts of <strong>the</strong> MCF functionality,<br />

such as a network layer routing function<br />

between <strong>the</strong> Q- interface and <strong>the</strong> ECC subnetworks.<br />

The central processor can vary in size from<br />

small and inexpensive one-board microprocessor-based<br />

systems, <strong>to</strong> high-capacity<br />

redundant computers.<br />

Each of <strong>the</strong> different printed board assemblies<br />

under <strong>the</strong> control of <strong>the</strong> Central Processor<br />

contains its own Unit Processor.<br />

The Unit Processor is a building block consisting<br />

of a microprocessor, memories,<br />

communication controllers and A/D and<br />

D/A converters when required.<br />

The Unit Processors perform routine tasks<br />

on each printed board assembly, such as<br />

alarm surveillance, collection of performance<br />

parameters, and self-diagnostics.<br />

The Unit Processors also control <strong>the</strong> lower<br />

layers of <strong>the</strong> ECC pro<strong>to</strong>col suite.<br />

The control system software modularity is<br />

ensured through <strong>the</strong> use of a layered structure<br />

combined with object-oriented techniques.<br />

Programs can be downloaded from an Operations<br />

System all <strong>the</strong> way down <strong>to</strong> a Unit<br />

Processor.<br />

Control of Switching <strong>Network</strong><br />

Besides Unit Processors and <strong>the</strong> Central<br />

Processor, <strong>the</strong> control circuitry of <strong>the</strong><br />

SDXC switch, SNC, is also connected <strong>to</strong><br />

<strong>the</strong> Internal Communication <strong>Network</strong>. This<br />

means that all processors can set up and<br />

release cross-connections in <strong>the</strong> SDXC.<br />

This facility comes in<strong>to</strong> use, for example,<br />

when rapid switching has <strong>to</strong> be performed<br />

due <strong>to</strong> a transport network fault discovered<br />

by an access unit. In this case <strong>the</strong> Unit<br />

Processor on <strong>the</strong> access unit immediately<br />

reconfigures <strong>the</strong> switch according <strong>to</strong> a predefined<br />

configuration previously communicated<br />

<strong>to</strong> <strong>the</strong> Unit Processor by <strong>the</strong> Central<br />

Processor. The Central Processor is<br />

always informed of <strong>the</strong> resulting configuration.<br />

Unit Processors<br />

Unit Processors typically occupy part of a<br />

board in <strong>the</strong> SDXC. A unit in <strong>the</strong> SDXC normally<br />

consists of only one board. A Unit<br />

Processor includes a microprocessor chip<br />

and memory. Its current implementation is<br />

a Mo<strong>to</strong>rola 68 302.<br />

A Unit Processor continuously performs<br />

tasks such as moni<strong>to</strong>ring of hardware and<br />

calculation of bit errors, but at <strong>the</strong> same<br />

time it must react quickly <strong>to</strong> events such<br />

as incoming alarms from <strong>the</strong> transport network.<br />

The unit processor is <strong>the</strong>refore<br />

equipped with a real-time operating system<br />

kernel, OS 68, which gives response<br />

times on a microsecond level.<br />

SDXC CONTROL SYSTEM<br />

ARCHITECTURE<br />

General<br />

The SDXC control system may be composed<br />

of a central processor and up <strong>to</strong> a<br />

couple of 100 Unit Processors. It uses a<br />

packet-switched Internal Communication<br />

<strong>Network</strong> (ICN) which is integrated with <strong>the</strong><br />

switch.<br />

Purchased hardware and software is used;<br />

<strong>the</strong> Central Processor being a UNIX computer,<br />

for example.<br />

Central processor<br />

The Central Processor is <strong>the</strong> coordina<strong>to</strong>r<br />

and master of <strong>the</strong> complete control system.<br />

It continuously moni<strong>to</strong>rs <strong>the</strong> Internal<br />

Communication <strong>Network</strong> <strong>to</strong> find new Unit<br />

Processors. Local opera<strong>to</strong>rs have access<br />

<strong>to</strong> <strong>the</strong> SDXC system via graphical workstations.<br />

These are connected <strong>to</strong> <strong>the</strong> Central<br />

Processor via an E<strong>the</strong>rnet Local Area<br />

<strong>Network</strong>, Fig. 5. Both <strong>the</strong> Central Processor<br />

and <strong>the</strong> opera<strong>to</strong>r workstations are<br />

UNIX machines based on SPARC architecture,<br />

<strong>the</strong> Central Processor being a<br />

ERICSSON REVIEW No. 3, 1992


71<br />

Fig. 5<br />

The CP is connected <strong>to</strong> <strong>the</strong> ICN by a separate<br />

E<strong>the</strong>rnet. All control signals are routed through<br />

a centralised, triplicated switch and embedded<br />

in <strong>the</strong> traffic signal<br />

CTU<br />

I<br />

Control Termination Unit<br />

Traffic signal<br />

Control signal<br />

Sun SPARC2 and <strong>the</strong> opera<strong>to</strong>r terminals<br />

Sun IPX.<br />

The Central Processor is connected <strong>to</strong> <strong>the</strong><br />

Internal Communication <strong>Network</strong> by a separate<br />

E<strong>the</strong>rnet. It is separate <strong>to</strong> ensure that<br />

<strong>the</strong>re is adequate capacity and security for<br />

communication with <strong>the</strong> Unit Processors.<br />

E<strong>the</strong>rnet, being a standardised interface,<br />

enables a change of Central Processor<br />

supplier without any interface boards having<br />

<strong>to</strong> be redesigned.<br />

The next version of <strong>the</strong> CP will consist of<br />

basic and optional plug-in units, forming a<br />

modular system connected <strong>to</strong> a highspeed<br />

VME backplane and mounted in a<br />

standard subrack suitable for telecommunication<br />

purposes.<br />

Internal Communication <strong>Network</strong><br />

All processors are connected <strong>to</strong> <strong>the</strong> Internal<br />

Communication <strong>Network</strong>. It enables all<br />

processors <strong>to</strong> communicate with each<br />

o<strong>the</strong>r and with <strong>the</strong> switch control circuitry.<br />

The Internal Communication <strong>Network</strong> employs<br />

packet-switching techniques which<br />

enable processors <strong>to</strong> communicate using<br />

different data rates. The Central Processor<br />

uses 2 Mbit/s and Unit Processors<br />

0.5 Mbit/s.<br />

The internal cabling that carries <strong>the</strong> SNI<br />

signals used for transport of traffic information<br />

within <strong>the</strong> SDXC is also used as a<br />

part of <strong>the</strong> Internal Communication <strong>Network</strong>."<br />

Thus, expansion of <strong>the</strong> SDXC with<br />

new equipment at <strong>the</strong> same time increases<br />

<strong>the</strong> capacity of <strong>the</strong> communication network<br />

<strong>to</strong> cater for new Unit Processors. Ongoing<br />

communication is not affected by<br />

this expansion.<br />

By utilising <strong>the</strong> SNI signals, <strong>the</strong> Internal<br />

Communication <strong>Network</strong> benefits from <strong>the</strong><br />

reliable triplicated structure of <strong>the</strong> Switch.<br />

A single failure cannot cause a complete<br />

failure of <strong>the</strong> Internal Communication <strong>Network</strong>.<br />

All control signals are routed through a<br />

centralised packet switch. This is triplicated<br />

and each part resides on a switch plane,<br />

Fig. 5.<br />

The Central Processor gives or refuses<br />

permission <strong>to</strong> communicate within <strong>the</strong><br />

Internal Communication <strong>Network</strong> and<br />

keeps a record of all ongoing communication.<br />

In fault situations, all <strong>the</strong> processors<br />

involved s<strong>to</strong>p communicating and <strong>the</strong> fault<br />

is cleared by <strong>the</strong> Central Processor.<br />

Communication is normally between <strong>the</strong><br />

Central Processor and <strong>the</strong> Unit Processors,<br />

but for time-critical operations direct<br />

communication between Unit Processors<br />

can be used, e.g. for fast protection switching.<br />

Broadcast messages can be distributed<br />

over <strong>the</strong> Internal Communication <strong>Network</strong>.<br />

In this case a message is sent from a processor<br />

and <strong>the</strong>n duplicated by <strong>the</strong> communication<br />

network and sent <strong>to</strong> all access<br />

points where a processor may be connected.<br />

This facility is utilised by <strong>the</strong> Central<br />

Processor <strong>to</strong> find out whe<strong>the</strong>r any new Unit<br />

Processors have been connected <strong>to</strong> <strong>the</strong><br />

Internal Communication <strong>Network</strong>. This<br />

happens, for example, when a magazine<br />

is equipped with a new interface unit. The<br />

Central Processor sends a broadcast message<br />

which is answered by all new Unit<br />

Processors. <strong>An</strong>o<strong>the</strong>r possibility with <strong>the</strong><br />

ERICSSON REVIEW No. 3, 1992


72<br />

Fig. 6<br />

Software <strong>Architecture</strong><br />

CSA<br />

CSP<br />

Control System Application<br />

Control System Platform<br />

broadcast facility is for <strong>the</strong> Central Processor<br />

<strong>to</strong> distribute calendar date and time <strong>to</strong><br />

all Unit Processors simultaneously.<br />

The use of broadcast greatly reduces <strong>the</strong><br />

load on <strong>the</strong> Central Processor, since <strong>the</strong><br />

single message sent by <strong>the</strong> Central Processor<br />

is multiplied and distributed by <strong>the</strong><br />

packet-switched nodes within <strong>the</strong> Internal<br />

Communication <strong>Network</strong>.<br />

A set of rules defines <strong>the</strong> way communication<br />

between <strong>the</strong> application programs in<br />

<strong>the</strong> Central Processor and <strong>the</strong> Unit Processors<br />

is accomplished. The rules specify<br />

<strong>the</strong> size of packages, priority handling<br />

between different messages, actions <strong>to</strong> be<br />

taken when messages or parts of messages<br />

are lost, etc.<br />

All <strong>the</strong>se rules are designed in<strong>to</strong> four pro<strong>to</strong>col<br />

layers. Toge<strong>the</strong>r <strong>the</strong>se four layers<br />

hide all Internal Communication <strong>Network</strong><br />

implementation details from <strong>the</strong> application.<br />

The pro<strong>to</strong>cols are implemented as<br />

separate software modules, and one layer<br />

can <strong>the</strong>refore be modified without affecting<br />

<strong>the</strong> o<strong>the</strong>rs. The same pro<strong>to</strong>col software<br />

is used both in <strong>the</strong> Unit Processors and in<br />

<strong>the</strong> Central Processor. It is written in <strong>the</strong> C<br />

programming language.<br />

Software <strong>Architecture</strong><br />

General<br />

By definition, all software in <strong>the</strong> SDXC belongs<br />

<strong>to</strong> <strong>the</strong> Control System. Its purpose is<br />

<strong>to</strong> provide Operation, Administration,<br />

Maintenance and Provisioning functions<br />

via <strong>the</strong> external control interfaces, namely<br />

<strong>the</strong> Graphical User Interface and <strong>the</strong> Q-interface.<br />

The software is physically distributed<br />

between <strong>the</strong> Central Processor and<br />

<strong>the</strong> Unit Processors.<br />

There is a basic architectural distinction<br />

between platform-oriented and application-oriented<br />

software (referred <strong>to</strong> in Fig. 6<br />

as CSP and CSA respectively).<br />

Control System Platform<br />

The Control System Platform (CSP), <strong>to</strong>ge<strong>the</strong>r<br />

with <strong>the</strong> computer hardware and its<br />

Operating System - currently UNIX<br />

SVR4.1.2 - provides a platform that offers<br />

<strong>the</strong> following services:<br />

- Internal Communication<br />

IPC, i.e. Inter-Process Communication<br />

(inter- as well as intra-processor) by<br />

way of 'sockets'<br />

- External Communication<br />

Provision of Application Programmer's<br />

Interface (API) for communication with<br />

external management applications.<br />

These services use open, standardised,<br />

7-layer OSI stacks<br />

- Inven<strong>to</strong>ry<br />

Services for checking consistency<br />

between software and hardware configuration<br />

- Program loading<br />

Services for loading of software packages<br />

from <strong>the</strong> local sites as well as<br />

from <strong>the</strong> centralised network management<br />

system (FMAS)<br />

- Process handling<br />

Functions <strong>to</strong> enable supervision of processes<br />

- Run-time fault reporting<br />

Functions <strong>to</strong> enable <strong>the</strong> reporting of<br />

internal DXC Control System faults <strong>to</strong><br />

a management system<br />

- Restart<br />

Upon <strong>the</strong> detection of a Control System<br />

fault, <strong>the</strong> nature and seriousness<br />

of <strong>the</strong> fault are evaluated and <strong>the</strong> system<br />

is subsequently restarted from a<br />

well defined point of execution.<br />

Fig. 7<br />

The CSA <strong>Architecture</strong>, with its layered structure<br />

Control System Application<br />

The DXC CS application software constitutes<br />

all <strong>Transport</strong> <strong>Network</strong> oriented functionality.<br />

For a static description of <strong>the</strong><br />

ERICSSON REVIEW No. 3, 1992


73<br />

CSA, three aspects of <strong>the</strong> functionality are<br />

taken in<strong>to</strong> account, <strong>the</strong>reby creating a<br />

layered structure, in order <strong>to</strong> isolate different<br />

dependencies and stimulate a modular<br />

design, Fig. 7.<br />

- The User Layer<br />

contains <strong>the</strong> functionality of <strong>the</strong> Graphical<br />

User Interface. This layer of software<br />

will develop and change considerably<br />

due <strong>to</strong> different market needs and new<br />

<strong>to</strong>ols for graphical presentation. It is<br />

<strong>the</strong>refore separated from <strong>the</strong> underlying<br />

software by an interface called <strong>the</strong> Controller<br />

Interface (CI) which can be said <strong>to</strong><br />

represent <strong>the</strong> functionality provided by<br />

<strong>the</strong> TMN Layer and offered <strong>to</strong> management<br />

systems, such as centralised OS<br />

or local GUI<br />

- The TMN Layer<br />

consists of functions for managing <strong>Network</strong><br />

Elements specified in an objec<strong>to</strong>riented<br />

Information Model. <strong>An</strong> Information<br />

Model is becoming <strong>the</strong> standard<br />

way of specifying <strong>the</strong> manageable resources<br />

of a <strong>Network</strong> Element in <strong>the</strong><br />

<strong>Transport</strong> <strong>Network</strong>. CCITT Recommendation<br />

G.774 describes <strong>the</strong> basis for <strong>the</strong><br />

SDH NE Information Model, and thus<br />

forms <strong>the</strong> foundation for <strong>the</strong> DXC CS<br />

Telecommunication Management Layer<br />

(TMN). It will necessarily develop over<br />

<strong>the</strong> years, particularly since different<br />

cus<strong>to</strong>mers will require <strong>the</strong>ir SDH NE deployment<br />

in stages not coherent with<br />

CCITT Recommendation G.774 releases.<br />

For <strong>the</strong> purpose of isolating <strong>the</strong>se dependencies,<br />

<strong>the</strong> Information Model aspects<br />

are singled out in <strong>the</strong> TMN Layer<br />

- The System Layer<br />

For SDH NEs, CCITT Recommendations<br />

G.781-G.783 specify-through <strong>the</strong><br />

use of Functional Block descriptions -<br />

<strong>the</strong> functionality that must be provided.<br />

The descriptions are a reasonably stable<br />

set of requirements and are singled out<br />

in <strong>the</strong> DXC CS in <strong>the</strong> System Layer so<br />

as <strong>to</strong> be distinguished from <strong>the</strong> management<br />

aspects. The object-oriented approach<br />

is used here <strong>to</strong>o, i.e. <strong>the</strong> System<br />

Layer consists of a number of objects.<br />

The software described above is mapped<br />

on<strong>to</strong> <strong>the</strong> computer platform as shown in<br />

Fig. 8.<br />

From an application programmer's point of<br />

view, <strong>the</strong> Q-interface and <strong>the</strong> F-interface<br />

are handled in a similar way. The pro<strong>to</strong>cols<br />

are specified in <strong>the</strong> Controller Interface.<br />

OSI standards for service specifications<br />

(CMISE) are utilised. The Q-interface<br />

is a full 7-layer OSI stack, while <strong>the</strong> F-interface<br />

uses <strong>the</strong> IPC mechanism provided<br />

by <strong>the</strong> Control System Platform.<br />

Fig. 8<br />

Mapping of <strong>the</strong> CSA software on<strong>to</strong> <strong>the</strong> computer<br />

platform


74<br />

Functional requirements<br />

Certain DXC Control System functional requirements<br />

must be taken in<strong>to</strong> account<br />

when designing <strong>the</strong> software:<br />

- Several users<br />

A DXC may be equipped with more than<br />

one GUI plus a Q-interface, and all of<br />

<strong>the</strong>se interfaces must be able <strong>to</strong> operate<br />

simultaneously. This requires concurrency,<br />

which is implemented through<br />

several processes working on <strong>the</strong> objects<br />

(<strong>the</strong> MIB)<br />

- Real-time characteristics<br />

The DXC Control System must be eventdriven,<br />

in <strong>the</strong> sense that alarm detection<br />

mechanisms and subsequent evaluation<br />

and filtering are in some cases required<br />

<strong>to</strong> result in au<strong>to</strong>nomous reconfigurations<br />

within a certain time<br />

- Data s<strong>to</strong>rage and consistency<br />

The information in <strong>the</strong> objects contained<br />

in <strong>the</strong> MIB - representing everything of<br />

relevance <strong>to</strong> <strong>the</strong> management system -<br />

adds up <strong>to</strong> a considerable amount of<br />

data, <strong>the</strong> volume of which requires <strong>the</strong><br />

use of disk s<strong>to</strong>rage. The information on<br />

<strong>the</strong> disk is also used for backup purposes.<br />

Typically, 1 GByte is used in a normal<br />

configuration. Since <strong>the</strong> MIB is <strong>the</strong><br />

image used by <strong>the</strong> management system<br />

<strong>to</strong> represent <strong>the</strong> <strong>Network</strong> Element's<br />

manageable resources, it is of course of<br />

extreme importance that <strong>the</strong> data is consistent<br />

with <strong>the</strong> current configuration<br />

- Availability and robustness<br />

It is through <strong>the</strong> DXC Control System that<br />

a management system exerts its network<br />

control. This control might for example<br />

involve <strong>the</strong> setting up of a digital<br />

path through <strong>the</strong> <strong>Transport</strong> <strong>Network</strong> <strong>to</strong><br />

provide end-users with data communication<br />

capacity. Unless <strong>the</strong> DXC CS has<br />

very high availability - a system which is<br />

robust, i.e. resistant <strong>to</strong> faults - <strong>the</strong> endusers<br />

will not receive <strong>the</strong>ir services,<br />

which ultimately leads <strong>to</strong> reduced revenues<br />

for <strong>the</strong> network opera<strong>to</strong>r.<br />

Since different opera<strong>to</strong>rs are allowed <strong>to</strong><br />

operate <strong>the</strong> system simultaneously, it is<br />

essential <strong>to</strong> ensure that a consistent MIB<br />

is maintained. The operations are regarded<br />

as one or several transactions. Should<br />

a system fault occur, preventing a transaction<br />

from being carried out, a roll-back<br />

<strong>to</strong> a well defined state is performed. This<br />

means, in some cases, that part of <strong>the</strong> MIB<br />

must be locked during a Transaction. What<br />

constitutes a Transaction and what has <strong>to</strong><br />

be locked is <strong>the</strong> responsiblity of <strong>the</strong> TMN<br />

Agent functionality.<br />

Also, <strong>to</strong> allow for <strong>the</strong> shortest possible reaction<br />

times, i.e. <strong>to</strong> provide <strong>the</strong> event-driv-<br />

Fig. 9<br />

The SMUX Control System <strong>Architecture</strong><br />

MCF<br />

SEMF<br />

S<br />

Message Communications Function<br />

Synchronous Equipment Management Function<br />

S-interfaces as specified in CCITT Recs. G.782-<br />

G.783<br />

ERICSSON REVIEW No. 3, 1992


75<br />

en real-time functionality, lengthy executions<br />

must be divided in<strong>to</strong> Transactions.<br />

Data related <strong>to</strong> <strong>the</strong> Information Model can<br />

be s<strong>to</strong>red by using one of several possible<br />

techniques:<br />

- Object DataBase Management System,<br />

ODBMS<br />

This solution is very attractive in <strong>the</strong><br />

sense that it provides persistent objects,<br />

which is exactly what <strong>the</strong> Information<br />

Model specification suggests. The difference<br />

between <strong>the</strong> implementation of <strong>the</strong><br />

system and <strong>the</strong> way it is described in an<br />

Information Model is less than in <strong>the</strong><br />

o<strong>the</strong>r solutions. Also, <strong>the</strong> amount of design<br />

and implementation work is considerably<br />

less<br />

- Relational Data-Base Management System,<br />

RDBMS<br />

This solution also limits <strong>the</strong> amount of<br />

design and implementation work because<br />

it provides mechanisms - such as<br />

an ODBMS - for data s<strong>to</strong>rage, transaction<br />

handling, roll-back, etc. However, a<br />

translation (design) must be made from<br />

<strong>the</strong> object-oriented specification <strong>to</strong> <strong>the</strong><br />

world of tables in a relational database<br />

- Ordinary files<br />

This solution is '<strong>the</strong> hacker's choice'. It<br />

is straightforward in that it uses only what<br />

<strong>the</strong> Operating System and <strong>the</strong> programming<br />

language provide<br />

- Class library<br />

This represents something of a compromise.<br />

It provides basic data persistence<br />

functionality for objects.<br />

Implementation Structure<br />

The network resource represented by a<br />

Managed Object is partly implemented in<br />

hardware, and <strong>the</strong> software is divided<br />

between <strong>the</strong> Central Processor and Unit<br />

Processors. To utilise <strong>the</strong> distributed processor<br />

structure, which in a DXC of normal<br />

size means one CP and some 100 UPs<br />

or more, as much as possible of <strong>the</strong> functionality<br />

is delegated <strong>to</strong> <strong>the</strong> UPs. The Internal<br />

Communication <strong>Network</strong>, ICN, allows<br />

UPs <strong>to</strong> communicate directly without involving<br />

<strong>the</strong> CP, which means that real-time<br />

functions such as protection switching and<br />

network synchronisation can be handled<br />

in <strong>the</strong> UPs.<br />

The DXC Control System developed for<br />

<strong>the</strong> German PTT by <strong>Ericsson</strong> <strong>to</strong>ge<strong>the</strong>r<br />

with <strong>the</strong> German companies FUBA and<br />

DeTeWe - its partners in <strong>the</strong> FLEXNODE<br />

consortium, which will install AXD 4/1 and<br />

AXD 1/0 in 6 sites - is regarded as a pro<strong>to</strong>type<br />

system and does not show all <strong>the</strong><br />

characteristics mentioned as requirements<br />

above.<br />

SMUX CONTROL SYSTEM<br />

ARCHITECTURE<br />

General<br />

The SMUX Control System is implemented<br />

mainly by programs executed on a<br />

Central Processor (<strong>the</strong> Support Unit) as<br />

well as Unit Processors (UPs) distributed<br />

on each transmission printed board assembly<br />

within <strong>the</strong> NE. The SU is a oneboard<br />

processor common <strong>to</strong> <strong>the</strong> whole<br />

range of <strong>Ericsson</strong> SDH multiplexers and<br />

optimised for small NEs. The SU may be<br />

common <strong>to</strong> a number - normally two - of<br />

SDH multiplexers.<br />

The SU has an overall responsibility for<br />

management of <strong>the</strong> NE and receives and<br />

evaluates management operations issued<br />

from an OS or a local opera<strong>to</strong>r. As a response<br />

<strong>to</strong> events detected in <strong>the</strong> network,<br />

<strong>the</strong> SU issues notifications, e.g. alarm<br />

messages, <strong>to</strong> <strong>the</strong> OS. The Q3 and ECC<br />

pro<strong>to</strong>col suites, and <strong>the</strong> F-interface, are<br />

also controlled by <strong>the</strong> SU.<br />

Functions of a simple nature but with a high<br />

repetition rate, e.g. scanning of binary indications,<br />

alarms, calculation of PM parameters<br />

and operations in close connection<br />

with transmission hardware, are<br />

performed by UPs.<br />

The implementation of <strong>the</strong> SMUX Control<br />

System mapped on<strong>to</strong> CCITT Recs. G.782-<br />

G.783, is shown in Fig. 9.<br />

The SEMF is a function block which sends<br />

and receives data on low-level management<br />

functions <strong>to</strong> and from transmissionoriented<br />

function blocks.<br />

The MCF is implemented as a pro<strong>to</strong>col machine<br />

on <strong>the</strong> SU, and <strong>the</strong> SEMF is implemented<br />

as software both on <strong>the</strong> SU and<br />

on UPs. The Management Information<br />

Base (MIB) is located on <strong>the</strong> SU, while <strong>the</strong><br />

S-interfaces, as specified in CCITT Recs.<br />

G.782-G.783, are implemented as an<br />

internal processor bus between <strong>the</strong> UP<br />

and hardware registers on <strong>the</strong> transmission<br />

printed board assemblies.<br />

Commercially available products supplied<br />

by Retix are used <strong>to</strong> implement <strong>the</strong> Q3 and<br />

ECC pro<strong>to</strong>col suites.


76<br />

Fig 10<br />

SMUX Control System Hardware implementation<br />

TAU<br />

MAU<br />

AUI<br />

LAN<br />

Termination Access Unit<br />

Medium Attachment Unit<br />

Attachment Unit Interface<br />

Local Area <strong>Network</strong><br />

Hardware Platform<br />

The management subsystem hardware<br />

platform consists of processors at two different<br />

levels:<br />

- Central Processor (SU)<br />

- Unit Processors (UP).<br />

In addition, <strong>the</strong> following equipment may<br />

be required for control and operation:<br />

- IBM-compatible 386/486 PC (Local<br />

Opera<strong>to</strong>r's Terminal)<br />

- MAU (E<strong>the</strong>rnet transceiver) for connection<br />

<strong>to</strong> a DCN of LAN type.<br />

The SMUX Control System Hardware <strong>Architecture</strong><br />

is illustrated in Fig. 10.<br />

The SU communicates with <strong>the</strong> UPs via an<br />

internal ISO 8482 bus, which is similar <strong>to</strong><br />

RS-485.<br />

The SU implementation is mainly based on<br />

<strong>the</strong> following circuits:<br />

-CPU<br />

- E<strong>the</strong>rnet controller (Q3-interface)<br />

- RS-232 Communication controller (Finterface)<br />

- LAPD Communication controllers (internal<br />

communication)<br />

- Relay contacts (Station alarm interfaces)<br />

- Detection logic (External alarm interfaces)<br />

- Program memory<br />

- Data memory<br />

- Backup memory for non-volatile data.<br />

The UP is a general hardware building<br />

block common <strong>to</strong> all transmission printed<br />

board assemblies. The UP implementation<br />

is mainly based on <strong>the</strong> following circuits:<br />

-CPU<br />

- LAPD communication controllers (ECC<br />

and internal communication)<br />

- A/D and D/A converters, for measurement<br />

of laser characteristics, such as<br />

input power and laser bias current<br />

- Test interface (gains access <strong>to</strong> UP software<br />

for an authorised user)<br />

- Program memory<br />

- Data memory<br />

- Backup memory for non-volatile data.<br />

Not all UP circuits have <strong>to</strong> be present on<br />

every transmission printed board assembly.<br />

The CPU used both for SU and UPs is <strong>the</strong><br />

Mo<strong>to</strong>rola 68 302, which is a microprocessor<br />

optimised for data communication<br />

(ISDN) purposes.<br />

<strong>An</strong> IBM-compatible 386/486 PC is used as<br />

Local Opera<strong>to</strong>r Terminal. The LOT provides<br />

<strong>the</strong> opera<strong>to</strong>r with a <strong>Network</strong> Element<br />

view. However, it is possible <strong>to</strong> manage<br />

small networks from <strong>the</strong> LOT, but of course<br />

without network view.<br />

To manage a network from an LOT without<br />

assistance from <strong>the</strong> FMAS, communication<br />

over <strong>the</strong> ECCs is used, Box 2. By<br />

using ECC it is possible for an LOT <strong>to</strong> exchange<br />

messages with any o<strong>the</strong>r<br />

SDH NEs within <strong>the</strong> SDH network. This<br />

possibility (of accessing remotely located<br />

SDH NEs) is referred <strong>to</strong> as Remote Login.<br />

It will be a valuable feature, especially for<br />

early field trials and installations without a<br />

complete <strong>Network</strong> Management System.<br />

Software <strong>Architecture</strong><br />

The SU and UP software in an SMUX<br />

forms a loosely coupled, distributed software<br />

system organised in<strong>to</strong> a layered<br />

structure, where each layer has a well defined<br />

task, Fig. 11.<br />

Additionally, PC software which is not indicated<br />

in Fig. 11 is required for <strong>the</strong> Graphical<br />

User Interface.<br />

The SU software layers and <strong>the</strong>ir tasks are<br />

as follows:<br />

ERICSSON REVIEW No. 3, 1992


Fig. 11<br />

SMUX Software <strong>Architecture</strong><br />

• The SU application software communicates<br />

peer-<strong>to</strong>-peer with <strong>the</strong> UP's application software<br />

• » The SU communicates with <strong>the</strong> UP's by using<br />

an ISO 8482 backplane bus<br />

References<br />

1 Tarle, H.: FMAS - <strong>An</strong> Operations Support<br />

System for <strong>Transport</strong> <strong>Network</strong>s.<br />

<strong>Ericsson</strong> Review 67 (1990):2, pp.<br />

163-182.<br />

2 Widl.W.: CCITTStandardisation of Telecommunications<br />

Management <strong>Network</strong>s<br />

<strong>Ericsson</strong> Review 68 (1991) :2, pp. 34-51.<br />

3 Widl, W. and Woldegiorgis, K.: In Search<br />

of Managed Objects. <strong>Ericsson</strong> Review 69<br />

(1992):1/2, pp. 34-56.<br />

4 Bergkvist, J. A., Evangelisti, G. and Hopfinger,<br />

J.: AXD 4/1, a Digital Cross-Connect<br />

System. <strong>Ericsson</strong> Review 69<br />

(1992):3, pp. 78-68.<br />

5 CCITT Draft Rec. M.3010: Principles for<br />

a Telecommunications Management<br />

<strong>Network</strong>.<br />

6 CCITT Rec. G.774: SDH <strong>Network</strong> Information<br />

Model for TMN<br />

7 CCITT Rec. X.200; Reference Model of<br />

Open Systems Interconnection for<br />

CCITT Applications.<br />

- User Access layer<br />

Provides external access for an OS or<br />

local opera<strong>to</strong>r <strong>to</strong> <strong>the</strong> management view<br />

of <strong>the</strong> SDH multiplexer. Contains <strong>the</strong><br />

data communication services for <strong>the</strong> F,<br />

Q3 and ECC interfaces<br />

• TMN layer<br />

Provides <strong>the</strong> generalised TMN management<br />

view of <strong>the</strong> SDH multiplexer in <strong>the</strong><br />

form of a Management Information Base<br />

with <strong>the</strong> managed objects, <strong>the</strong>ir attributes,<br />

actions, and emitted notifications<br />

• SMUX layer<br />

Contains a logical SDH multiplexer as<br />

specified in CCITT Recs. G.782-G.783.<br />

This logical multiplexer can be controlled<br />

from <strong>the</strong> TMN layer and has <strong>the</strong> standardised<br />

au<strong>to</strong>matic behaviour for protection<br />

switching and change of synchronisation<br />

source<br />

- Magazine layer<br />

Manages all hardware units in <strong>the</strong> magazine<br />

so that <strong>the</strong>y provide <strong>the</strong> logical<br />

SDH multiplexer transmission services<br />

requested by <strong>the</strong> SMUX layer using <strong>the</strong><br />

available units, and signal interconnections<br />

in <strong>the</strong> magazine<br />

- Unit layer<br />

Manages individually each hardware<br />

unit in <strong>the</strong> magazine, ordering changes<br />

in <strong>the</strong> unit, receiving events from <strong>the</strong> unit<br />

and ensuring that <strong>the</strong> U P software is consistent<br />

with <strong>the</strong> SU software<br />

- Base layer<br />

Provides process management and<br />

communication, drivers for SU I/O ports<br />

and communication services between<br />

<strong>the</strong> SU and UPs<br />

- Virtual machine layer<br />

Provides a real-time, multi-task virtual<br />

machine on bare machine hardware.<br />

Contains Operating System kernel and<br />

low-level hardware interfacing <strong>to</strong> <strong>the</strong> SU<br />

hardware.<br />

The UP software layers and <strong>the</strong>ir tasks are<br />

as follows:<br />

- Unit layer<br />

Manages each hardware unit individually,<br />

making changes on <strong>the</strong> unit according<br />

<strong>to</strong> orders from <strong>the</strong> SU, and reporting<br />

events from <strong>the</strong> unit<br />

- Base layer<br />

Provides process management and<br />

communication, drivers for I/O ports on<br />

<strong>the</strong> unit, and communication services<br />

between <strong>the</strong> UP and SU<br />

- Virtual machine layer<br />

Provides a real-time, multi-task virtual<br />

machine on bare machine hardware.<br />

Contains Operating System kernel and<br />

low-level software interfacing <strong>to</strong> <strong>the</strong><br />

transmission circuits.<br />

The PC application software and its purpose<br />

are as follows:<br />

- Graphical User Interface (GUI)<br />

Provides <strong>the</strong> local opera<strong>to</strong>r with a graphical<br />

user interface.<br />

Summary<br />

The control and operation of SDH <strong>Network</strong><br />

Elements is and will continue <strong>to</strong> be adapted<br />

<strong>to</strong> <strong>the</strong> evolving TMN standards. This<br />

facilitates <strong>the</strong>ir connection <strong>to</strong> centralised<br />

Operations Systems.<br />

The implementation of <strong>the</strong> control system<br />

takes in<strong>to</strong> consideration <strong>the</strong> various demands<br />

of <strong>Network</strong> Elements, ranging in<br />

size and complexity from small SMUXs up<br />

<strong>to</strong> large SDXCs.<br />

ERICSSON REVIEW No. 3, 1992


AXD 4/1, a Digital Cross-Connect<br />

System<br />

Jan A Bergkvist, Giovanni Evangelisti and Jan Hopfinger<br />

AXD 4/1 is one of <strong>the</strong> new cross-connect products in <strong>the</strong> ETNA - <strong>Ericsson</strong> <strong>Transport</strong><br />

<strong>Network</strong> <strong>Architecture</strong> - concept. The system is designed <strong>to</strong> meet <strong>the</strong> different needs<br />

of an evolving network that will include both SDH and PDH equipment.<br />

The authors describe <strong>the</strong> system architecture and <strong>the</strong> present system implementation.<br />

Fast service provisioning, supervision providing<br />

guaranteed high-quality leased<br />

lines with capacities ranging from 2 Mbit/s<br />

up <strong>to</strong> VC-4 (155 Mbit/s), fast network configuration<br />

and better network administration<br />

are some of <strong>the</strong> benefits of an<br />

AXD 4/1. 3<br />

digital communication systems<br />

telecommunication networks<br />

telecommunication network management<br />

The AXD 4/1 is a digital Cross-Connect<br />

system which terminates digital signals<br />

and cross-connects <strong>the</strong>se signals or <strong>the</strong>ir<br />

tributary parts. The extensive switching<br />

and supervision facilities of <strong>the</strong> AXD 4/1<br />

system makes it suitable for a wide range<br />

of applications involving network provision<br />

and network protection.<br />

The AXD 4/1 is a vital component in <strong>the</strong><br />

ETNA network solution, offering fast service<br />

provisioning and high availability.<br />

The AXD is a switch that differs from an ordinary<br />

telephony switch in three ways:<br />

- It is controlled by commands from an operating<br />

system or an opera<strong>to</strong>r and not by<br />

embedded control information in <strong>the</strong><br />

transmitted signals<br />

- The holding times for a connection are<br />

days or weeks, as compared with minutes<br />

for a telephony switch<br />

- The bandwidth of <strong>the</strong> switched signals is<br />

in <strong>the</strong> range 1.5-155 Mbit/s, compared<br />

with 64 kbit/s.<br />

Functions<br />

The AXD 4/1 system replaces <strong>the</strong> manual<br />

distribution frames and multiplexers used<br />

in <strong>the</strong> present network.<br />

The AXD 4/1 cross-connects signals at all<br />

VC levels (VC-12 <strong>to</strong> VC-4) according <strong>to</strong> <strong>the</strong><br />

SDH standard, corresponding <strong>to</strong><br />

2-155 Mbit/s. Both PDH and SDH signals<br />

can be terminated and cross-connected simultaneously<br />

in <strong>the</strong> same system, Table 1.<br />

The specification of <strong>the</strong> system's internal<br />

interfaces allows all sixty-four 2 Mbit/s and<br />

all four 34 Mbit/s signals in a 140 Mbit/s<br />

signal <strong>to</strong> be used. This means that <strong>the</strong> introduction<br />

of SDH in <strong>the</strong> PDH network will<br />

entail no network restrictions.<br />

In addition <strong>to</strong> <strong>the</strong> cross-connect functions,<br />

<strong>the</strong> AXD 4/1 adds extensive supervision<br />

functionality <strong>to</strong> <strong>the</strong> network. All terminated<br />

signals are continuously supervised for<br />

performance and fault. Combined with a<br />

central network management system,<br />

Fig. 1<br />

Opera<strong>to</strong>rs working with <strong>the</strong> AXD system<br />

ERICSSON REVIEW No. 3, 1992


JAN A BERGKVIST<br />

JAN HOPFINGER<br />

<strong>Ericsson</strong> Telecom AB<br />

Sweden<br />

GIOVANNI EVANGELISTI<br />

<strong>Ericsson</strong>-FATME Spa<br />

Italy<br />

Abbreviations<br />

ASIC<br />

AXD<br />

BCP<br />

Application Specific Integrated Circuit<br />

<strong>Ericsson</strong>'s DXC products<br />

Basic Control Pro<strong>to</strong>col<br />

BiCMOS Bipolar and CMOS<br />

CEPT<br />

CMI<br />

CP<br />

CS<br />

CSB<br />

CTU<br />

DMB<br />

DXC<br />

ECC<br />

ESU<br />

ETNA<br />

FMAS<br />

FPGA<br />

HCB<br />

HCMOS<br />

HPB<br />

European Telecom Standard<br />

Code Mark Inversion<br />

Central Processor<br />

Control S<strong>to</strong>re<br />

Clock generation and Synchronisation<br />

Board<br />

Control Termination Unit<br />

Distribution Matrix Board<br />

Digital Cross Connect<br />

Embedded Control Channel<br />

External Synchronisation Unit<br />

<strong>Ericsson</strong> <strong>Transport</strong> <strong>Network</strong> <strong>Architecture</strong><br />

Facility Management System<br />

Field Programmable Gate Array<br />

Horizontal Control Board<br />

High Speed CMOS<br />

Horizontal Power Board<br />

HW<br />

ID<br />

MV<br />

NAS<br />

PBA<br />

PCB<br />

PDH<br />

PROM<br />

PSU<br />

RAM<br />

SDH<br />

SMB<br />

SN<br />

SNI<br />

SS<br />

STM<br />

sw<br />

swc<br />

SWM<br />

TAU<br />

TCU<br />

TS<br />

UP<br />

VC<br />

Hardware<br />

Identity<br />

Majority Vote<br />

North American Standard<br />

Printed Board Assembly<br />

Printed Circuit Board<br />

Plesiochronous Digital Hierarchy<br />

Programmable Read Only Memory<br />

Protection Switching Unit<br />

Random Access Memory<br />

Synchronous Digital Hierarchy<br />

Switch Matrix Board<br />

Switching <strong>Network</strong><br />

Switching <strong>Network</strong> Interface<br />

Speech S<strong>to</strong>re<br />

Synchronous <strong>Transport</strong> Module<br />

Software<br />

Switching Cabinet<br />

Switching Module<br />

Terminal Access Unit<br />

Terminal Connection Unit<br />

Time Space Switch<br />

Unit Processor<br />

Virtual Container<br />

such as <strong>Ericsson</strong>'s FMAS, <strong>the</strong> network can<br />

be strikingly improved<br />

Features<br />

The AXD 4/1 is a powerful component for<br />

use in a variety of transport network applications.<br />

Fast delivery of broadband services<br />

as well as better utilisation of network<br />

resources are <strong>the</strong> most important benefits.<br />

Some paramount features of <strong>the</strong> AXD 4/1<br />

are:<br />

- Non-Blocking Synchronous Switching<br />

- <strong>Network</strong> Access at Various Levels<br />

- PDH/SDH Gateway Facility<br />

- <strong>Network</strong> Synchronisation Transparency<br />

- High Availability.<br />

To ensure flexibility in <strong>the</strong> network, <strong>the</strong><br />

AXD 4/1 system has full connectivity<br />

between all input and output ports by using<br />

a non-blocking switch structure. The nonblocking<br />

characteristic is guaranteed for all<br />

mixtures of signal levels and also when different<br />

mixtures of broadcast connections<br />

are handled. It is of particular importance<br />

<strong>to</strong> avoid restrictions in <strong>the</strong> network when<br />

introducing cross-connect systems.<br />

The AXD 4/1 is based on synchronous<br />

switching, using <strong>the</strong> SDH concept, and<br />

<strong>the</strong>refore capable of cross-connecting<br />

both PDH and SDH signals without introducing<br />

slips or violating <strong>the</strong> timing information.<br />

This makes for great flexibility in terms<br />

of cross-connect functionality. The internal<br />

format of <strong>the</strong> system allows <strong>the</strong> use of all<br />

34 Mbit/s and all 2 Mbit/s tributaries in a<br />

140 Mbit/s signal, or any mix of <strong>the</strong>se tributaries.<br />

This means that <strong>the</strong> AXD 4/1 system<br />

is well suited <strong>to</strong> function as a gateway<br />

between <strong>the</strong> plesiochronous and <strong>the</strong> synchronous<br />

networks without introducing<br />

any restrictions in <strong>to</strong>day's PDH network.<br />

The large traffic capacity makes availability<br />

a key concept; a triplicated structure<br />

Table 1<br />

Access and switch levels<br />

ERICSSON REVIEW No. 3, 1992


80<br />

Fig. 2<br />

AXD 4/1, System <strong>Architecture</strong><br />

has been chosen <strong>to</strong> achieve extremely<br />

high availability. The triplicated structure<br />

includes both <strong>the</strong> Switch and <strong>the</strong> internal<br />

connections throughout <strong>the</strong> system. The<br />

triplication is terminated by majority vote<br />

on bit level and results in a system with efficient<br />

fault detection and fault isolation<br />

and excellent availability. The cross-connected<br />

signals are completely unaffected<br />

by any single fault within <strong>the</strong> triplicated<br />

structure.<br />

System <strong>Architecture</strong><br />

Unlike most switching systems, <strong>the</strong><br />

AXD 4/1 is functionally divided in<strong>to</strong> only<br />

two structural parts, Devices and <strong>the</strong><br />

Switch, Fig. 2. In order <strong>to</strong> support this structure<br />

<strong>the</strong> system architecture has been built<br />

on three corner-s<strong>to</strong>nes:<br />

- Standardised interfaces<br />

- Integrated control paths<br />

- Triplicated Switch.<br />

The purpose of <strong>the</strong> Devices is <strong>to</strong> interface<br />

external signals and produce a logic,<br />

switchable signal, whereas <strong>the</strong> Switch performs<br />

<strong>the</strong> task of switching this latter type<br />

of signal between different Devices. Thus,<br />

switching is transparent and does not involve<br />

any data processing, since this is<br />

handled by <strong>the</strong> Devices. The Central Processor<br />

of <strong>the</strong> AXD 4/1 is also connected<br />

as a Device <strong>to</strong> ensure full flexibility.<br />

Internal Interfaces<br />

The 184 Mbit/s SNI-4, Box 1, has been designed<br />

<strong>to</strong> function as <strong>the</strong> internal interface<br />

between <strong>the</strong> Devices and <strong>the</strong> Switch. This<br />

interface includes both traffic and control<br />

data.<br />

The SNIs function as connecting-devices<br />

<strong>to</strong> <strong>the</strong> Switch by carrying all information<br />

needed, i.e. both traffic, timing, and control<br />

data. The transmission signals are carried<br />

and cross-connected in a circui<strong>to</strong>riented<br />

way, while <strong>the</strong> system's internal<br />

control data is transported and switched<br />

as packets.<br />

One <strong>to</strong> eight columns of <strong>the</strong> SNI-4 can be<br />

configured <strong>to</strong> carry ei<strong>the</strong>r packet-switched<br />

or circuit-switched data. The rest of <strong>the</strong><br />

SNI-4 capacity, except one column, can be<br />

circuit-switched.<br />

In order <strong>to</strong> effectively support Devices<br />

whose bandwidth demand is substantially<br />

lower than 155 Mbit/s, an interface version<br />

with lower bandwith (SNI-3) is supported<br />

by <strong>the</strong> Switch. The SNI-3 is structured in<br />

<strong>the</strong> same way and carries <strong>the</strong> same type<br />

of information as <strong>the</strong> SNI-4. <strong>An</strong> adap<strong>to</strong>r,<br />

Terminal Connection Unit (TCU), extends<br />

<strong>the</strong> architectural structure of <strong>the</strong> triplicated<br />

Switch, Fig. 4. The TCU provides multiplexing<br />

of four SNI-3 signals <strong>to</strong> a single<br />

SNI-4 interface.<br />

Integrated Control Paths<br />

The communication between different processors<br />

in <strong>the</strong> system (a common Central<br />

Processor for <strong>the</strong> whole AXD and one Unit<br />

Processor at each Device) uses capacity<br />

in <strong>the</strong> internal SNI interfaces. By using<br />

<strong>the</strong>se integrated paths for distribution of<br />

control information, <strong>the</strong> same redundancy<br />

and maintenance as for traffic signals are<br />

used, which gives more reliable communication<br />

than any bus structure regardless<br />

of redundancy structure. Since <strong>the</strong> control<br />

capacity of <strong>the</strong> SNI-4 can be adjusted,<br />

each Device will have exactly <strong>the</strong> capacity<br />

it needs. In addition, installation and<br />

internal cabling is easy. Only one connection<br />

has <strong>to</strong> be established <strong>to</strong> put a new Device<br />

in<strong>to</strong> operation, and <strong>the</strong>re is only one<br />

connection <strong>to</strong> supervise.<br />

Triplicated Hardware Structure<br />

The Switch is implemented by using a system<br />

structure consisting of three identical<br />

planes working in parallel. All three planes<br />

use <strong>the</strong> same input and perform <strong>the</strong> same<br />

functions in perfect synchronisation.<br />

The triplication is originated and terminated<br />

at Devices and at all processors within<br />

<strong>the</strong> triplicated structure. Termination of <strong>the</strong><br />

triplication is by majority vote at bit level,<br />

continuously for all signals.<br />

The triplicated Switch structure gives <strong>the</strong><br />

following characteristics:<br />

- 100% fault detection<br />

- 100% fault localisation<br />

-100% fault isolation.<br />

All single faults within <strong>the</strong> triplicated structure<br />

are immediately detected by <strong>the</strong> majority<br />

vote circuit, and <strong>the</strong> affected plane is<br />

directly indicated. The majority vote isolates<br />

<strong>the</strong> fault, and any disturbances originating<br />

from a plane are au<strong>to</strong>matically filtered.<br />

This means that <strong>the</strong> switched signal<br />

will never be disturbed, which is of <strong>the</strong> utmost<br />

importance for high-quality network<br />

service <strong>to</strong> be obtained. When <strong>the</strong> faulty<br />

plane has been identified, a plane com-<br />

ERICSSON REVIEW No. 3, 1992


81<br />

parison/majority vote is performed on sections<br />

of <strong>the</strong> planes <strong>to</strong> indicate <strong>the</strong> faulty<br />

board.<br />

Triplication results in a more robust<br />

system, with less complex maintenance<br />

functionality compared with that of a duplicated<br />

system, and makes it easier <strong>to</strong> upgrade<br />

functionality in <strong>the</strong> future.<br />

Processor distribution<br />

The Control System in AXD 4/1 consists<br />

of one common Central Processor (CP) for<br />

<strong>the</strong> whole AXD and one Unit Processor<br />

(UP) at each Device, i.e one board or a<br />

Switch plane. The CP performs all control<br />

functions and handles <strong>the</strong> communication<br />

<strong>to</strong>/from FMAS via a Q-interface. The CP<br />

can be a redundant or single machine with<br />

Box 1<br />

Switching <strong>Network</strong> Interfaces<br />

SNI - General<br />

To minimise <strong>the</strong> number of internal interfaces<br />

in <strong>the</strong> AXD, only Switching <strong>Network</strong> Interfaces<br />

(SNI) are allowed between system parts. Two<br />

different SNIs are specified <strong>to</strong>day.<br />

SNI-4<br />

<strong>the</strong> only format allowed between SN and High<br />

Speed Devices and between SN and TCUs<br />

SNI-3<br />

<strong>the</strong> only format allowed between TCU and<br />

Low Speed Devices.<br />

SNI-4<br />

Characteristics<br />

Data rate 163.84 Mbit/s<br />

Bit rate 184.32 Mbit/s<br />

Line code 8B9B, which means that each octet<br />

is completed with <strong>the</strong> ninth bit, which is an inversion<br />

of <strong>the</strong> eighth bit.<br />

Physical realisation<br />

SNI-4 contains data, timing and synchronisation<br />

information in one signal. One 50 ficoaxial<br />

pair is used as transmission medium in<br />

each direction.<br />

SNI-4 is self-adjustable <strong>to</strong> <strong>the</strong> cable length<br />

delay between <strong>the</strong> SN and device/TCU.<br />

Logical format<br />

The frame format of SNI-4 is called IVC-4,<br />

which stands for Internal Virtual Container at<br />

level 4. IVC-4 can carry both SDH and PDH<br />

signals at bit rates ranging from 1.5 Mbit/s <strong>to</strong><br />

155 Mbit/s, Fig. A.<br />

SNI-3<br />

Characteristics<br />

Data rate 40.96 Mbit/s<br />

Bit rate 20.48 Mbit/s<br />

Not line-coded.<br />

Physical realisation<br />

SNI-3 contains data, timing and synchronisation<br />

information in three signals. Two data<br />

wires and one clock wire with CMOS levels<br />

are used for transmission of SNI-3.<br />

Logical format<br />

The frame format of SNI-3 is called IVC-3,<br />

which stands for Internal Virtual Container at<br />

level 3. IVC-3 can carry PDH signals at bit<br />

rates ranging from 1.5 Mbit/s <strong>to</strong> 34 Mbit/s,<br />

Fig. B.<br />

Fig. A<br />

The IVC-4 frame consists of 2560 octets, which<br />

are divided in<strong>to</strong> 284 rows and 9 columns, plus 4<br />

octets for framing information. One column<br />

forms a switching entity. One SNI-4 can consist<br />

of four SNI-3s<br />

Fig. B<br />

The IVC-3 frame consists of 640 octets, which<br />

are divided in<strong>to</strong> 71 rows and 9 columns, plus 1<br />

octet for framing information. Four SNI-3s can be<br />

mapped in<strong>to</strong> one SNI-4<br />

ERICSSON REVIEW No. 3, 1992


82<br />

Fig. 3<br />

Time Space Switch structure<br />

SS<br />

Speach S<strong>to</strong>re<br />

single or redundant connection <strong>to</strong> <strong>the</strong> rest<br />

of <strong>the</strong> system <strong>to</strong> suite different availability<br />

needs. The CP is described in ano<strong>the</strong>r article<br />

in this issue of <strong>Ericsson</strong> Review.<br />

The UPs handle<br />

- Communication between <strong>the</strong> CP and <strong>the</strong><br />

Hardware (HW)<br />

- Maintenance and Control of <strong>the</strong> Device/-<br />

Switch plane<br />

- Alarm from <strong>the</strong> Device/Switch plane.<br />

The UPs are equipped with so-called<br />

Flash-PROMs, which means that one part<br />

of <strong>the</strong> program memory can be write-protected<br />

and one part can be updated. This<br />

makes it possible <strong>to</strong> use central, loadable<br />

software (SW) -an advantage in future upgradings<br />

of <strong>the</strong> system - and still have nonvolatile<br />

s<strong>to</strong>rage of <strong>the</strong> programs.<br />

Program-load and inven<strong>to</strong>ry are two important<br />

features of <strong>the</strong> processor block.<br />

New UP programs can be loaded ei<strong>the</strong>r<br />

from <strong>the</strong> CP through <strong>the</strong> SNI interfaces or<br />

through a special test port, located at <strong>the</strong><br />

board front. Programs are loaded without<br />

disturbing traffic in progress. Thanks <strong>to</strong> <strong>the</strong><br />

Flash-PROMs, <strong>the</strong> program will be kept<br />

during power-off, for example when shifting<br />

<strong>the</strong> board between two magazines.<br />

The product number, revision state and individual<br />

serial number of each board are<br />

s<strong>to</strong>red during production tests at <strong>the</strong> fac<strong>to</strong>ry<br />

and can be read by <strong>the</strong> CP. <strong>An</strong>o<strong>the</strong>r<br />

function handled by <strong>the</strong> UP is <strong>the</strong> reading<br />

of <strong>the</strong> physical board position in <strong>the</strong> magazine<br />

and <strong>the</strong> magazine ID. This information<br />

will be sent <strong>to</strong> <strong>the</strong> CP on request <strong>to</strong><br />

give a complete picture of <strong>the</strong> system's<br />

functional and physical configuration.<br />

The control system is based on an open<br />

architecture using object-oriented system<br />

design and language. Mo<strong>to</strong>rola type<br />

MC 68 302 computers are used as UPs.<br />

Sun or Stra<strong>to</strong>s UNIX machines can be<br />

chosen as CP, depending on reliability requirements.<br />

Switching<br />

The switching network performs all connections<br />

necessary for communication<br />

between <strong>the</strong> Devices. The Switch provides<br />

two types of function for <strong>the</strong> Devices:<br />

cross-connection of transmission signals<br />

and switching of control information.<br />

128 STM-1 equivalent ports, corresponding<br />

<strong>to</strong> 8196 2 Mbit/s signals, is <strong>the</strong> maximum<br />

capacity of <strong>the</strong> present AXD 4/1<br />

Switch.<br />

The AXD 4/1 is based on column switching,<br />

which allows all types of VC-n switching<br />

(n = 1,2,3 or 4) using <strong>the</strong> basic switching<br />

element of one SDH column, i.e.<br />

9 x 64 kbit/s = 576 kbit/s. These columns<br />

are <strong>the</strong>n combined <strong>to</strong> form <strong>the</strong> VC-ns that<br />

are <strong>to</strong> be cross-connected. The SDH and<br />

PDH signals that have pre-defined crossconnections<br />

are:<br />

VC-4, VC-3, VC-2, VC-12, VC-11 and<br />

140, 34, 2, 45 and 1.5 Mbit/s.<br />

A number of VC-ns, or columns, can be<br />

grouped <strong>to</strong>ge<strong>the</strong>r <strong>to</strong> form o<strong>the</strong>r cross-connections,<br />

e.g. concatenated VC-2s.<br />

Different types of connection can be established<br />

in <strong>the</strong> AXD 4/1 Switch, regardless<br />

of <strong>the</strong> hierarchical level or mix of <strong>the</strong><br />

connections:<br />

- Simplex, one way signals<br />

- Duplex, bidirectional signals<br />

- Broadcast, multi-destination signals.<br />

In <strong>the</strong> AXD 4/1 Switch, <strong>the</strong> number of destinations<br />

in a broadcast connection - as<br />

ERICSSON REVIEW No. 3, 1992


83<br />

well as <strong>the</strong> number of simultaneous broadcast<br />

connections - is unlimited. This<br />

means that an AXD 4/1 Switch is suitable<br />

for services on leased or switched circuits<br />

from 1.5 Mbit/s up <strong>to</strong> 140/155 Mbit/s.<br />

SNI-4 column, <strong>the</strong> basic switching<br />

entity<br />

<strong>An</strong> SNI-4 column, 9 x 64 kbit/s, is <strong>the</strong> smallest<br />

possible switchable entity. The bandwidth<br />

of one SNI-4 column is regarded as<br />

sufficiently small for internal applications<br />

<strong>to</strong>o.<br />

Each external signal terminated by <strong>the</strong><br />

AXD 4/1 and each standard tributary of that<br />

signal is mapped in<strong>to</strong> an integer number<br />

of columns. As a maximum, a whole<br />

STM-1 channel (155 Mbit/s) including<br />

overhead (OH) can be switched.<br />

To be able <strong>to</strong> provide connections that use<br />

a number of columns, <strong>the</strong> integrity of <strong>the</strong><br />

frame of columns is guaranteed, i.e. <strong>the</strong><br />

time sequence integrity.<br />

Non-blocking, full connectivity Switch<br />

matrix<br />

The Switch matrix is designed <strong>to</strong> provide<br />

circuit-switched connections without any<br />

internal congestion and <strong>to</strong> ensure full connectivity<br />

between any input and any output<br />

port, independently of <strong>the</strong> hierarchical<br />

level of <strong>the</strong> cross-connected signals and<br />

of <strong>the</strong> mix of hierarchies (CEPT, NAS,<br />

SDH, simplex, broadcast).<br />

The Switch is a Time-Space-switch (TS)<br />

consisting of a number of Speech S<strong>to</strong>res<br />

(SS), arranged in rows and columns: one<br />

row for each input and one column for each<br />

output, Fig. 3.<br />

Every time a signal is <strong>to</strong> be fed <strong>to</strong> a certain<br />

output, it can be chosen from among <strong>the</strong><br />

signals s<strong>to</strong>red in one of <strong>the</strong> SS (n) associated<br />

exclusively with that output. The<br />

choice is made through <strong>the</strong> use of Control<br />

S<strong>to</strong>res (CS) associated with that same output.<br />

No switch action will affect connections already<br />

established.<br />

Characteristics of <strong>the</strong> TS structure<br />

Since no internal routing is necessary and<br />

only one T-stage is involved, <strong>the</strong> TS switch<br />

has <strong>the</strong> following characteristics:<br />

- No rearrangement of established connections<br />

is needed <strong>to</strong> permit <strong>the</strong> set-up<br />

of a new connection, capable of being<br />

used for broadcast also<br />

- The delay introduced by <strong>the</strong> switch is<br />

minimised.<br />

This results in a switch with short set-up<br />

time and very low transmission delay,<br />

which is necessary in a large network with<br />

stringent service requirements.<br />

Multicolumn switching<br />

Multiconnection switching is characterised<br />

by <strong>the</strong> connection of a number of columns<br />

executed at <strong>the</strong> same time. The number of<br />

columns that can be connected in a multiconnect<br />

switching operation has no upper<br />

limit. As an extreme, <strong>the</strong> whole switch can<br />

be reconfigured at <strong>the</strong> same time.<br />

Internal communication<br />

The AXD 4/1 Switch is capable of handling<br />

packet-oriented signals. This function<br />

is used both for internal control communication<br />

and <strong>to</strong> handle Embedded Communication<br />

Channels (ECC) in SDH signals.<br />

A special communication pro<strong>to</strong>col is developed<br />

<strong>to</strong> handle <strong>the</strong> internal communication<br />

- <strong>the</strong> Basic Communication Pro<strong>to</strong>col<br />

(BCP). BCP is a self-addressing<br />

pro<strong>to</strong>col that uses <strong>the</strong> tree-structure of <strong>the</strong><br />

integrated control paths <strong>to</strong> give reliable internal<br />

communication. Information is carried<br />

in 'packet format'. The packet-handling<br />

equipment and <strong>the</strong> circuit switch are<br />

triplicated, so as <strong>to</strong> fulfil general requirements<br />

for failure immunity and fault detection<br />

capability.<br />

The supervision of internal communication<br />

is handled by BCP functionality and by <strong>the</strong><br />

triplication.<br />

Devices<br />

All types of equipment connected <strong>to</strong> <strong>the</strong><br />

Switch are called Devices. The most common<br />

types are Termination Access Units<br />

(TAU), which terminate transmission signals.<br />

The Central Processor, <strong>to</strong>o, is connected<br />

as a Device.<br />

In addition <strong>to</strong> <strong>the</strong> termination of signals, i.e.<br />

line signal access and multiplexing, a Device<br />

terminates <strong>the</strong> switching network. Majority<br />

vote is used <strong>to</strong> terminate <strong>the</strong> signals<br />

from <strong>the</strong> triplicated Switch. This is <strong>the</strong> basic<br />

function that forms a single fault-<strong>to</strong>lerant<br />

system.


84<br />

Fig. 4<br />

AXD 4/1 Devices<br />

Devices of AXD 4/1, Fig. 4, can be divided<br />

in<strong>to</strong> four different groups:<br />

- Terminal Access Units (TAU) interfacing<br />

plesiochronous transmission systems:<br />

• TAU 140, interfacing 140 Mbit/s signals<br />

• TAU 34, interfacing 34 Mbit/s signals<br />

• TAU 16x2, interfacing sixteen 2 Mbit/s<br />

signals<br />

- TAU interfacing transmission signals belonging<br />

<strong>to</strong> <strong>the</strong> Synchronous Digital Hierarchy<br />

(SDH) family:<br />

•TAU STM-1E, interfacing 155 Mbit/s<br />

electric signals<br />

• STM-1 and STM-4 optical TAU will be<br />

provided later on<br />

- Devices interfacing special external signals:<br />

• ESU, External Synchronisation Unit for<br />

2 MHz synchronisation signals<br />

- Devices for internal use:<br />

• CTU, Control Termination Unit, access<br />

unit for <strong>the</strong> Central Processor.<br />

Power distribution<br />

Power distribution in AXD 4/1 is decentralised.<br />

All boards containing electronic components<br />

are equipped with <strong>the</strong>ir own<br />

DC/DC converters, which produce <strong>the</strong> necessary<br />

voltages from <strong>the</strong> duplicated -48 V<br />

power source. This makes it possible <strong>to</strong><br />

have truly duplicated power supply all <strong>the</strong><br />

way from <strong>the</strong> exchange battery <strong>to</strong> <strong>the</strong> PBAs.<br />

Hardware Structure<br />

Basic Technology and Methodology<br />

The major part of <strong>the</strong> Hardware (HW) of<br />

AXD units is realised with Application Specific<br />

Integrated Circuits (ASIC). Texas BiC-<br />

MOS is used for high-speed applications.<br />

BiCMOS is a 0.8 micron process whose<br />

available speed in AXD applications exceeds<br />

200 MHz. Gate arrays with a maximum<br />

of 100,000 usable gates are available.<br />

Mo<strong>to</strong>rola HCMOS 0.7 micron circuits<br />

are used for o<strong>the</strong>r system parts. Max.<br />

speed in AXD applications is above 60<br />

MHz, and arrays with 127,000 usable<br />

gates are used.<br />

Structured HW design methodology has<br />

been used when designing ASICs. This<br />

means that functions are described in a<br />

High Level Language (Verilog); <strong>the</strong> code<br />

is <strong>the</strong>n syn<strong>the</strong>sised <strong>to</strong> gates and flip-flops<br />

by a syn<strong>the</strong>sis program (Synopsys). This<br />

procedure increases design efficiency and<br />

facilitates <strong>the</strong> structuring of large arrays.<br />

The occurrence of a number of high-speed<br />

signals between different units places extremely<br />

stringent requirements on <strong>the</strong> analog<br />

parts. <strong>An</strong>alog parts handle <strong>the</strong> termination,<br />

generation and regeneration of line<br />

signals, and phase and frequency control<br />

of signals. A library of analog blocks with<br />

schematics, component specifications<br />

and layouts has been created, which allows<br />

identical functions <strong>to</strong> be implemented<br />

in <strong>the</strong> same way in <strong>the</strong> different Devices.<br />

The use of one standard internal<br />

interface reduces <strong>to</strong> <strong>the</strong> absolute minimum<br />

<strong>the</strong> problems with analog design.<br />

A new metric building practice conforming<br />

<strong>to</strong> ETSI standard is used. Boards with 6 <strong>to</strong><br />

10 layers are placed in cabinets with a<br />

board spacing down <strong>to</strong> 16 mm. Circuit<br />

ERICSSON REVIEW No. 3, 1992


85<br />

Fig. 5<br />

Functional partioning of Switching Cabinet<br />

Only one plane is shown<br />

CSB<br />

HCB<br />

VCB<br />

DMB<br />

SMB<br />

UP<br />

DP<br />

CSG<br />

SWP<br />

Rx<br />

Tx<br />

Clock generation and Synchronisation Board<br />

Horizontal Control Board<br />

Vertical Control Board<br />

Distribution Matrix Board<br />

Switching Matrix Board<br />

Unit Processor<br />

Device Processor<br />

Clock and Synchronisation Genera<strong>to</strong>r<br />

Switch Port<br />

Receiver<br />

Transmitter<br />

packages with a pin spacing down <strong>to</strong> 0.5<br />

mm are used for some ASICs. Boards can<br />

be placed both vertically and horizontally.<br />

This allows <strong>the</strong> use of <strong>the</strong> crossboard technique:<br />

vertical boards are interfacing horizontal<br />

boards in <strong>the</strong> same subrack.<br />

The technology and methods described<br />

above offer a number of advantages:<br />

- Compact design, which gives high functional<br />

density per volume unit<br />

- Low power dissipation<br />

- High testability<br />

- Simple verification<br />

- Short design time.<br />

Switch Fabric<br />

The Switch Fabric which is <strong>the</strong> complete<br />

switch with all three planes, is implemented<br />

through single-type cabinets. The HW<br />

in one cabinet belongs <strong>to</strong> one Switch Fabric<br />

Plane.<br />

The maximum capacity of <strong>the</strong> Switch Fabric,<br />

in <strong>the</strong> present implementation, is<br />

128STM-1 equivalent ports.<br />

Modularity<br />

The AXD 4/1 features two types of mechanical<br />

modularity:<br />

- Board modularity<br />

- Cabinet modularity.<br />

Each Switch Cabinet (SWC) has a capacity<br />

of 64 x 128 ports. Two SWCs are required<br />

<strong>to</strong> arrange a 128 x 128-port Switch<br />

Fabric Plane. The minimum capacity is<br />

four ports, and extensions can be made in<br />

steps of four.<br />

Board Functions<br />

Switching <strong>Network</strong><br />

Fig. 5 shows <strong>the</strong> structuring of <strong>the</strong> main<br />

functions of <strong>the</strong> Switch Fabric on <strong>the</strong> different<br />

board types.<br />

The Clock generation and Synchronisation<br />

Board (CSB) generates all <strong>the</strong> clock<br />

and synchronisation signals required in <strong>the</strong><br />

system. Each CSB receives clock frequency<br />

and phase information from <strong>the</strong> o<strong>the</strong>r<br />

CSBs. Information from <strong>the</strong> selected synchronisation<br />

sources is also received.<br />

The CSB boards of <strong>the</strong> three planes are<br />

interconnected <strong>to</strong> ensure synchronisation<br />

of <strong>the</strong> system. The three signals are also<br />

compared for supervisory purposes.<br />

The Horizontal Control Board (HCB) is<br />

used for distribution of clock and control<br />

signals. The clock signal from CSB is distributed<br />

through HCB <strong>to</strong> all vertical boards.<br />

ERICSSON REVIEW No. 3, 1992


86<br />

The control signals, which come from <strong>the</strong><br />

Vertical Control Board (VCB), are distributed<br />

<strong>to</strong> all SMBs. In addition, HCB collects<br />

external alarms, e.g. temperature, fans,<br />

etc.<br />

The VCB is <strong>the</strong> core of <strong>the</strong> control functions<br />

in <strong>the</strong> AXD 4/1. Packet handling and<br />

control of <strong>the</strong> Switch plane are <strong>the</strong> two<br />

main tasks performed by <strong>the</strong> VCB, which<br />

contains <strong>the</strong> Unit Processor (UP). The<br />

packet handling is performed by a Router.<br />

The Router continuously looks for a packet<br />

<strong>to</strong> be routed and keeps track of <strong>the</strong> packet<br />

load.<br />

The UP handles <strong>the</strong> control of <strong>the</strong> entire<br />

Switch plane by controlling <strong>the</strong> Device Processors<br />

(DP) located on <strong>the</strong> respective<br />

boards in <strong>the</strong> Switching Cabinet.<br />

The Distribution Matrix Board (DMB) terminates<br />

four incoming SNI-4s from Devices,<br />

and separates <strong>the</strong> traffic part of <strong>the</strong> signal<br />

from <strong>the</strong> control information carried in<br />

<strong>the</strong> internal overhead of <strong>the</strong> SNI-4. The<br />

traffic signals are distributed <strong>to</strong> all <strong>the</strong><br />

Switching Matrix Boards (SMB) in <strong>the</strong> cabinet.<br />

The VCB handles <strong>the</strong> separated control<br />

information for routing. One SWC can<br />

be equipped with up <strong>to</strong> sixteen DMBs.<br />

The SMB is <strong>the</strong> heart of <strong>the</strong> Switch. It contains<br />

<strong>the</strong> Switch for four outgoing SNI-4<br />

signals. In addition, four outputs used for<br />

expansion are located at <strong>the</strong> SMB. These<br />

ports are used <strong>to</strong> interconnect two SWCs<br />

when expanding <strong>the</strong> switch. Each SMB is<br />

connected <strong>to</strong> all DMBs in <strong>the</strong> cabinet. Up<br />

<strong>to</strong> sixteen SMBs can be accommodated in<br />

one SWC.<br />

Each board is equipped with a microprocessor<br />

for board supervision (DP).<br />

Devices<br />

TAU 140, Figs. 6 and 7, interfaces<br />

140 Mbit/s plesiochronous line signals according<br />

<strong>to</strong> CCITT G.703. TAU 140 can be<br />

set <strong>to</strong> different modes, which allows for<br />

branching at 2, 34, and 140 Mbit/s level<br />

and makes switching at <strong>the</strong>se levels possible.<br />

TAU 140 contains five different types of<br />

ASIC. Toge<strong>the</strong>r <strong>the</strong>y handle CMI-decoding<br />

of 140 Mbit/s line signals, framing and<br />

maintenance at <strong>the</strong> 140,34,8 and 2 Mbit/s<br />

levels, buffering and frequency justification<br />

between line signals, synchronisation,<br />

generation of <strong>the</strong> synchronous IVC-4<br />

frame and sending of SNI-4 <strong>to</strong> all three<br />

switch planes. TAU 140 has <strong>the</strong> corresponding<br />

functionality in <strong>the</strong> transmit direction,<br />

completed with majority vote for data<br />

and clock signals from three switch planes.<br />

TAU 140 contains approximately 1.5 Million<br />

used gates, processor block with<br />

256 kWord-PROM, 512 kWord-RAM,<br />

DC/DC converters and a number of analog<br />

high speed blocks.<br />

TAU STM-1E interfaces 155 Mbit/s electric<br />

line signals belonging <strong>to</strong> <strong>the</strong> Synchronous<br />

Digital Hierarchy, and performs demap-<br />

Fig.6<br />

TAU 140<br />

ERICSSON REVIEW No. 3.1992


87<br />

Fig. 7<br />

TAU 140 functionality<br />

FS<br />

Frame Synchronisation<br />

ping of an STM-1 line signal <strong>to</strong> VC-12s and<br />

maintenance of <strong>the</strong>m, pointer adjustment,<br />

buffering, mapping in<strong>to</strong> IVC-4 which is sent<br />

<strong>to</strong> <strong>the</strong> three planes of <strong>the</strong> Switching <strong>Network</strong>.<br />

TAU STM-1 E has <strong>the</strong> corresponding<br />

functionality in <strong>the</strong> transmit direction<br />

completed with majority vote. TAU STM-<br />

1E contains six different types of ASIC.<br />

TAU 140 and TAU STM-1 E are placed in<br />

<strong>the</strong> same magazine which can be equipped<br />

with up <strong>to</strong> sixteen TAUs, arbitrarily<br />

placed and mixed.<br />

TAU 34 interfaces 34 Mbit/s plesiochronous<br />

line signals. Two different types of<br />

ASIC, common <strong>to</strong> TAU 34 and TAU 140,<br />

are used. The signal from TAU 34 in <strong>the</strong><br />

direction <strong>to</strong>wards <strong>the</strong> switch is a triplicated<br />

SNI-3.<br />

I n order <strong>to</strong> make optimal use of <strong>the</strong> capacity<br />

of <strong>the</strong> Switching <strong>Network</strong>, a unit called<br />

Terminal Connection Unit (TCU) and located<br />

between TAU 34 and <strong>the</strong> switch has<br />

been designed, Fig 4. This unit multiplexes<br />

four SNI-3s <strong>to</strong> one SNI-4. Each TCU<br />

board handles two SNI-3/SNI-4 multiplexers<br />

in <strong>the</strong> direction <strong>to</strong>wards <strong>the</strong> Switch and<br />

two SNI-4/SNI-3 demultiplexers in <strong>the</strong> direction<br />

from <strong>the</strong> Switch. TCU is triplicated,<br />

and all Devices with an SNI-3 interface can<br />

be connected <strong>to</strong> <strong>the</strong> Switch by means of a<br />

TCU.<br />

TAU 16x2 interfaces sixteen 2 Mbit/s plesiochronous<br />

line signals. The same<br />

TAU 16x2 is used both in AXD and SDH<br />

systems. Since as many as sixteen line<br />

signals are handled by one board, a protection<br />

function may be used. For every<br />

four active TAU 16x2 is it possible <strong>to</strong> have<br />

an additional standby TAU 16x2. All sixteen<br />

signals can be switched <strong>to</strong> this board<br />

when any one of <strong>the</strong> ordinary boards is<br />

faulty. TAU 16x2 contains three different<br />

ASICs.<br />

The protection function is controlled from<br />

<strong>the</strong> TCU and is performed by <strong>the</strong> Protection<br />

Switching Unit (PSU). The PSU has a<br />

minimum number of components and,<br />

hence, no UP. The PSU is supervised from<br />

<strong>the</strong>TCUs, Fig. 8.<br />

ERICSSON REVIEW No. 3, 1992


88<br />

Fig. 8<br />

Protection Switch Unit structure<br />

The triplication of <strong>the</strong> Switch structure includes<br />

<strong>the</strong> TCUs<br />

ESU is used for synchronisation of <strong>the</strong><br />

AXD with external 2 MHz synchronisation<br />

signals. ESU compares <strong>the</strong> frequency of<br />

<strong>the</strong> incoming 2 MHz signal with <strong>the</strong> divided<br />

signal from AXD's internal clock. The<br />

result of this comparison is sent <strong>to</strong> <strong>the</strong> Central<br />

Processor (CP) for control of <strong>the</strong> clock<br />

modules. The generation of <strong>the</strong> 2 MHz timing<br />

information is also performed by <strong>the</strong><br />

ESU. Frequency division logic is performed<br />

in a Field Programmable Gate<br />

Array (FPGA), and one ASIC is used for<br />

SNI-3 communication.<br />

CTU is used for connection of <strong>the</strong> CP <strong>to</strong><br />

<strong>the</strong> AXD. The CP is connected via E<strong>the</strong>rnet,<br />

and control information is converted<br />

<strong>to</strong> <strong>the</strong> internal Basic Control Pro<strong>to</strong>col<br />

(BCP) embedded in SNI interfaces. One<br />

ASIC is used for SNI-3 communication.<br />

TAU 34, TAU 16x2, ESU and CTU are located<br />

in <strong>the</strong> same magazine. In this magazine<br />

<strong>the</strong>re are three TCUs used by <strong>the</strong><br />

SNI-3 Devices for communication <strong>to</strong>/from<br />

<strong>the</strong> Switching <strong>Network</strong>. Devices can be<br />

mixed arbitrarily, except for <strong>the</strong> CTUs,<br />

which have two specific slots. This flexibility<br />

permits efficient utilisation of available<br />

space when <strong>the</strong> exchange layout is made.<br />

Conclusions<br />

The AXD 4/1 system presented is a system<br />

with great flexibility and a functional<br />

distribution that makes it useful in a wide<br />

variety of network applications. High availability,<br />

thanks <strong>to</strong> <strong>the</strong> use of triplicated HW,<br />

as well as an open architecture have been<br />

key concepts in <strong>the</strong> system design.<br />

The system handles accesses ranging<br />

from 1.5 Mbit/s up <strong>to</strong> 155 Mbit/s, PDH (both<br />

NAS and CEPT) as well as SDH. Switching<br />

can be performed at all levels from<br />

576 kbit/s (9x64 kbit/s) <strong>to</strong> 155 Mbit/s in<br />

steps of 576 kbit/s including all standardised<br />

signal levels.<br />

References<br />

1 CCITT Rec. G.703<br />

2 CCITT Rec. G 707-709<br />

3 <strong>An</strong>dersson, JO.: Digital Cross-Connect<br />

Systems - a System Family for <strong>the</strong> <strong>Transport</strong><br />

<strong>Network</strong>. <strong>Ericsson</strong> Review 67<br />

(1990):2, pp. 72-83.<br />

4 Breuer, H-J. and Hellstrom, B.: Synchronous<br />

<strong>Transport</strong> <strong>Network</strong>s. <strong>Ericsson</strong> Review<br />

67(1990):2, pp. 60-71.<br />

ERICSSON REVIEW No. 3. 1992


ERICSSON<br />

ISSN 0014-0171 Telefonaktiebolaget LM <strong>Ericsson</strong> 92434 Ljungforetagefl. Orebro 1992

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!