Frame Relay - for Faster and More Efficient Data Communications ...

ERICSSONREVIEW1-21992Frame Relay - for Faster and More Efficient Data CommunicationsComputerised System for Quality Inspection in Optical-Fibre Cable ProductionHuman Factors - A Key to Improved QualityCell-voltage Equalisers Series BMP 160In Search of Managed Objects

ERICSSON REVIEWNumber 1-2 • 1992 • Volume 69Responsible publisher Bo HedforsEditor Per-Olof ThyseliusEditorial staff Martti ViitaniemiSubscription Peter MayrSubscription one year $30Address S-126 25 Stockholm, SwedenPublished in English with four issues per yearCopyright Telefonaktiebolaget L M EricssonContents2 • Editorial Changes3 • Frame Relay - for Faster and More Efficient Data Communications12 • Computerised System for Quality Inspection in Optical-Fibre CableProduction20 • Human Factors - A Key to Improved Quality28 • Cell-voltage Equalisers Series BMP 16034 • In Search of Managed ObjectsCoverDehlfi is a computerised system for qualityinspection in optical-fibre cable production. TheDehlfi system is available in three languages:English, Spanish and Swedish. The pictureshows cable attenuation testing at Ericsson'sFibroco factory in Spain.

ERICSSON REVIEWBO HEDFORSPublisher 1992-GOSTA LINDBERGPublisher 1979-1991EDITORIAL CHANGESOn March 1, 1992, Gosta Lindberg andGoran Norrman retired from their respectiveposts as publisher and editor of EricssonReview. The new publisher is Bo Hedfors,senior vice president in TelefonaktiebolagetL M Ericsson and Director of CorporateSystems and Technology. Per-OlofThyselius has been appointed as the neweditor.Gosta Lindberg succeeded Dr ChristianJacobaeus as publisher of Ericsson Reviewin 1979, following his appointment asTechnical Director of TelefonaktiebolagetL M Ericsson. Gosta Lindberg has takenan active part in making Ericsson Reviewan objective and informative technical journaland has been a valuable support to thesuccessive editors during his 13 years aspublisher.Goran Norrman leaves his position as editorof Ericsson Review with this issue.When he succeeded Gosta Neovius, in1986, he had a long and successful serviceto the company behind him. His broadtechnical knowledge and interest intechnology has been very valuable in hiseditorial work, which has been much appreciatedby the publisher and management,and not least by the readers.From the editor's deskThis, the first issue of Ericsson Review aftermy appointment as editor, is also thefirst issue to be published in English only.Subscribers who have previously receivedthe French, Spanish or Swedish editionswill from now on receive only the Englishedition. When the Board of Ericsson Reviewrecently decided to stop publishing thenon-English editions, their prime motivewas the awkward delay between the firstand last editions - even considering thatEricsson Review has never had the ambitionto be a technical news magazine.Another basis for the decision was, of cource,our firm belief that the majority of ourFrench, Spanish and Swedish readers arefamiliar with English.Ericsson Review was first published in1923, and as early as 1924 it became afive-language magazine. Since the discontinuanceof the German edition, it hasbeen issued in four languages. The ambitionfrom the start was to publish a technicalmagazine to spread knowledge ofEricsson products and Ericsson technologyto our customers' technical staffs andto the technical staffs within our own worldwideorganisation. This objective has neverbeen changed. The major content ofEricsson Review has always been descriptionsof our different systems and productsand outlines of our technical objectives.Occasionally, the magazine has also containednews from Ericsson's business operations,large telecom exhibitions andconferences.Scientific material from our applied researchactivities had once an exclusiveoutlet in the form of Ericsson Technics.This magazine was integrated with EricssonReview in 1978. Unfortunately, it turnedout that scientific material met with difficultiesin competing for space in EricssonReview. The readers are the only oneswho can change this situation, and we willcertainly appreciate it if they inform us oftheir views on Ericsson Review and itscontents.During my 38 years of working with differentparts of Ericsson, I have myself beena regular reader of Ericsson Review. Takingoffice as editor I will do my best tocreate a readable and informative magazinein the years to come.GORAN NORRMANEditor 1986-1991To avoid the situation of our English readersreceiving the first issue of Ericsson Review1992 prior to the last 1991 issue, wehave chosen to publish a double-issue:1-2, 1992.PER-OLOF THYSELIUSEditor 1992-ERICSSON REVIEW No. 1-2, 1992

Frame Relay - for Faster and MoreEfficient Data CommunicationsKajsa LundfallKAJSA LUNDFALLEricsson Business CommunicationsToday, workstations and personal computers have become everyday tools. This fachas radically changed the conditions for data communications networking. Theprocessing power has moved from computer centres to users' desks. Thecommunications pattern is also different: it is no longer a matter of exchangingtransactions of the question/answer type, but of transferring large volumes of data inthe form of "bursts". At the same time, technological improvements in the transportnetworks have resulted in better data transmission quality. This, in turn, reduces theneed for error-correcting functions in the communication protocols; instead, effortsfocus on the ever-growing need for higher transmission speed. Efficient handling ofthe new situation requires new protocols and new communications systems.The author describes the way technological advances and a growing market demandcontribute to the development of new communications protocols such as FrameRelay. The article compares Frame Relay to packet switching (X.25) and describeshow Frame Relay works.data communication systemscomputer networkspacket switchingprotocolsFig. 1Traditional centralised multi-user systemThe environment is primarily designed for moderatetransmission speedIn the last few years we have been witnessinghow developments in the data processingand data communication fieldshave changed the use of computers inmany companies. Due to the low cost ofpersonal computers and workstations,practically every desk is equipped withthese tools today. The typical informationsystem is no longer a centralised multiusersystem, but a decentralised systemmade up of interacting Local Area Networks(LAN). These inter-networks (composedof various LANs) place new demandson the Wide Area Networks (WAN).Centralised systemsThe core of a typical transaction-orientedsystem is a central host computer whichserves a large number of terminal usersconnected via concentrators and communicationprocessors, Fig. 1. All exchangeof information between a terminal and itsenvironment takes place via the host computer,which also accommodates all processingpower. Most of today's applicationsand communication networks havebeen designed for operation in this environment.The typical transaction is a text string (aquestion) which is sent to the host computer,and a full screen (an answer) which issent from the host computer to the terminal.Very little graphics - if any at all - isused, and the amount of information transferredis small. Moderate transmissionspeed can be used, even when the requirementsfor short response time are stringent.Distributed systemsThe present development in computertechnology is creating a new computer environment- based on Local Area Networksand distributed, interacting applications- which supplements the traditionalhost computer configuration. LANs usesimple protocols and provide high transmissionspeed for data communicationsover short distances. Gone is the centralisedarrangement, with a master-slave relationshipbetween computer and terminal,in which all information must pass througha mainframe.Local Area Networks and personal computershave developed along parallel lines.ERICSSON REVIEW No. 1-2, 1992

4Fig. 2Distributed computer environment based on LocalArea Networks using internal transmissionspeeds of several Mbit/s. This places exactingdemands on the datacom network, which is requiredto provide cost-effective LAN-to-LANcommunicationsFig. 3The transfer time varies with the link transmissionspeed, which in turn is dependent on thetype of file transferred. When programs or filescontaining graphics are to be transferred, a linktransmission speed of 64 kbit/s will result in unsatisfactoryresponse timeFig. 4A client-server application makes efficient use ofthe combined processing power provided by thehost computer and the personal computer/workstation.This makes exacting demands on thenetworkType of file 64 kbit's 2 Mbit*2 text pages 1/3s 1/100 s1 page of spreadsheet 6 s 1/5 s1 drawing page 15 s 1/2 sLarge program file 1 min 2 sThe design of protocols for Local Area Networkshas been based on the demand forthese networks to be capable of handlingshort response times and transferringlarge amounts of data (file transfers, etc).This has been achieved by building LANswith internal cabling for data transmissionat high speeds. The Wide Area Networks,on the other hand, use the existing transmissioninfrastructure, which is primarilydesigned for telephone traffic.Increasingly, applications in the LAN environmentare introducing more graphics,which accentuates the need for large fileinformation transfer at high speed. An LANcan manage transmission speeds of 4,10or 16 Mbit/s, or even more. The transfertime is short even for large information volumes,and databases, program librariesand advanced I/O devices serving the entirenetwork have become a reality. In atypical LAN, data is processed on theuser's own personal computer or workstation.The user connects himself to the networkonly to print out files or to transfer filesor programs to a server or retrieve themfrom that server. Information is transferredin the form of short bursts at a relativelylow frequency.The need for communications outside thelocal environment requires adistributed infrastructure,as exemplified in Fig. 2. ButLAN-to-LAN traffic in Wide Area Networksis not altogether uncomplicated. Both technologyand economy make the transmissionspeed on the links between local environmentsa restricting factor. Fig. 3 illustratesthe differences in transfer time fordifferent file types transferred on a link at9.6 kbit/s, 64 kbit/s and 2 Mbit/s.New architecturesThe processing power offered by today'sPCs and workstations is sufficient for mostapplications. Databases, program librariesand many other applications are characterisedby the need for common resourcesfor storing information or for tellingusers where a certain type of informationcan be accessed. In LANs, thesecommon resources are available in servers.In most PC networks, however, servershave so far been used only as an extrahard disk for storage and printout of data.All processing of data has taken placein the local workstation. In order to makemore efficient use of the processing powerERICSSON REVIEW No. 1-2, 1992

Fig. 5The X.25 protocol places the users' data In packets,which are sent one after the other on theline. In this way, maximum use Is made of the capacityof the connection. Each connection is virtual,which means that several sessions can usethe same physical line simultaneouslyPacket, virtual channel 1Packet, virtual channel 2Fig. 6X.25 switching permits maximum utilisation ofthe capacity of the connection, all of which isavailable for use by the party who needs it forthe momentBandwidthprovided by the server, a new type of applicationhas emerged in which the serverand the workstation share the processingwork. This new way of designing applicationsis called the client-server model,Fig. 4.In the future, the client-server concept maypermit use of the network as an externaldata bus between workstation and server,a configuration which will place very highdemands on fast and safe informationtransfer in the network.The client-server model does not requirethat the user knows where the requestedinformation is stored. It may be in a serverin his own LAN or in another LAN thatforms part of the network. The location ofdata is transparent to the user: he addressesthe service he wants to use, leaving itto the network to keep track of where it isexecuted. These conditions set the requirementsfor LAN-to-LAN traffic thatcharacterise future development of longdistancenetworks.Improved transport networksThe ongoing modernisation of public telecomnetworks replaces analog transmissionto a great extent with more robust digitaltechnology, including extensive useof fibre optic cables that are immune to interference.The result is significantly decreasederror rate and a dramatic reductionin the leasing cost of digitalconnections for transmission speeds of 64kbit/s and upwards.Planning for highertransmission speedThe Local Area Networks have the capacityrequired to meet the need for highspeedconnections with very short delay,which is the result of increased workstationprocessing power and larger informationvolume in each transfer. The digitalisationof public networks makes it possibleto meet this need in Wide Area Networkstoo. However, a traffic pattern characterisedby short information bursts and hightransmission speed means poor utilisationof point-to point connections between twoLocal Area Networks. Communication protocolswhich permit interleaving of trafficfrom a large number of users over a singlecommunication line, are a prerequisitefor cost-effective handling in Wide AreaNetworks.X.25So far, X.25, or packet switching, has beenthe leading established communicationsprotocol capable of interleaving data frommany users over a single line. Packetswitching is used both for public servicesand in private networks in order to ensuresatisfactory economic utilisation of networktrunks, Fig. 5.X.25 is an effective method for meetingintermittent data communication needs,and today's X.25 networks have worldwidecoverage. X.25 does not stipulate pre-allocationof a specific portion of the bit flowto a specific connection between two users.As opposed to time division multiplexing,this means that the whole bit flow isavailable to the user who has data to sendat the moment, Fig. 6. This also means thatseveral so-called "virtual connections" areset up in parallel over the same physicallink. Each of these connections is allocateda logical channel: a unique identity indicatedin the data packet header.Two types of connection are defined inX.25: Permanent Virtual ConnectionsERICSSON REVIEW No. 1-2, 1992

Fig. 7A company's data network should be designedso as to permit both existing and future applicationsto use the same physical networkBox 1Standardisation of Frame RelayFrame Relay as a communications standard wasfirst proposed by CCITT in the late 80's and thenpresented in Recommendation 1.122. At that time,the US standardisation body ANSI had also beganto take an interest in this field, but the resultof both bodies' work was rather long in coming.Northern Telecom, DEC, CISCO, and Stratacomtherefore took the matter in their own hands, submittingat the end of 1990 a proposal for a standardbased on the existing - but still incomplete -proposals drawn up by CCITT and ANSI. This proposalalso contained a "Local Management Interface",LMI, which describes the exchange of informationon link status between the DTEs andthe network. To complete the work and force thepace of standardisation, Frame Relay Forum wasestablished in January, 1991.More than 70 companies including Ericsson arenow members of Frame Relay Forum, which doesnot do any standardisation work of its own but submitsproposals to existing standardisation bodiessuch as CCITT, ANSI and ETSI. The formation ofFrame Relay Forum speeded up the standardisationwork, and the CCITT and ANSI standardshave now reached an appreciable degree of development,including the specification of LMI functionality.CCITT's recommendations include I.223 (whichdescribes the service), Q.922 (describing theframe) and 0.933 (signalling). ANSI's versions ofthese standards, which are basically equivalent tothose specified by CCITT, are called T1.606 (theservice), T1.618 (frame) and T1.617 (signalling).(PVC) and Virtual Connections (VC). ThePermanent Virtual Connection is a networkpath which is predefined by the networkoperator. The procedure for setting up aVirtual Connection includes indication ofthe address by means of a Network TerminalNumber, which is unique to each setof equipment connected. No network operatoris involved in this set-up procedure,which results in the establishment of a networkpath that remains connected until thesender or receiver orders its release.The X.25 protocol is based on the storeand-forwardprinciple. Packets enteringthe network are stored in a buffer andpassed on in the order of their arrival. Thetransfer over each individual link is monitoredseparately, and an error results in initiationof retransmission on the link. In thisway, correct data transfer is possible evenin networks with heavily disturbed connections.There is one disadvantage, however;the protocol handling requires muchprocessing power, which causes delayand limits the available transfer speed. Today,a typical X.25 network cannot efficientlyutilise link speeds above 64 kbit/s.But the development towards higher transmissionspeed is evident in the X.25 networks,too, and follows the developmentof microprocessor capacity.Due to their limited transmission speed,X.25 networks have not been consideredeffective enough for the interconnection ofLANs in a wide-area environment. Pointto-pointconnections have been preferredwhen high transmission speed has beenthe prime requirement. Although point-topointconfigurations provide the requiredtransmission speed, they are not usuallycost-effective. The reason for this is thatLAN interconnect applications generatedata in the form of bursts; that is, they utilisethe available capacity for very short periodsonly. In fact, the connections are passivemost of the time, resulting in a low degreeof utilisation.Network solutions of today andtomorrowThe new and the traditional, centraliseddatacom environments usually coexist ina company's organisation. Today, manycompanies therefore have a datacom environmentmade up of separate infrastructures.A uniform communications environmentmust be able to meet the needs ofboth LAN-to-LAN connections and of thelarge number of terminals connected tohost computers, Fig. 7.Frame Relay - a newalternative in datacommunicationsThe purpose of introducing Frame Relayis to meet the new requirements, and mainlythose resulting from the development ofLAN-to-LAN communications. The standarddefines a simplified packet switchingservice which meets the need for a simpleFig. 8The Frame Relay standard specifies an interfacebetween a user and a Frame Relay networkERICSSON REVIEW No. 1-2, 1992

Fig. 9A typical Frame Relay network with DTEs, networknodes and public Frame Relay servicesFig. 10Frame Relay networks use permanent virtualconnections, which means that a physical connectioncan provide several logical channels forsimultaneous usedata transmission protocol focusing onhigh speed and requiring a minimum oferror-correcting and flow-controlling functions.The standard, which is described in bothCCITT and ANSI recommendations, definessignal and data transmission at linklevel, OSI level 2, in the interface betweenuser equipment and network, Fig. 8. (Cf.X.25, which comprises the first three OSIlevels and uses a frame at level 2 for thetransmission of a packet at level 3.) Thestandardisation work is briefly described inBox 1.How Frame Relay worksA Frame Relay network is made up of networknodes and user equipment, DTE.s(Date Terminal Equipment) connected tothe network. The DTE, e.g., a personalcomputer, a gateway, a router or a hostcomputer, is provided with the interfacedefined for Frame Relay, Fig. 9.The sending DTE transmits frames to thenetwork. Each of these frames contains anidentification code (Data Link ConnectionIdentifier, DLCI). All network nodes alongthe path to the final destination contain informationindicating the outgoing channelto which a frame with a specific identificationcode is to be sent. The path betweenthe sending and receiving DTEs has beenpredefined by the network operator. Thistype of connection - a Permanent VirtualConnection - is so far the only one definedin the standard. Virtual connections are notincluded in today's Frame Relay function,but future versions of the protocol are expectedto allow such connections too.The network node routes to the right destinationsthe frames sent from a DTE. Thenetwork node reads the identification codeof the incoming frame and sends the frame(without changing it) on the outgoing channelindicated in the node's routing table.This outgoing channel can either be a connectionto another network node (in whichcase the procedure described above is repeated)or a connection directly to the terminatingDTE. However, the way theframes are handled internally in the networkis not defined in the standard.As in X.25 switching, the use of severalidentification codes permits several parallelsessions in different directions to coexiston one physical connection, Fig. 10.In this way, a DTE can communicate simultaneouslywith different destinationsover the same physical connection to thenetwork. This is necessary if the DTE is acommunications port in an LAN, but it is alsoan attractive solution in cases where theDTE is a personal computer or workstationthat uses several simultaneously activewindows.Handling is simple because the protocoldoes not include any error-correctingmechanism. Speeds of several Mbit/s canbe used without requiring unreasonableprocessing capacity for link handling.Today's version of the standard stipulatesa maximum link transmission speed ofERICSSON REVIEW No. 1-2, 1992

8Fig. 11The frame format used for Frame RelayFig. 12The address field of the frame used for FrameRelay consists of a DLCI and control /check informationfor flow handlingC/REAFECN6ECNDECommand/Response indication bitExtended Address bitsForward Explicit Congestion Notification bitBackward Explicit Congestion Notication bitDiscard Eligibility bit2 Mbit/s, but considerably higher speedscan be expected.Frame StructureThe frame format used for Frame Relay isbased on LAPD (Link Access Protocol D),which has been specified by CCITT(Q.922) and ANSI (T1.618). The frame,Fig. 11, is used to carry both user data andidentification code (DLCI). The header ofthe LAPD frame consists of two byte andcontains - in addition to the 10-bit DLCI -information bits for flow control (BECN,FECN and a DE bit; see below).The length of the information field in theLAPD frame is adjustable to a maximumvalue defined for the service concerned.The length is normally set to a value atwhich the information from the application- a TCP/IP packet, an SDLC frame, anX.25 packet, etc - can be carried withouthaving to be split.The frame ends in a Frame Checking Sequence,FCS, which is used to check thatthe frame has been transferred correctly.As opposed to X.25, where an erroneousFCS triggers a retransmission procedureover the link, the network cannot guaranteethat all frames will reach their destination.Frames indicated as erroneous by theFCS are discarded, but this does not necessarilymean that the user loses any data.Checking to ensure that the DTE-to-DTEtransmission has been error-free is usuallya mandatory function in the higher-levelprotocol used for information transfer inthe application concerned. If an error occurs,data must be retransmitted throughthe entire network, and this obviouslytakes longer time than retransmission onlyover the link on which the error occurred.An approximate limit where Frame Relayis effective is a Bit Error Ratio (BER) of nomore than 1/1,000,000 on individual links.This means that modern, digital transportnetworks are well suited for the simplifiedprotocol, but if the chain contains an analoglink, a short delay time for each linkmust be weighed against the expected frequencyof end-to-end retransmission.Flow control and handling of overloadsituationsThe handling of an overload situation in aFrame Relay network demands less of theswitching equipment than handling the samesituation in an X.25 network. In the lattercase, the flow is controlled by the networknode sending confirmation messagesin the backward direction to the DTE,to the effect that sending may go on. If nomessage is forthcoming, the DTE will stopsending new packets until it has receiveda confirmation. This rule ensures a controlledand stable system, at the cost of delayedtransfer.ERICSSON REVIEW No. 1-2, 1992

9Fig. 13Applications which use sophisticated graphicsand therefore generate very large files are becomingmore and more frequent. Transferthrough the network requires a transmissionspeed of several Mbit s to obtain a reasonable responsetime. Local networks are well adapted tothis requirement, but in wide area network transferthere may be inconveniences, since thesenetworks did not initially have the infrastructurerequired for high transmission speeds.In pace with the digitalisation of the wide areanetworks, new communication protocols are nowevolving, the development of them being to alarge extent driven by the demand for efficientLAN-to-LAN communication. Frame Relay is oneexample of this trendFrame Relay is based on the opposite principle:the network node sends signals tothe DTE only in overload situations. Thismeans simplified handling, and no delayswill occur as long as the traffic is undisturbed.A Frame Relay network reacts tooverload in two steps:1 When the node registers a tendency towardsoverload, the network requeststhe DTE to reduce its data flow by sendinga Forward/Backward Explicit CongestionNotification (FECN/BECN) consistingof the check bits of the frameheader, Fig. 12. The network takes noaction but leaves it to the DTE to decideon the best way of controlling its dataflow.2 If overload occurs, the network will discardframes.The standard does not say how the networkwill define the concept "tendency towardsoverload". To exemplify, the nodecan count the number of frames queuedfor access to outgoing lines, or check theCPU load, and then send an FECN/BECNsignal when a preset limit value is exceeded.As has been mentioned already, a discardedframe in the network will not result inloss of information. The higher-level userapplication protocol checks that the terminatingDTE sends a confirmation of datahaving arrived. If no such confirmation isreceived, retransmission through the entirenetwork will be initiated, resulting in undesireddelays, i.e. longer response time.Transferring the flow control to the DTEclearly involves a risk, since most oftoday's frame-relay-compatible DTEs cannothandle FECN and BECN signals. Somesuppliers of equipment capable of handlingFECN/BECN signals recommend usersto switch off the function in order toavoid falling behind when competing fortransport capacity with equipment withoutthat function.For the simplified flow handling to functionproperly, each connection from a DTE isallocated a Committed Information Rate,CIR. This rate is affiliated to the virtual connectionthrough the network, and the ratevalue determines how much data the DTEcan send on this connection during a givenperiod of time.The standard does not give a strict definitionof how to use the CIR value. It could,for example, be chosen so that each PVCwould be guaranteed access to a given bitflow, which would basically be the sameas using a leased line. A 2 Mbit/s connectionwould then be able to carry 32 PVCsof 64 kbit/s each. However, a more rationaluse will be ensured by starting from thestochastic traffic characteristics and interpretingCIR as a stochastic value of the averageamount of data that a PVC is expectedto carry during a given period of integration.Since the traffic is in the form ofbursts, there is little probability of all sourcessending at the same time. In otherwords, we can allow a PVC to momentarilyutilise a relatively large portion of theavailable transmission capacity, providedthe stipulated average value is kept. At anintegration time of, say, 0.5 s, a CIR valueof 256 kbit/s for a PVC will mean that theDTE can use this PVC for sending one ormore bursts of altogether 128 kbit duringeach 500 ms interval.Assigning CIR values will facilitate calculationof the traffic flow through the network.When the CIR value is exceeded,the Discard Eligibility bit in the frame head-ERICSSON REVIEW No. 1-2. 1992

10er (DE, Fig. 12) should be set to 1. If thisis not done in the DTE, it will be done bythe access node of the network. If overloadoccurs, queued frames whose DE bitis set will be discarded before other, unmarkedframes. The DTE is allowed to exceedthe assigned CIR by apredeterminedvalue, the Excess Burst Size. If largeramounts of data are being transmitted, allexcess frames will be discarded, regardlessof the load situation in the network,Fig. 14.An overload situation may remain in thenetwork even after all frames with set DEbits have been discarded. In order to furtherorganise the continued discarding, itis also possible to establish an order of priorityfor the different virtual connections.This is up to the manufacturer or operatorto decide, since the Frame Relay recommendationdoes not define any priorities.Supervision of the interfaceSupervisory frames with reserved DLCIaddresses are used for two purposes: forsupervision and for exchange of informationbetween the network and the DTEs.The link status and the information indicatingactive or inactive link are transferredin these frames. Indication of the status ofpermanent virtual connections and executionof changes in the DLCIs used areother functions of these informationframes.Implementation of FrameRelay-Frame Relay can be implemented in differentways:• As part of a private packet switching networkthat also supports other types ofcommunication, such as X.25 and SNA.• As a virtual private network consisting ofan operator network section dedicated touse by a specific customer, a companyor an organisation.• As a public Frame Relay service providedby the PTT or some other operator.• As a hybrid solution: a private networkcovering the primary geographical areaof a company, and peripheral units connectedvia public Frame Relay services.Fig. 14Handling an overload situation in a Frame Relaynetwork. Each Virtual Connection through thenetwork is assigned a Committed InformationRate (OR). The CIR value is generated by thenetwork and indicates that the DTE on this connectionis capable of sending a given amount(Committed Burst Size, Be) of data during a givenperiod of time (Tc), CIR=Bc Tc. Provided thenetwork is not overloaded, the DTE is allowed tosend a somewhat larger amount of data (ExcessBurst Size, Be) on the connection. But if an overloadsituation occurs, this extra amount will becancelled and any frames above the CIR valuediscarded. If the DTE exceeds the maximum permissiblequota (Be + Be) for a given connection,the network will always discard the excess numberof framesCommitted RateMeasurement IntervalERICSSON REVIEW No. 1-2, 1992

11Frame Relay in ERIPAXEricsson's packet switching system ERIPAX iscontinuosly improved to meet the emerging demandsfor an enhanced packet switching service.Frame Relay services is currently added toERIPAX functionality.The implementation of Frame Relay in ERIPAXconforms to applicable international standards butalso includes additional services of value to theuser. To follow and influence the Frame Relaystandardisation process, Ericcson has been amember of the Frame Relay Forum since itsfounding in 1991.Frame Relay will be offered as an integrated softwarefunction in the ERIPAX node, allowing packetswitching and Frame Relay services to co-existin the node. Packet-switched services, such asX.25 and SNA/SDLC, may use Frame Relay as abearer service within the network and - withoutany additional investments - benefit by the higherspeed and capacity of the Frame Relay service.If so required, certain parts of a network can berestricted for use of one of the services only.Frame Relay services in ERIPAX will be describedby Goran Ingemarsson and Bo Karlander in an articlein the next issue of Ericsson Review.A study of the 1000 largest companies inthe USA, made by the Yankee Group andpublished in Data Communications International,January 1992, shows that 64 %of these companies were planning to installcommunications systems based onFrame Relay technology within the nexttwo to five years. Western Europe is probablysome years behind the developmentin the US, where deregulation has playedan important role and where LANs are inmore frequent use than in Europe.Advantages of Frame RelayFrame Relay is basically a simplified implementationof the X.25 protocol, aimedat facilitating a marked increase of thetransmission speed on network links andDTE lines. The advantages of the X.25 protocolremain unchanged, i.e. efficient trafficinterleaving and the possibility of establishingseveral connections over thesame physical channel. The consequencesof the removal of the error-correctingfunction and the greatly simplified flowcontrol on individual network links aremore than compensated for by the improvedtransmission quality obtained byextensive digitalisation of the public transmissionfacilities.Frame Relay is therefore a logical consequenceof the technological development— in two respects. Firstly, in terms of thenew needs emerging in the increasinglyadvanced computer environment. Secondly,in terms of faster and cheaper transmission(characterised by a lower level ofsensitivity to interference), which is the resultof improved digital technology andwhich is probably also enhanced by the ongoingderegulation of public telecom operations.Another important factor is the ease withwhich Frame Relay can be introduced intoday's data communications environment.A PC, a workstation or a server becomesa Frame Relay DTE after addingsoftware of low complexity. Frame Relay,just as X.25, is a specification of the interfacebetween the DTE and the network,and many of the functions offered today bythe X.25 networks (such as alternativerouting and charging) can be directly appliedwhen implementing Frame Relayfunctionality in an X.25 node.No network-to-network interface is specifiedin the present version of the standard,but the standardisation work is in progress.The result of this work is likely to furtherstrengthen the position of Frame Relay asan attractive public and private service.Effective LAN-to-LAN communication isthe driving force behind the introduction ofFrame Relay, but once the Frame Relaynetworks have been implemented it will bepossible to offer better solutions to otherdatacom needs too. Therefore, it is not surprisingthat Frame Relay has been widelyaccepted, both by operators and by suppliersof equipment for corporate datacomnetworks.ERICSSON REVIEW No. 1-2, 1992

Computerised System for QualityInspection in Optical-Fibre CableProductionGoran Nilsson and Solweig ViklundIn order to meet the market's increasingly stringent quality requirements - thosespecified in SS-ISO 9001, for example - Ericsson Cables AB in Hudiksvall, Sweden,has developed a computerised system for quality inspection in optical-fibre cableproduction. The system, called Dehlfi, has been in operation at the Hudiksvall factorysince 1988; another version has been used at Ericsson Cables' Spanish factory sinceits inauguration in July, 1991.The authors describe the system's design, functions and field of application.optical cablesoptical fibresquality controlinspectionstatistical process controlFig. 1Workplaces, terminals and test stations in theDehlfi systemQuality is a concept of vital importance toEricsson Cables. For a product to be improvedand optimised, its characteristicsmust be known. Knowledge of a product'squality is largely a matter of information -in design, sales and administration - andthis information must be easy to understandand make use of.A considerable number of parameters areneeded in the design and manufacture ofoptical fibre and cable. Large amounts ofmeasurement data have to be presentedin an attractive and comprehensible way -even to our customers. To achieve ourquality objectives we must have a systemthat can receive, store, process andpresent the necessary information. That iswhy Ericsson Cables has developed acomputerised system called Dehlfi (Databaseat Ericsson Hudiksvall for optical Fibreand cable).Some requirements have been given toppriority in the development of the system:- data must admit of being automaticallyentered and stored in the database- the database must be user-friendly andeasy to modify- the database must allow the use of complexrelations- printouts - especially those in the formof customer documents- must be of highquality.The manufacture of fibre and cable is describedin References 1 and 2.System descriptionDatabaseThe core of Dehlfi is a minicomputer. It isan HP 9000/835S, with RISC architectureERICSSON REVIEW No. 1-2, 1992

13GORAN NILSSONSOLWEIG VIKLUNDEricsson Cables AB(Reduced Instruction Set Computing), capableof performing 14 million instructionsper second (14 MIPS). It has an 8 Mbyteprimary memory and uses two hard disksof 571 and 670 Mbyte respectively. Twomagnetic tape stations are provided forbackup copying.The HP machine uses a UNIX operatingsystem. The database is developed in HPAllbase/4GL, which is a special databaselanguage, and has ISAM data files (IndexSequential Access Method). A number ofexternal routines are written in the two programminglanguages C and Pascal.Fibre-optic networkA star-shaped network connects the minicomputerto terminals, test stations and otherequipment, Fig. 1. Cable designers,system developers at the technical departmentand operators at the manufacturingand quality inspection departmentscan access the system via this network. In1991,12 test stations, 10 terminals and variousPCs and printers were connected tothe network.The network is made up of fibre-optic modemsand fibre-optic multimode cables,and designed according to ElA's standardRS 232 (CCITT V24/V28).The units of a test station are interconnectedvia the instrument bus HP-IB (Hewlett-Packard Interface Bus), by other instrumentmanufacturers called GPIB (GeneralPurpose Interface Bus). This is a standardised,general-purpose bus which connectsone or more test instruments to acontrol unit, thus forming a computer-controlledtest system.The communication routines are written inHP Basic, and data is transferred in theform of text sequences in ASCII code(American National Standard Code for InformationInterchange). This is not the fastestcommunication protocol available, butthe transfer time is insignificant in comparisonwith the test time.Test stationsA test station is equipped with a test instrument;for example, an OTDR (Optical TimeDomain Reflectometer) and the associatedcomputer HP 9000/332, a printer and aplotter, Fig. 2. Each test station has a specifictest program in HP Basic. The menucontrolledtest program instructs the operatorand reads the instrument. A teststation may either communicate directlywith the database or operate in standalonemode.User interfacesThe Dehlfi system has menu-based userinterfaces. Each operator category has aseparate, tailor-made set of menus, andeach operator has a unique password.Personal computers have Windows userinterfaces, Fig. 3.The introduction of Dehlfi meant considerablesaving of labour, for one thing becausethe system eliminates the time-consumingand monotonous manual recordingof test data at terminals.Fig. 2Example of test station equipmentPrinterComputerDiskette driveTest instrumentHP ThinkjetHP 9000/332HP 9122OTDRERICSSON REVIEW No. 1-2,1992

Box 1Code39, the first alphanumeric bar-code ever tobe designed, is now the most frequent code in industrialapplications. The code consists of 44 discrete,unique patterns. Each pattern is made upof 9 elements: 6 narrow and 3 wide; hence thename Code 3 of 9. Each sequence starts and endswith an asterisk, "*".An enhanced version, Extended Code39, hasbeen introduced to permit coding of all 128 charactersin the ASCII table. Here, $, %, / and + arecombined with other characters to form a completeASCII table. Example: "a" is coded as "+A".Data flowTraceability is important when producingoptical fibre and cable. The Dehlfi systemprovides an efficient tracing function: it candetermine to which cables the fibres drawnfrom a given preform belong and, in theopposite direction, the original preformsused for a given cable, Fig. 4.PreformsPreforms delivered to the factory are registeredon receipt, i.e. each preform is givena unique identification number - a preformid.This number, together with data onmanufacture, supplier and type of fibre, isstored in the database. The number willthen form part of the fibre identity.Fibre drawingWhen the fibre is drawn, a drawing recordis prepared which contains run data andmachine data. The finished fibre is thensubjected to a strength test, and the fibrelengths obtained are registered in the database,which generates a unique identificationnumber, a fiberid, for each fibre. Thefiberid is printed - both in plain text and inbar-code - on a label printer and stuck onthe fibre drum. All registration of preformsand fibres is done by staff serving at thedrawing towers.Fibre testingThe registration of the fibre is followed byquality inspection (primary testing) at fourdifferent test stations in the test room:- Attenuation test, OTDR- Spectral attenuation and mode field test,FOA (Fibre Optical Analyser)- Dispersion test, CD3 (Chromatic Dispersion)- Geometrical test, i.e. testing of the geometricalcharacteristics of the fibre.The operator carries out the tests asprompted by the test program. Test datacan be stored in three ways:- Automatically in the database- On diskettes- In the form of printouts only.A check is made to ascertain that the fibrehas been registered, and then the test datais stored in the database. Storage on diskettesis primarily a backup routine to safeguardthe system against faults, such asbreaks in the communication betweendatabase and test station. A program routineis provided which automatically readstest data from diskettes to the database.The system automatically compares theprimary test results of each fibre with spec-Fig. 3Windows' Spanish user interfaceERICSSON REVIEW No. 1-2, 1992

15Fig. 596-fibre ribbon cableified values. Approved fibres are put instock.Purchased fibrePurchased fibre is accompanied by test datastored on diskettes, whose contents areread to the database via a PC. At the sametime, a cross-reference register is preparedwhich shows the supplier's andEricsson's fiberids. A label showing fiberid,length and storage-location is printed onthe label printer and stuck on the fibre drum.All labels - Ericsson's as well as those ofthe fibre suppliers - have bar-code informationwhich facilitates handling and minimisesthe risk of mistakes.Code39 is the standard bar-code (symbology),and Extended Code39 is also usedwhere required. See Box 1.RibbonsRibbons are manufactured for use in ribboncable.The ribbons are placed in thecable core, with up to four ribbons in eachslot, which gives 96 fibres per cable, Fig. 5.In the manufacturing phase, each ribbonis given a unique identification number - aribbonid. The quality of a ribbon is inspectedby comparing its test data with the dataof each fibre in the ribbon.Storage managementThe Dehlfi system also handles the storageof fibres. The database indicates thephysical storage-location of each fibre. Itonly takes a couple of minutes to producean inventory report showing the currentquantity of each type of fibre. This informationis sent once a week to the productionplanning department. The financial departmentalso receives weekly reports onstorage transactions, i.e. the amount of fibresreceived and withdrawn.Secondary coating, stranding andjacketingFibres are withdrawn from store for secondarycoating and stranding. For eachcable, the fibres selected in the databasetally exactly with the customer's specification.For this purpose, the system uses asearch routine which selects fibres with therequired properties and indicates theirphysical storage-location.

16Fig. 6Cable attenuation test at Ericsson's Fibroco factoryin SpainAfter the secondary coating has been applied,the fibre is subjected to quality inspection(secondary testing). This test iscarried out according to the same routinesas those applied in the primary OTDR test.The test data is usually stored directly inthe database.Terminals at the stranding lines registerthe fibre contents of each cable, and thecable is given a unique identification number— a cableid. The cable line operatorsalso handle the registration of cables. Tofacilitate work and minimise the risk of incorrectregistration, all terminals have abar code reader and a bar-code menu.Stranding is followed by intermediate testing,jacketing and delivery inspection.Intermediate testing comprises attenuationmeasurements of all fibres in thecable. The test data is stored in the database.The values obtained in these testsand in secondary attenuation tests arechecked against the primary attenuationvalues. The program issues a warning ifthe difference exceeds the tolerancesspecified.Delivery inspectionA cable which has passed the delivery inspectionis ready for delivery. A deliveryrecord for the customer is printed on a laserprinter.The delivery record documents the opticaland physical parameters of the cable, Fig.7. Since a cable may contain hundreds offibres, the delivery record will take up atleast two size-A4 pages. The first pagecontains statistical data, and the subsequentpages give test data of each individualfibre.The layout of the delivery record will vary,depending on the customer's requirementsand on the type of cable delivered.To rationalise handling, customers may receivecable parameters stored on diskette.StatisticsTwo main system requirements were definedwhen Dehlfi was put into service inApril, 1988: Automatic transfer of test datafrom test stations to the database, andprintout of complete delivery records. Duringthe subsequent development of theDehlfi system, a large number of quality inspectionparameters have been added,and the possibilities of using these parametersin quality work have also increased.Three program categories are available forthe processing of data extracted from thedatabase:- Search, processing and presentationComplete reports for recurrent follow-upare prepared- Search and processingData is prepared for presentation bymeans of a special graphicsprogram, e.g. Graph-in-the-Box- SearchRoutines which only locate predetermineddata for processing and presentationlater on by means of statistics software,e.g. Minitab.Minitab is a software package for advancedstatistical processing. EQFOS(Ericsson Quality, Design and StatisticalAnalysis of Experiments) is one of its users.Graph-in-the-Box is a very handygraphics program which also contains somestatistics functions.The software packages developed byEricsson Cable's own specialists includeone for collecting fibre data during a pre-ERICSSON REVIEW No. 1-2, 1992

VK 1523-402033CERICSSON CABLESRECORD OPTICAL CABLEPrepared I Date Order no:HL/ECA/T/VK Margit Dahlberg 17 Dec 91 29595-080Approved [ Checked Manufacture no: ' KB no:| S40E00700 7779Cable : GNLWLV 24x10/125 EM Length ordered. . : 4000tested. . . : 3978Fibre : OVD/Q shipped. .: 3973Cable id.. . : 9035007 Customer :Drum no....: 3483 order..: KI 199-6755spec no:SPECIFIED VALUES [ TEST VALUES| mean | min | maxMode field diam P1I 9.4 ± 0.8 Jim 9.2 9.6H _ Cladding diam 125 ± 3 Jim 124 126J I Concentricity error ^ 1.0 iim 0.32BBI " Cladding non circ

18Fig. 9Installing the Dehlfi system at Ericsson's Fibrocofactory in SpainFig. 8Example of printout for monitoring fibre production.The program was developed by EricssonCable's own software specialistsFi le name: er2Date : 92-03-03Mean: 0.34Standard deviation : 0.007Number of values.. .: 1732ATTENUATION at 1310 nmWeek xx - yy, supplier AA.classpart r%l123456789101112 0.50 :0.00.071.726.31.40.50.00.00.00.00.00.0ERICSSON REVIEW No. 1-2, 1992

19determined period, for presenting certainstatistical information, and for showing (bymeans of histograms) how the data is distributed,Fig. 8. Quarterly reporting of fibrestatistics is one application of this software.A production report shows the output of thefibre manufacture at Ericsson Cable's ownfactories, both for defined periods of theyear and on a per-year-basis. The figuresshow the approved amount of fibre, expressedeither in number of metres or asa percentage of the total length of drawnfibre. Report items include the number ofdrawn preforms, the relation between actualand theoretical results, and the lengthof approved or rejected fibre.Expected future developmentFlexibilityToday, 30 peripheral units (terminals, teststations, etc) can be connected to the mini-computerof the Dehlfi system. Since thesystem would be useful to more technical,production and inspection staff, an extensionis being planned according to twomain options:- Extension of the present star-shapednetwork by adding serial ports on themini-computer. This will permit connectionof 80 users- Installation of an LAN (Local Area Network).The advantage of this option istwofold: a larger number of users can beconnected, and the LAN can be connectedto the existing PC network.Both options require extension of the primarymemory to 16 Mbyte.A result of the in-house development of thesoftware is high flexibility. Programs in thedatabase and in the test stations can easilybe modified in step with the introductionof changes to design and production procedures.International applicationThe Dehlfi system is available in threelanguages: English, Spanish and Swedish.In addition to the Swedish Hudiksvallfactory, the system has been in operationat Ericsson's new Fibroco factory in Spainsince that factory was inaugurated in July,1991. The English version is used whenguiding visitors from abroad and as a platformfor translations into other languages(only Spanish, so far).All further Dehlfi development work is doneat Hudiksvall. The English and Spanishversions are updated in parallel with thedevelopment of the Swedish software. Ourobjective is to be able to provide three mutuallycompatible versions.ReferencesLarsson, A. and NygSrd-Skalman, K.: SlottedCore Optical Fibre Cable. Ericsson Review65 (1988) :3, pp. 100-107Stensland, L. and Gustafsson, P.: FibreManufacture Using Microwave Technique.Ericsson Review /65(1988):4, pp. 152-157.

Human Factors - A Key to ImprovedQualityThomas Backstrom and Nigel ClaridgeTelecom services are being developed rapidly, and introduced at an alarming pace.Both end-users and telecom network operators will soon experience how theseservices and other emerging possibilities place new and greater demands on theircommunication with the telecom system. Technical solutions must take human issuesinto account - technology is used by people in their daily lives. As the number ofservices offered increases, it will become more and more important to provideefficient interaction between the individual and the technology.Internationally, the art of creating an integrated man-machine system is known asHuman Factors Engineering, or Human Factors for short.The authors describe what Human Factors is, Ericsson's corporate policy within thearea, and the process required to create good and usable products from the humanviewpoint.human factorstelecommunication servicesstandardisationFig. 1In a changing world, where telecommunicationsis becoming a part of everyday life of the individual,variations in culture, traditions and languageplace demands on the design of telecom productsand services. Human Factors increases thedegree of usability and acceptability by makingproducts simple and easy to useEricsson manufactures and supplies productswithin a wide variety of areas - fromspecialised and complex products, suchas information systems used in fighter aircraft,to public and private telephone exchangesand consumer products such astelephone sets. Each product has its ownset of specific design requirements andconditions, based on the characteristicsand limitations of the intended users.Although there are few similaritiesbetween the design specifications for thecockpit of a modern fighter and those fora desk-top telephone set, all products - irrespectiveof the level of technology - haveone basic demand in common: they mustbe designed so as to perform well in theenvironment for which they are intended.Clearly, a fighter cockpit or air traffic controlcentre requires significantly more developmentresources than a simpler product,but how many of us have beenfrustrated, if not injured, when using a canopener. After several unsuccessful attemptswe have probably thrown it away.To avoid these problems, the way a productwill be used - and by whom - must bethoroughly analysed and the results integratedearly into relevant product specifications.In principle, there is little differencebetween the Human Factors methods appliedwhen designing a cockpit and thoseapplied in the design of a telephone set.The variation lies in the number and complexityof relevant factors, resulting incockpit design being significantly more difficultand more extensive.Historical backgroundThe need for new specialist knowledgearose because both users and manufacturersbecame more and more specialisedin their activities, and because demandsfor safety and efficiency increased. Thisknowledge identified demands to be includedin product design - demands thatwere based on the users' (operators') cognitiveand physical limitations and the worksituation. It also formulated these demandsso that they were understood andusable by the systems engineer.The need for Human Factors knowledgewas accentuated during the Second WorldWar. Technical developments advancedto such a degree that individuals operatingsystems and vehicles experienced seriousdifficulties and were exposed to personaldanger. It became necessary todesign control systems and to present informationin a way that permitted efficientoperation, yet avoided overstressing theindividual and eliminated the risk of accidents.This development has continued, and HumanFactors knowledge is now applied notonly to complex and sophisticated productsbut also to a wide range of everydayproducts, such as cars, computers, toolsand machinery and consumables. Aids forthe elderly and the handicapped is anotherarea where Human Factors knowledge isapplied.ERICSSON REVIEW No. 1-2, 1992

21NIGEL CLARIDGENomos Management ABTHOMAS BACKSTROMEricsson Telecom ABBox 1Definition of Human FactorsHuman Factors is defined as "The study of the relationbetween man and his occupation, equipmentand environment, and particularly the applicationof anatomical, physiological andpsychological knowledge to problems arisingtherefrom".Therefore, "Human Factors in Information Technology"(IT) is considered to be the study of, andapplication of Human Factors knowledge to, allaspects of the relation between the human, themachine and the environment which directly influencethe safe, efficient, acceptable and satisfyingusage of the IT system.Box 2HUMAN FACTORSENGINEERINGEricsson's PolicyPurpose To ensure high Human Factors'quality in all Ericsson productsPolicy Human Factors Engineeringshould be applied to all product design,thus making the productsprofessional also from the individualuser's and handler's point ofviewResponsibility Business Area and/or Local Companymanagement is responsiblefor the issuance of application rulesfor Human Factors Engineering.Corporate Systems andTechnology is responsible forcoordination activitiesGuiding Principles- Products and systems to be used in similar userenvironments should be given uniform economicor Human Factors properties- In product design work, great attention shouldbe paid to the concepts "user-friendliness" and"suitability for purpose"- Our professional knowledge in Human FactorsEngineering should be utilised in our marketing- Qualified application of Human Factors Engineeringshould be emphasised as importantquality improvements- Existing standards in the Human Factors fieldmust be followed and measures must be takento further develop such standardsERICSSON REVIEW No. 1-2,1992What is Human FactorsAs defined in Box 1, Human Factors is aninterdisciplinary science. The terms HumanFactors and Ergonomics are oftenused synonymously when defining boththe applied area of research and the methodused to create products. In Scandinavia,the term Ergonomics has beenstrongly identified with the physical designof products and workplaces, i.e. associatedwith man's physical characteristics relevantin this context: anthropometry,stress, vision and hearing. The term HumanFactors is used to encompass theseaspects - and also the cognitive aspects- in the design process, such as man's abilityto process, interpret and act upon information.Cognitive issues have grown inimportance as information technology hasbroken into more and more areas of application.The changing world ofTelecommunicationsAs a result of increased competitionthrough deregulation, network operators'cost situation has become strained at thesame time as subscriber demand for servicehas increased. Subscribers have becomeincreasingly dependent on the telecomservices they use. Individuals andcompanies are emphasising the need formobility; yet demanding access to the samefunctions and services irrespective oflocation. Many of these demands are nowfulfilled by both public and private exchangesand by fixed and mobile services.Common to all these changes is the demandfor simplified methods of handlingan increased range of complex services.Services must be easy to learn and easyto use for end-users, subscribers and networkoperator service staff. Handling proceduresfor a given service should be thesame irrespective of the application, whichmeans that efficient, uniform handlingcharacteristics will become one of the mostimportant competitive edges in the future.This applies both to products which arepart of the network and to products usedby subscribers when utilising network services.Efficient handling characteristicsare a result of products that are well adaptedto user requirements and the intendedtask.There are further advantages to be gainedthrough the application of Human Factors.Variations in culture, tradition and languageare taken into account, which resultsin products becoming accessible toa wider range of end-users. The applicationof Human Factors will reduce trainingtime, lead to more reliable data input andhigher productivity, and ensure that productswill be better adapted to the environmentsin which they will be used. Facedwith the trend (noticeable in many countries)of increased supplier responsibilityfor product safety, the need to designgood, safe products supported by usabilitytesting becomes paramount. HumanFactors input in product development willbecome more and more essential.Ericsson's PolicyEricsson's goal as regards Human Factorsis that all Ericsson products should meetuser requirements for comfort and safety,and at the same time promote user andsystem performance efficiency. The Ericssonview of Human Factors is summarisedin a corporate policy which has been approvedby the Corporate Executive Committee,Box 2. The policy emphasises theneed for all Ericsson products to have aprofessional design with regard to user requirements.It also expresses the need forco-ordination in these issues between thedifferent corporate units.A principal criteria put into practice is thatspecifications about usability and functionalityare integrated, at an early stage, intothe development process of completelynew or existing products.The Technical Committee forHuman FactorsEricsson Technology Council (ETC) coordinatesthe follow-up of technical developmentsin general, and also the provisionof new technology within Ericsson. Thistask is performed via a number of researchcentres, application labs and a series ofspecialised technical committees. TheTechnical Committee for Human Factors(TUHF) is one of these committees.TUHF's task is to strengthen Ericsson'scompetence within the area of HumanFactors and to ensure a uniform corporateprofile.TUHF, through a network of contacts, bothwithin Ericsson and with external institu-

221 Use the user's model2 Introduce through experience3 Design the product to be user-centred4 Allow the user to observe and control thesystem5 Log activities during testing6 Ensure consistency and uniformity7 Provide immediate feedback8 Avoid the unexpected9 Validate entered data where possible10 Provide selected help informationFig. 3Ten fundamental principles to keep the userfriendly! (British Telecom, 1990)Fig. 2The main elements in the computer systemIt is important to view the system not merely as aset of technological items, but as the totality ofpeople at work with equipment in a working environment.The challenge is to provide sufficientflexibility and variety of interfaces to suit thewide range of users and task situationstions, organisations and specialists, receivesvital input for its development programmeand in this way ensures increasedHuman Factors knowledge about new technologyand market demands. Ericssoncooperates with Nomos Management AB,which provides access to current specialistknowledge in the Human Factors area.Human Factors and Industrial Design havemany areas of mutual interest. AlthoughIndustrial Design tends to work with thephysical characteristics of products, emphasisingaesthetic values, more andmore work is being done on the graphicaldesign of user interfaces, notably the designof symbols, pictograms and icons.TUHF, therefore, works in close cooperationwith The Technical Committee for Design.4International CooperationCooperation has been established withBritish Telecom's Human Factors Divisionin England. The purpose of this cooperationis - through exchange of knowledgeand experience - to strengthen HumanFactors know-how in both companies andto ensure that Ericsson products fulfil BritishTelecom's usability requirements. It alsoaims at developing a broader base forusability testing, involving end-users fromseveral markets.Working with Human FactorsThe user, the task, the system and the softwareare the main elements which, togetherwith procedures and decision rules,form an efficient system, Fig. 2. The userinterface, where a given task is to be performed,is the common boundary betweenthe user and the technical system. The primarygoal of Human Factors is to specifyand design the interface so that the userwill be given the best possible environmentto carry out his task safely and efficiently(Box 3).The general approach is the same whendesigning all types of interfaces, althoughmethods and the amount of input data inthe different project phases may vary. Likeall other development work, Human Factorsis an iterative process. The designprinciples presented in Fig. 3 should be followedthroughout the development of userinterfaces.Start with the usersThe end-user determines how a system isused in day-to-day operation and, hence,the design of the interface. The initial stageis to identify the users and describe theirknowledge, skills and experience. It is importantto note the differences and extremeshere, and design the interface accordingly.More complex products andsystems require an understanding of theway users process information, interpretdisplays and make decisions. Faulty assumptionsabout user needs can have anegative effect on usability. If the interfaceis not designed to meet user requirements,it is unlikely that the product will beused efficiently, if at all.Identify the task to be performedA task analysis is made to describe usertasks in detail and to collect informationabout working environments. This providesfundamental information about userneeds. Several task analysis methods areavailable for making the process of collectingand abstracting information aboutusers more systematic. They range from"macro" methods - the whole system isanalysed in terms of organizational, socialand environmental issues - to "micro"methods, through which sub-tasks are reducedinto small cognitive units.ERICSSON REVIEW No. 1-2, 1992

23Box 3User interfacesUser interfaces can be defined as the boundarybetween the user and the system and wheresome task is performed. Interfaces can be- dialogues with computers and other products- presentation of information on a display and/orother medium- entry of data/commands from keyboard or otherinput device- physical design of products and workplaces- sight and sound characteristics- procedures, rules and routines- icons, symbols and pictograms- systems for marking cables and connectors- instructions, manuals and training materialFig. 4Two generations of Ericsson mobiletelephonesMobile telephony will be a major part of futuretelecom services. Human Factors will play an integralrole in designing services based on userrequirements and interfaces with regard to cognitiveand physical issuesExamples of task factors are-ease- complexity- novelty- time limits-whether tasks are repetitive, overlapping,monitoring- what skills a task requires- whether a task is performed by a groupor by an individual.Define evaluation criteriaWhen usability testing is integrated into thedesign cycle, it should be quantitativelyspecified in advance. Performance goalsfor usability should be measurable. Keyfactors in usability testing are- learnability- or ease of learning - is thetime and effort required to reach a predefinedlevel of user performance- performance - or ease of use - is thenumber of tasks accomplished by experiencedusers, the speed of task executionand the number of mistakes made- flexibility, the extent to which users canadapt a system to new ways of interactionas they become more experienced- attitude, subjective assessment engenderedin users by the system and the interface.Test and evaluateTesting and evaluation should be carriedout with users early in the design processto ensure that products and interfaces aredesigned to meet user needs and requirements.The conceptual design of an interface maystart with the creation of sketches of the interfacewhich demonstrate surface features.These can be tested for appropriateness;for example, the design of ametaphor or symbol can be checked withrespect to how well it is understood by users.As design develops it will be transformedthrough simple models and simulations toprototypes, full-scale models and, eventually,the final product (implementation).There are many computer-based designtools which permit quick and easy evaluationof user-interface options. Applicationof these tools is essential if interface designis to be cost-effective. Well conductedtrials ensure that the design matchesthe users' mental model of the system andthe tasks.Trials should be conducted in environmentswhich are as realistic as possible;preferably the real situation. It is importantto establish whether the product is reallyusable or not. The principle rule is to designiteratively, with many cycles of "design- test with users - redesign". The aimis not to produce one correct solution whichis subsequently not changed, but an evolvingsystem which - with each iteration - isfurther tailored to users' needs.Experience gained from evaluating theproducts, especially in the real situation,will provide input when planning futuregenerations of the product, or completelynew products. "Future-proofness" is of keyimportance, and interface design guidelinesprovide a consistent approach toproduct and system uppgrades.Variation in the development processThe development process is seldom simpleand regular. The process rarely startsfrom square one, i.e. the development ofa completely new product or system.There is invariably some form of historyfrom existing products which must be takeninto account. This means that HumanFactors efforts in each phase will vary fromproduct to product.When a new product is created, based onnew technology, it is important to identifyuser attitudes and requirements so that theproduct is used, understood and acceptedby users once the novelty value wears off.When combining products from severalvendors into a new and unique configuration,qualities such as homogeneity andcontinuity in use are important. In general,the primary Human Factors considerationis consistency of operation of all componentsrather than optimisation of a singlecomponent.Manufacturers sometimes buy productsfrom another vendor. Human Factors effortwill depend on several factors, such asthe quantity purchased, the importance ofthe product to the corporate strategy, andthe ease with which usability can be improved.There is a golden opportunity to improveusability when redesigning an existingproduct: by obtaining usability data fromthe users of the existing product. Featuresand functions that have been well receivedERICSSON REVIEW No. 1-2. 1992

24Fig. 5Advanced Human Factors design - J AS cockpitThe cockpit ot JAS 39 is designed to provide thepilot with maximum support during flying, navigationand tactical applications. Displayed informationis either presented automatically or canbe selected by the pilot.The display in the centre is an electronic mapproviding navigational information. Human Factorsspecified the quantity of information presentedand designed and tested symbols. Simplicityis the key. All information must be easilyunderstood under the most difficult conditionspossiblemay be retained, while the redesign effortfocuses on solutions to problems that usershave identified as being the most serious.StandardisationCurrent standardisation activities in thearea of Human Factors are directed mainlytowards the user interface and software.A number of relatively similar de-factostandards for information display (CUA,OPEN LOOK, Motif, Apple) are used bymanufacturers when they design their applications.This illustrates the difficulties inestablishing a uniform standard duringrapid technical developments. A user interfacestandard must be verified in a widerange of applications before it can begenerally accepted. During this time, furtherdevelopments occur and the conditionschange. No quick breakthrough resultingin one particular standard dominatingover the others is therefore expected.In the telecommunications area, where thelarge network operators have a significantinfluence in the standardisation bodies,standards for completely new services arebeing worked out. The European TelecommunicationsStandards Institute (ETSI)has twelve committees, one of which dealswith Human Factors. Issues dealt with concernusability in telecommunications, suchas Universal Personal Telephony (UPT)and video telephony. The procedures,symbols and icons used in these applicationsare studied. Other issues concerntelecommunications for people with specialneeds, including the handicapped; theobjective being to ascertain that publictelephone services are available to thesegroups. ETSI also produces manuals,handbooks and check-lists for the designof services and products.Current Human FactorsprojectsTelecom servicesThe total number of telecom services is expandingrapidly and each user is going tobe confronted with many more servicesthan today. Services are going to be broadin nature and not, as today, mainly oftelephony character. Thus, the demand forsimple, intuitive and reliable use of serviceswill become more and more important.There must be a significant degree ofuniformity between closely related services,and a clear and consistent differencebetween different groups of services sothat users understand the distinctions. Toachieve this is no easy task, especiallywhen cultural variations in the interpretationof concepts and use of telephony mustbe considered.Human Factors is intensively engaged indesigning new services for public, privateand mobile telephony. Examplesof currentactivities are- Development of system-independentmethods of specifying and describingservices based on user perspectives ofthe services. (The design of services incurrent networks is determined by theway in which the telecom system is constructed)- Studies of the use of current serviceswith detailed analysis of motivation andjustification for each service- Evaluation of prototypes of new serviceswithin different user groups- Test and evaluation of symbols andicons for services.Presentation of information inadvanced vehiclesOne of the most stressful workplacesimaginable is the cockpit of a modernfighter aircraft. Pilots must interpret situationsquickly and accurately and act accordingly.They are often exposed to acombination of mental and physical stress,typically g-forces. To support the decisionmakingprocess, the pilot receives information- from several different sources -which is continuously adapted and combinedin a suitable way before being presented.The visual display unit is the mostefficient and suitable means of presentingthis information.Ericsson has designed the informationsystem for the Swedish Airforce's newfighter, JAS 39 Gripen (Fig. 5). The cockpitof JAS has been the subject of HumanFactors effort in four major areas: presentationof information, design of controls,workstation design and environmental adaptation.The pilots' physical characteristics,capabilities and limitations have beenanalysed in detail, as have the type andamount of information they need. Testswith mock-ups and various forms of prototypeshave been used to increase knowledge.The establishment of the final designof the cockpit, with informationsystems and controls, has been based onERICSSON REVIEW No. 1-2, 1992

25Fig. 6Cellular network configurationsIt is essential to see information about all or partof a cellular network in order to manage it efficiently.TMOS User Interface Design Standards(TUIDS) have been applied to interface designand have specified rules for the selection anduse of colour, for exampleresults from simulation and full-scale trials.Only then has the system been installed inits correct environment.Telecommunications Managementand Operation Support-TMOSTMOS is a family of systems whose differentproducts together offer operative supportfor the whole telephone network andits services. Based on TMOS, Ericssonhas developed a series of products:NMAS Network Management System forthe switched telecommunicationsnetworkSMAS Service Management System forintelligent networksCMAS Cellular Management System formobile telephone networksFMAS Facility Management System fortransport networksBMAS Business Management System forCentrex and Virtual Private NetworksIn a family of systems, characteristics andconfigurations should be based on a uniformbasic structure, partly to illustrate therelationship but mainly so that users recognisesituations and procedures. This isof particular importance for the user interface.TMOS is based on workstation technologythat provides products with powerfulgraphical user interfaces, Fig. 6. Inorder to create a homogeneous interface,Ericsson has developed a standard forTMOS "look and feel" - TMOS User Inter-ERICSSON REVIEW No. 1-2, 1992

Fig. 7User interface for PBX attendantworkstationsHuman Factors has been involved in the analysisand simulation of new PBX workstations. Thisfigure illustrates a normal traffic situation. Incomingcalls are presented in the left-hand field;outgoing calls in the right-hand field. The arrowin the centre indicates voice contact. The displayis void of redundant information making the requiredinformation easy to find and selectFig. 8User interface for PBX attendantworkstationsThis figure illustrates a situation in which the attendantcan obtain information about the requiredextension using any one or a combinationof several search profiles available in the directory.Directory support is located in the lower partof the displayface Design Standards, TUIDS. TUIDSconsists of two equally important parts:- Part 1, which describes the method of integratingHuman Factors into the specificationand design process- Part 2, which provides general and specificrules for the design of TMOS userinterfacesWhen designing interfaces, trade-offsbetween conflicting demands must bemade. TUIDS provides information whicheases this situation. It is designed to worktogether with commercially available userinterface style guides, where all interfacedetails are described (windows, windowframes, buttons, menus, etc). TUIDS specifieshow these elements should be used,combined and grouped to create an efficientinterface. The style guide selected forTMOS is Open Look.Work is under way to present a corporatestandard for designing user interfaces. Althoughdeveloped for the TMOS application,TUIDS will be adapted to form theback-bone of this standard.PBX attendant workstationsTechnical developments have changedthe application of Human Factors in thede-sign of attendant workstations: the emphasishas shifted from physical issues tocognitive ones. Prime requirements in thedevelopment of Ericsson Business Communications'futureworkstationforMD110were increased user efficiency, availabilityand flexibility. The functionality withinthe PBX has become more complex andmore sophisticated, accentuating the needfor easy and logical access to the differentfeatures.If basic and frequent tasks are awkward toperform and less frequent tasks difficult toremember, then system credibility will below - as will user efficiency. Simplicity istherefore essential, especially for the moresophisticated functionality available; forexample, the integration of supportsystems with basic call-handling procedures.The task of designing the user interfacewas complicated by the fact that the productwas to be sold worldwide. This meanttaking different business cultures and traditions- and different languages - into accountin the design, and having the resultstested and evaluated not only by specialistsin the team but also by experiencedPBX attendants on both sides of the Atlantic.The participation of users in the processof designing, simulating and developingthe user interface has had aproductive impact on the design work.The user interface specification was theguiding factor for hardware design. Onceit had been established what informationwas to be presented and how the PBX attendanthandled this information, hardwareissues such as display technologyand size, and keyboard layouts, could bespecified.One specific area has been the design andtest of symbols and pictograms for PBX attendantworkstations. Symbols and pictogramsare interpreted differently in differentcountries and also by individuals withina country. This is a challenge to the HumanFactors specialist, since the userERICSSON REVIEW No. 1-2, 1992

Fig. 9User interface for desk-top telephone setsThe advent of the display in desk-top telephoneshas solved many Human Factors problems and,at the same time, has created new ones. Initially,the type and quantity of information displayedwas specified and the way in which users accessedthe various functions and features wereidentified.Fig 9a illustrates the "Idle Mode" with informationsuch as time and date, subscribers' extensionnumber and subscriber name. The third rowis reserved for menu selection only. A maximumof four alternatives are presented, each correspondingto a menu key located directly underneath.Figure 9b represents the "Message Mode". Theuser is informed that there are messages andcan either call a person direct, leaf through othermessages or delete messagesReferences1 Dialogergonomi - Effektiv interaktionmanniskor-dator. Mekanforbundet 1989,Mekanresultat 89005 ISBN 91-524-1021-8.2 Asker, B.: Graphic User Interfaces. EricssonReview 67, (1990):3, pp 138-146.3 Brantberg, H., Ostman, J.: VDU Maps inSophisticated Craft. Ericsson Review 68(1991):1,pp 10-20.4 Lindhe, R. Wiklund J..Ericsson Design. -Policy, Guidelines and Methodology.Ericsson Review 66 (1989):1, pp 40-46.5 A Guide to Usability. 1990. The Open University,Milton Keynes, England.interface must work efficiently and beunderstood irrespective of whether it isused by a Russian or an Australian operator.Design of desk-top telephone setsThe desk-top telephone set has alwaysbeen the subject for Human Factors application.Traditionally, Human Factorshas addressed the physical issues, suchas numeric keyboard layout, keytopshape and size, receiver weight, size andbalance.With the advent of digital phones, the emphasishas shifted from these physical issuesover to the analysis of procedures requiredto obtain the wide variety offunctions and features now available.The procedures required to obtain thesefeatures/functions have been none toouser-friendly. The everyday user has hadto enter long unfamiliar codes, such as*21 * number #, in order to use the simple"follow-me" feature. The net result of theseawkward codes was that many of the features/functionsoffered were not used becauseprocedures were long and difficultto remember.The situation was partly remedied by increasingthe number of programmablekeys on the telephone set. Instead of thecomplicated codes, the user only neededto press one key. This was convenient forthe user but resulted in telephone sets witha vast number of keys, which was not particularlycost-efficient. The advent of thedisplay in the telephone set eased manyof the Human Factors problems mentioned.The aim was to minimise the useof long codes and reduce the number ofkeys. But the display created new problems.Human Factors work addresseduser-interface design, in particular the typeand quantity of information displayed andthe way in which the user accessed thevarious features/functions available. Onthe basis of this analysis, then and onlythen, could the size of the display be established.Figs 9a and 9b illustrate the informationcontent in the telephone display. The exemplifieddisplay is made up of three rowswith 40 characters in each row.The analysis of the user interface is followedby a visualisation process whereend users participate in the simulation andevaluation of traffic situations. User participationand feedback is an integral part ofthis design and development work.SummaryIn all development work there is a desireto improve user efficiency and technicalperformance and to rationalise production.In addition, products and systems are increasinglybeing combined into new, evenmore sophisticated and complex creations.For a market-orientated technical corporationlike Ericsson, the usability of its productsis an important part of the company'sbusiness and marketing strategy. It is vitalthat products and systems are designedso that they meet customers'requirementsand expectations.The Human Factors work contributes activelyto the usability, acceptability andlearnability of Ericsson products andsystems, and to the application of thecompany's Human Factors policy withinthe scope of the Ericsson Human Factorsprogramme.ERICSSON REVIEW No. 1-2, 1992

Cell-voltage EqualisersSeries BMP 160Johan FrandforsIn telecom installations, batteries are used to guarantee uninterrupted service in theevent of a power outage. The dimensioning of battery capacity is based on estimatesof the frequency and duration of outages and other disturbances. The condition of thebatteries is also an essential parameter. Inadequate charge of individual battery cellswill reduce the time electronic equipment can be powered by batteries. In 1982,Ericsson Components AB introduced cell-voltage equalisers as a means of securingfull charge of all cells in a battery.The author describes Ericsson Components' new cell-voltage equalisersBMP 160 003, BMP 160 004 and BMP 160 005 and how they can be employed tosecure proper charging and increase battery service life.consequence, that these cells will be discharged.Other cells are supplied with acurrent in excess of their need and areovercharged. To improve the condition ofthe first-mentioned cells, the battery mustbe subjected to equalising charge at regularintervals, with a current that is capableof recharging those cells which havenot been compensated for their self-discharge.Other cells will inevitably be overcharged,which has a negative effect ontheir service life.equaliserscells (electric)power supplies to apparatusThe principle of cell-voltageequalisersA 48 V battery contains 23 or 24 cells connectedin series. Each cell is subjected toa self-discharge process, the intensity ofwhich depends on manufacturing tolerancesand the battery's age. Float-chargingof the battery, in normal operation, is oneway to compensate for this self-discharge.If no cell-voltage equalisers are used, allbattery cells are supplied with the samefloating current, which means that for someof the cells the floating current is lowerthan the self-discharge current and, as aThe cell-voltage equaliser is a current/voltageregulator (Fig. 2), which is connectedacross each single battery cell. The rectifierthat feeds the battery is set at an outputvoltage equal to the nominal cell floatingvoltage multiplied by the number ofseries-connected cells in the chain. If allthe cells were identical, all of them wouldhave the nominal voltage and, as appearsfrom Fig. 3, an additional current of 200mA would pass through the shunt regulatorin each cell-voltage equaliser. If a cellhas a voltage below the nominal value, thecell-voltage equaliser will reduce the currentthrough the shunt regulator, and thecorresponding higher current will passthrough the cell. If the cell voltage exceedsthe nominal value, the equaliser will increasethe current through the shunt reg-Rg. 1Cell-voltage equaliser BMP 160 005/1, connectedto a battery cellERICSSON REVIEW No. 1-2, 1992

29JOHAN FRANDFORSEricsson Components ABulator and so reduce the current throughthe cell. In this way, each individual cell willhave a current whose value is determinedby the deviation of the cell voltage from thenominal value. This means that, in a fullycharged battery, each individual cell will besupplied with the floating current neededto compensate for its self-discharge.Cell-voltage equaliser BMP 160Fig. 1 shows a BMP 160 cell-voltage equaliserconnected to a battery cell. The equaliserweighs 40 g and its dimensions are30x23x9 (HxWxD). It is connected directlyto the cell poles by means of one red andone blue 270 mm cable provided with ablade-type contact. BMP 160 comes inthree mechanically identical designs whichare in different colours: BMP 160 003/1 for2.25 V is red, BMP 160 004/1 for 2.23 V isblue, and BMP 160 005/1 for 6.75 V isblack. The equalisers for 2.25 V and 2.23 Vare intended for individual cells: the onefor 6.75 V is intended for three-cell batteries.The design is simple and rugged. Theregulating capacity, 0-350 mA, is sufficientfor batteries with a nominal floating currentof up to 200 mA. If greater capacity is needed,several equalisers can be connectedin parallel across the cell.Technical data of the three versions aregiven in Box 1.Fig. 2The battery floating current is regulated in eachindividual cell and is not affected by the conditionof other cellsFig. 3Current l s through shunt regulator for cell-voltageequaliser BMP 160BMP 160 004, 2.23 VBMP 160 003, 2.25 VBMP 160 005, 6.75 VERICSSON REVIEW No. 1-2, 1992

30Box1CELL-VOLTAGE EQUALISER BMP 160 - TECHNICAL DATADimensionsWeightConnectionConnecting cableMax voltage without destructionTemperature range30x23x9 mm40 g6 mm blade270 mm2.9 V0 to +55°C(1.2"x0.9"x0.4")(1.4 oz)(32 to 131F)ModelTrimming levelColourShunt current attrimming levelBMP 160 004/12.23 +0.007Blue200 mABMP 160 003/12.25 ±0.007Red200 mABMP 160 005/16.75 ±0.015Black150 mAFig. 4, below leftCell voltages measured In a large-size batterywithout cell-voltage equalisers (ICE)Fig. 5, below rightCell voltages measured in the same battery whenit has been provided with cell-voltage equalisersfor a short periodExperiences of cell-voltageequalisersBackup timeFig. 4 shows the distribution of cell-voltagevalues measured in a large-size batterywhich has been used in a plant, underobservance of normal maintenance routinesbut without cell-voltage equalisers.Fig. 5 shows the corresponding valueswhen the battery has been provided withcell-voltage equalisers for a short period.The distribution between the cells has decreasedfrom the order of ±200 mV toabout ±20 mV; that is, by a factor of 10.This means a marked improvement of thecondition of the battery.The proper charging of a battery cell canonly be checked through several hours ofloading tests and voltage measurements.The fact that the voltage across a cell innormal float-charging is close to the nominalvoltage does not guarantee a satisfactorycondition of this cell. Nor is the oppositeself-evident: if the voltage across a cellin a battery without equaliser is up to 50mV below the nominal voltage, this doesnot necessarily mean that the cell is in abad condition. The latter, on the otherhand, is an indication that the cell mayhave a higher self-discharge than othercells in the battery and that the floating currenttherefore has not been sufficient tokeep the cell fully charged. If the worstcomes to the worst, the charging of sucha cell may be only 50-60 % of the capacityof a fully loaded one. In an eight-hourtest, the battery capacity would drop to thelower voltage limit after less than fivehours. This means a 50 % reduction of thebattery backup time in spite of the fact thatall the cells are fully capable of functioning.EconomyThe above example may seem somewhatdrastic, but low cell voltages caused by insufficientfloat-charging occasionally appeareven in relatively new batteries. Theempty circles in Fig. 6 show the cell voltagesin a 48 V, 3,000 Ah battery that hasbeen in operation for four years. The valuesfor about one third of the cells are 35m V or more below the nominal voltage. Unfortunately,this battery was not tested withrespect to capacity when the voltageswere measured, and its real condition istherefore not known. However, plant managementconsidered boost charging necessaryin order to restore the condition ofthe battery but decided to choose the alternativesolution: cell-voltage equalisers.Fig. 6 shows the voltages measured beforethe cell-voltage equalisers were installedand the corresponding values aftersix weeks. An eight-hour discharging test,in conformity with US standards, wasmade after another three months. The bat-ERICSSON REVIEW No. 1-2, 1992

Fig. 6Cell voltages measured in a 3,000 Ah battery afterfour years' operation without cell-voltage equalisers,and six weeks after equalisers havebeen installed. The cells in the diagram are arrangedaccording to the values recorded in thefirst measurement• Initial reading• Reading after 6 weekstery voltage was measured every fifteenminutes; the results are shown in Fig. 7.As is evident from the power drawn fromthe battery - 104 % of the nominal eighthourcapacity - it had been restored to perfectcondition.In a battery without cell-voltage equalisers,the balance between the cells is restoredthrough periodic equalising charges.Since the battery backup time in practiceis determined by that battery cell which isin the worst state of charge, the battery willnot - in the intervals between the additionalcharges - meet the backup-time requirementscalculated on the basis of nominalbattery data. Correct dimensioning, therefore,requires that a battery without cellvoltageequalisers should have somewhatgreater capacity than a battery providedwith such equalisers. It is difficult to quantifythe need for overcapacity: it dependsvery much on the maintenance of batteriesat each specific plant. But a plausibleestimate is that the cost price of batterieswith cell-voltage equalisers is roughly thesame as that of batteries without equalisers.The two alternatives - greater batterycapacity or automatic control - are in allprobability equally good. On the otherhand, cell-voltage equalisers make thebattery more resistant to insufficient maintenanceand in this way help to assure thatit is in good condition when its capacity isneeded.This means that battery maintenance canbe made easier. Experience has shownthat the number of voltage readings can behalved, and substantial reductions of costshave been demonstrated in many cases.Relatively large cost savings can also bemade when new batteries are installed andwhen battery cells have to be exchanged.When a new battery is installed, it is normallycycled three or four times, in orderfor all its cells to be in the same condition.In this procedure, which takes three to fourdays, the battery is measured each hour.It has turned out, though, that a battery providedwith cell-voltage equalisers seldomrequires this cycling and measurementprocedure. Such a battery is in perfect conditionafter six weeks of normal float-charging.When the equipment to be powered isinstalled and tested, the battery is well balancedand in good condition.If the capacity of a cell in an installed batteryhas been considerably reduced, dueto premature ageing or some kind of dam-Fig. 7Capacity test of the battery shown in Fig. 4, threemonths after installation of cell-voltage equalisersERICSSON REVIEW No. 1-2, 1992

32age, the faulty cell or the entire batterymust be exchanged. In most cases, a newcell has much lower self-discharge thanthe ordinary battery cells. Replacement ofindividual cells in a battery therefore causesbalance problems and is not recommendedin industry standards. In a batteryprovided with cell-voltage equalisers, theimbalances between the cells are automaticallycompensated for, which meansthat individual cells can be replaced withoutany risk of complications.An interesting alternative, when replacinga cell in a battery without cell-voltage equaliser,is to provide only the new cell with anequaliser. This is one way of avoiding bothovercharge of the new cell and voltagedrops in other cells in the series-connectedfloating-current chain. But if a batterycell needs to be replaced, it is likely thatother cells too are in a bad condition andthat the battery backup time has been reducedto a value well under what was projected.If an assessment of the battery'scondition indicates that satisfactory backuptime can be achieved by providing allthe cells with equalisers, then this solutionis often preferable to scrapping the battery.When it is eventually scrapped, after havingserved its time, the cell-voltage equaliserscan be shifted on to the new battery.Fig. 8Measurement of cell voltages in batteries installedverticallyERICSSON REVIEW No. 1-2, 1992

33Thermal run-awayPlacing batteries in cramped spaces involvesa risk ot thermal run-away. Whenthe temperature in a battery rises, the selfdischargeprocess is intensified, which willraise the floating current. The power generationin the battery increases and causesthe temperature to rise further. Theeel Is in the centre of a battery as a ru le can -not carry off heat to the same extent as theouter cells and will therefore have a highertemperature. The cell voltage rises with increasedtemperature, which means thatthe hottest cells account for a greater portionof the total power generated in the battery.The temperature in these cells riseseven more. If a cell cannot carry off enoughheat, the process will get out of control,causing the electrolyte to boil away anddestroying the overheated cell. The plantwill no longer have any standby power.If the battery is provided with cell-voltageequalisers, even a slight voltage increaseleads to a substantial reduction of the currentthrough the cell. The feedback is negative.The risk of thermal run-away - andthus a potential source of trouble - hasbeen eliminated.Valve-regulated batteries - intheory and in practiceCell-voltage equalisers have been successfullyused in valve-regulated batteriesalso. However, some experts - basingtheir arguments on theoretical models ofthe chemical processes involved - claimthat cell-voltage equalisers should not beused in this type of battery. If low cell voltagesare caused by sulphation in the cells,the increase of the floating current achievedby means of cell-voltage equalisersclearly has a positive effect. If, on the otherhand, the low cell voltages are due tothe fact that cells have dried out, and thata recombination process is going on, thenincreased floating current will have a negativeeffect. Correspondingly, high cellvoltages may indicate that the recombinationprocess is incomplete, in which casethe floating current should be kept at a highvalue.In the operational data of valve-regulatedbatteries that Ericsson Components ABhas access to, there are no indications thatcell-voltage equalisers have any negativeeffects, even though batteries from differentmanufacturers might possibly have differentcharacteristics in this respect.SummaryIn a power supply system, the battery isthat component which has the shortest lifeand needs most maintenance. This nottoo-glamorousmaintenance also requiresthat the applicable instructions are followedin every detail.When the primary power supply fails, theproper functioning of a fully charged batterysuddenly becomes vitally important.The cell-voltage equaliser has been developedto assure that the battery is alwaysin the best possible condition and to reducethe consequences of temporary deviationsfrom stipulated maintenance routines.The costs of cell-voltage equalisers arecompensated for by cost savings. Firstly,maintenance routines can be simplified,and, secondly, a battery with cell-voltageequalisers need not be cycled when it isinstalled. In a longer perspective, the batteryeconomy will be improved thanks toideal float-charging conditions, which increasethe life of the battery cells, andthanks to the possibility of replacing damagedor prematurely aged battery cellswithout having to include any additionalmaintenance measures.References1 B.W.B Battery Systems Inc. ICE MANU­AL Rev 3.0 1991.12 Bjork, D.: Maintenance of Batteries. IN-TELEC-86.13 Andersson, H. and Bergvik, S.: SealedLead-Acid Batteries for Small Telecommunicationsplants. Ericsson Review601 (1983):4,s 222-225.4 Bjork, D.: Operation and Maintenance ofPower Supply Equipment. Ericsson Review64 (1987):1, s 9-15.ERICSSON REVIEW No. 1-2, 1992

In Search of Managed ObjectsWalter Widl and Kidane Woldegiorgisare laid out in Recommendation M.3010.Management of the telecommunication network across standardised interfaces M.3020 identifies the management servicesthat represent the managementrequires that the necessary resources - in the form of Managed Objects - should alsobe standardised. A methodology for describing tools and processes is needed to tasks and describes the methodology to bedefine the Managed Objects that represent the manageable resources of the used in the specification of the interfaces.telecommunication network.M.3400 contains the list of TMN managementfunctions for the different TMN func­The derivation of the Managed Objects for the transport network is the main focus ofthis article.tional areas. Two recommendations thatprovide information models are in the finalprocess of approval. M.3100 provides ageneric network information model, andG.744 is the information model for SDHbasedtransport networks.telecommunication networkstelecommunication network managementstandardisationFig. 3Relations between CCITT's TMN RecommendationsIntroductionEfficient management of new and existingtelecommunication networks is one of themain concerns of administrations, manufacturersand standardisation bodies. Intensivestudies during the last few yearsseem to have finally resulted in workabletools for the description of networks andtheir management, as appears from the relationsbetween CCITT's TMN Recommendations,Fig. 3. The methodology andthe driving forces behind the definition ofthe standardised interfaces (both communicationprotocols and information model)Figs. 1 and 2 show an overview of themethodology and the processes involvedin deriving the Managed Objects. Fig. 1shows the associated tasks, which - in abroad sense - are seen as the functionaland information modelling processes.A number of TMN Management Servicesare defined by CCITT. "Management oftransport networks" is an example of sucha service. The functional modelling of thetransport network consists of a process,Fig. 2, in which - for each transport networklayer or sublayer - Trails and Con-ERICSSON REVIEW No. 1-2, 1992

35WALTER WIDLKIDANE WOLDEGIORGISEricsson Telecom ABnections are identified, along with the correspondingTermination Points (ConnectionPoints and Termination ConnectionPoints). 2 In this process, Manageable Resourcesin the vicinity of the TerminationPoints are derived.In the information modelling, the Resourcesare combined into Managed Objects.This article demonstrates these processesfor the Termination Point and crossconnectionaspects of the transport network.Fig. 1Development of transport network Managed ObjectsFig. 2Development process for transport network ManagedObjectsNETCPCPMONetwork ElementTermination Connection PointConnection PointManaged ObjectERICSSON REVIEW No. 1-2, 1992

Fig. 4The telecommunication network can be consideredto be divided intoa the physical transmission networkb the logical transport networkcdDESDXCLSTANTNSSPSCPFNUwsOSQthe intelligent services networkthe telecommunication managementnetwork (TMN)Digital ExchangeSynchronous digital cross-connect equipmentLine SystemTerminalAccess NodeTransit NodeService Switching PointService Control PointFeature NodeUserWorkstationOperating SystemInterface between the Transmission ManagementNetwork and the Managed ObjectsIn a follow-up article, the derivation of ManagedObjects for other aspects will be presented.No attempt is made in this articleto define the Managed Objects in detail,since the standards referred to give completedefinitions of the objects in question.A layered view of thetelecommunication networkBasically, networks and their managementare described according to the principle ofdividing complex problems into less complexones, which can be solved independentlyof each other. Partial solutions cancontribute in an iterative manner to the solutionof a complex problem.If these principles are followed, the telecommunicationprocess can be allocatedto various networks, Fig. 4,- the physical transmission network- the logical transport network- the intelligent services network (IN)-the telecommunications managementnetwork (TMN).In addition to the telecommunication functions,each of the networks also containsfunctions for network management. ManageableResources are represented byManaged Objects in the Information Model.The Transmission Network, Fig. 4a, containsthe physical means of transfer of information,i.e. telecommunications equipmentand connecting transmission media:typically, switching matrices of switchesand cross-connect systems, modems,muldexes, line systems and cables. Thetransmission network contains varioustransmission network layers, as specifiedin Recommendations G.702 and G.707,for example. 5 Typical physical managementresources that are part of the transmissionnetwork are: signal level detectors,error counters and error informationstorage.The Transport network, Fig. 4b, performslogical functions related to transmission,such as network configuration and protectionswitching, and ensures reliable transferof information between service users,regardless of the information content. Thetransport network contains various TransportNetwork Layers, each of which carriesa distinct characteristic signal, i.e. a signalwith a particular data rate and format. 2Manageable Resources are the functionalrepresentations of the physical resourcesimplemented in the transmission network.The Service network, Fig. 4c, performsfunctions related to switching and signalling.The purpose of the intelligent ServiceNetwork is to offer users a particular service.The user initiates a service via a ServiceSwitching Point (SSP) using signallingsystem No. 7 or D-channels of theISDN system. The service also needs datafrom a Feature Node (FN). The process iscontrolled by the database of the ServiceControl Point (SCP). 6 Typical ManageableResources are concerned with traffic management,customer services and manyother management functions related toswitching and signalling.The Telecommunications ManagementNetwork (TMN), Fig. 4d, connects the networkoperators and administrators - viaworkstations (WS), operation systems(OS) and standardised Q-interfaces - toManaged Objects which represent the resourcesin the various networks. 3 For themanagement of TMN, typical resourcesare concerned with communicationchecks and protocol conversions.ERICSSON REVIEW No. 1-2, 1992

37Fig. 5Transport Processing Functions, Transport Entitiesand Reference PointsTRPFSTransport Reference PointTransport Processing Function or TransportEntityManagement Reference PointTelecommunication network functions andresources appear in all the networksshown in Fig. 4. Each network can consistof several layers, each of which can bemanaged individually by a Manager-AgentProcess. 3 The resources and their propertiesare represented in a standardised wayas Managed Objects in databases.Management methodologyNetwork management requires the selectionof a TMN Management Service - i.e.a distinct task for an operator - which involvesseveral TMN Management Layers- i.e. the area of application of the task,Fig. 2. The task requires a functional analysisleading to Resources that are suitablefor management. The Resources aredescribed in the functional modelling process.The information modelling process isbased on the management tasks and organisationof the management provider.The selection of a particular TMN ManagementService and related TMN ManagementLayers leads to the selection of TMNManagement Service Components andTMN Management Functions. TMN ManagementFunctions are performed onManaged Objects. 3The information modelling process specifieswhich Managed Objects are neededto perform the desired management tasks.The functional modelling process leads tothe Manageable Recources offered by thenetwork. Harmonisation of these requirementsnormally leads to iterative processes.The Information Model contains all data relatedto the Managed Objects, their relationshipsand naming structure, as seenacross a particular management interfacebetween the managing and managedsystems. In the Information Model, whencompleted, all required Managed Objectsappear in a standardised form.The process of deriving transport networkManaged Objects consists of functionalmodelling and information modelling. Inthese analyses, a given physical transmissionnetwork supports a transport network.Each transport network contains layers definedby a characteristic signal. 2 Connectionsand Connection Points carrying thecharacteristic signal are determined foreach transport network layer. This processleads to a detailed specification of managementaspects of the Network Elementsforming the transport network (NetworkElement View). Each transport networklayer can be partitioned into various geographicaland administrative regions,which offers various degrees of detail inthe network views.The management of Trails and Connectionsis based on Resources in the vicinityof the end points of Trails and Connections.The anomaly detectors and controllabledevices in the telecom equipment arerepresented in the Functional Model asproperties and Resources; these in turncorrespond to transport network ManagedObjects in the Information Model. The followingchapters go into more detail concerningfunctional modelling and informationmodelling for transport network-relatedTMN Management Services.Functional modellingThe physical implementation of digital telecommunicationnetworks comprisesequipment such as- switching matrices- modems and muldexes- line, radio and satellite systems- transmission media for line systems andinterfaces belonging to the transmissionnetwork.Functions performed by the above-mentionedequipment include- cross connection and switching- multiplexing and demultiplexing- signal conversion between physicaltransmission media and interface cables- signal transmission via physical media.The functions of a telecommunication networkcan be modelled with the aid of theERICSSON REVIEW No. 1-2,1992

Fig. 6Functional Model illustrating the use of TransportFunctionsA Connection can be a Link Connection, a Sub-Network Connection or a Matrix connectionATCLCAPCPTCPClient-to-server adaptationTrail TerminationConnectionLink ConnectionAccess PointConnection PointTermination Connection PointFig 7Simplified Functional Model illustrating the useof Transport FunctionsGeneric Transport Network Architecture."The basic elements of this model are- Transport Entities, such as Link Connection(LC), Sub-Network Connection(SNC) and Trail. Transport Entitiestransfer information transparently- Transport Processing Functions, suchas Adaptation (A) and Trail Termination(T). Transport Processing Functionsprocess the transferred information-Transport Reference Points, such asConnection Points (CP), TerminationConnection Points (TCP) and AccessPoints (AP).Some of the Transport Processing Functionsand Transport Entities are defined asfollows:-The Transport Entity "Trail" is responsiblefor the transfer of characteristic information(signals with specified datarates and formats) between the AccessPoints to a layer (AP). A Trail containsnear-end and far-end Trail Terminationsand one or several Network Connections- The Transport Entity "Network Connection"is responsible for the transfer of informationbetween Termination Connec-ERICSSON REVIEW No. 1-2, 1992

39Fig. 8Levels of abstraction in a Layered Network; exampleSubnetworktion Points (TCP). A Network Connectioncontains either Link Connections (LC) orSub-Network Connections (SNC), orboth- The Transport Entity "Connection" is responsiblefor the transfer of informationbetween Connection Points (CP). AConnection can be a Sub-Network Connectionor a Link Connection- The Sub-Network Connection transfersinformation transparently across a subnetwork.The connectivity of a Sub-NetworkConnection can be controlled by amanagement process- The Link Connection transfers informationtransparently across a link betweentwo subnetworks. The connectivity of aLink Connection cannot be changed bya management process.Fig. 5 shows that Transport Functions, i.e.Transport Processing Functions andTransport Entities, are separated by oneTransport Reference Point. In addition, amanagement reference point S is providedfor management information to andfrom each Transport Function. The input/outputconcept is related to TransportFunctions and should not be interpreted asindicating signal directions. "Input" representsthe input, "output" the output part ofthe function.The use of the Transport Functions is illustratedin Fig. 6." Fig. 7 is a simplified versionof Fig. 6 and does not show the AccessPoints (AP) explicitly.The Functional Model can be used to modelthe telecommunication network to thedesired degree of detail. Forexample, NetworkElements and parts of Network Elementscan be represented. Fig. 8 showsSub-Network Connections and Link Connectionswhich belong to the same transportlayer, and which in turn can be partitionedinto subnetworks showing variousdegrees of detail.The NET 1 level shows SNC1s, LC1s,CP1 s and TCP1 s, i.e. the greatest degreeof detail. The NET 2 level - less detailed- shows SNC2s, LC2s, CP2s and TCP2s.In this way various degrees of detail (levelsof abstraction) can be chosen, dependingon the purpose of the Functional Model.The network connection in the NET 4level is that of a single SNC4 between twoTCPs. From a management point of view,NET 1 provides the details required at theERICSSON REVIEW No. 1-2, 1992

40Fig. 9Combination of layering and partitioning; exampleLRDXCLocal networkRegional networkDigital cross connectFig. 10PDH-based Transport Network Layered Model(CEPT hierarchy)local level, NET 2 at the district, NET 3 atthe regional and NET 4 at the national level.Signal transport in real networks leads toa combination of layering and partitioning,Fig. 9. A 2 Mbit/s signal passes throughdigital cross connects (DXC) located in alocal network, is transmitted via a 140Mbit/s line between the cross connectsand via an STM-1 system between the localnetworks, and then passes againthrough DXCs in a local region. Three ofthe DXCs use the 2 Mbit/s and one of themthe 140 Mbit/s switching level. The exam-pie shows that all Reference Points andTransport Processing Functions involvedin the transport of information can be identifiedas belonging to a certain layer of thetransport network and a certain level of abstraction.In CCITT draft Recommendation G.snal,the transport network is divided into threelayers, which in turn can be divided intosublayers. The layers are as follows:-The Circuit Layer network is concernedwith the transfer of information betweenusers, in direct support of telecommunicationservices. This circuit definition isspecifically used to describe network architectureand differs from CCITT circuitdefinitions used for other purposes-The Path Layer network supports theCircuit Layer network and may consist ofvarious sublayers- The Transmission Media Layer networksupports the Path Layer network andconsists of a function for transfer of informationbetween two nodes and afunction concerned with the physical media.The concepts introduced above are applicableto transport networks based on PDHand SDH.Plesiochronous Digital Hierarchy(PDH)In this article, the following PDH characteristicsignal types are considered:DSn Binary digital signals with datarates and framing according to theorder n in Rec. G.702. (Signalsgenerated and processed in terminalequipment)Dn Digital signals with data rates andpulse masks according to Rec.G.703. (Signals transmitted viastandardised interfaces).ERICSSON REVIEW No. 1-2, 1992

41Fig. 11Functional Model of PDH multiplexerFramed 2 Mbit/s signalHDB-3 coded 2 Mbit's signalClient-to-server AdaptationTrail TerminationIn PDH, the following transport networklayers are defined:- The Path Layer network consists of sublayersfor various data rates (e.g. DSn =2, 8, 34 and 140 Mbit/s sublayers)- The Transmission Media Layer networkconsists of a Signal Layer (e.g. Dn = 2,8, 34, and 140 Mbit/s) and a media-dependentPhysical Media Layer (for thetransmission of non-standardised 2, 8,34, 140 and 565 Mbit/s line signals).The use of the Functional Model will be illustratedby a number of examples basedon the Transport Network Layered modelfor PDH, Fig. 10. The lines between theblocks represent different alternatives fortransporting information. The Physical MediaLayer belonging to the transport networkrepresents the physical media appearingin the transmission network.Plesiochronous digital networks containthe following Transport Processing Functions:PPA Plesiochronous Path Adaptationfor the client-server relationshipsDSn-1/DSn required for multiplexing/demultiplexingand framingPPTSLASLTPPIPlesiochronous Trail Terminationsfor the DSn Trails required for overheadinsertion and extractionSignal Layer Adaptation for theclient-server relationships DSn-1/Dn required for line coding/decodingSignal Layer Trail Terminations forthe Dn Trails required for line codecheckingPlesiochronous Physical Interfacesrequired for the adaptation ofsignals to cables, and vice versa.Fig. 11 shows in detail the unidirectionalsignal transport of 30x64 kbit/s circuitsbetween a multiplexer and a demultiplexer.The lower part of the figure shows thephysical representation in the transmissionnetwork; the upper part the functionalrepresentation in various layers of thetransport network. The send terminal containsthe multiplexing function, generatesthe 2 Mbit/s framed signal, and containsfunctions for signal conversion to HDB-3coding and for connection of the interfacecable. The receive terminal contains thecorresponding functions. The TransportFunctions appear in various transportERICSSON REVIEW No. 1-2, 1992

42Table 2Relationship between Functional and InformationModelFunctional Model,see Fig. 6Managed Objects in InformationModel, see Fig. 18TransportFunctionTrailConnectionTrail (Managed Object)Connection (Managed Object)TransportReferencePointsEnd point of trail APEnd point of a connection CPor network connection TCPTrail Termination Point TTP(Managed Object)Connection Termination Point CTP(Managed Object)Fig. 12Functional Model of PDH muldex and line systemDS1D1PHTFramed 2 Mbit/s signalHDB-3 coded 2 Mbit s signalPhysical Media TrailRelationshipsbetweenTransportReferencePointsand relatedManagedObjectsOne-to-one relationshipbetween AP and CPOne-to-one relationshipbetween CP/TCP and CPOne-to-n relationshipbetween AP in a serverlayer and CP in aclient layerOne-to-one relationshipbetween TTP and CTPexpressed by connectivitypointersOne-to-one relationshipbetween CTP and CTPexpressed by connectivity pointersOne-ton relationshipbetween I l P in a server layerand CTP in a client layerexpressed by namingClient-to-server AdaptationTrail TerminationERICSSON REVIEW No. 1-2, 1992

43layers, Table 1. Each function is characterisedby a number, which appears in tablesand figures in a consistent way. (Thesame number is used for the send and receiveparts).The example in Fig. 12 is based on the examplein Fig. 11 and shows the connectionof the terminals via a line system consistingof line terminals and one regenerator.Thirty 64 kbit/s Circuit Trails are transmittedvia a DS1 Trail which consists ofTrail Terminations, two Sub-Network Connections(TCP-CP) and one Link Connection(CP-CP) in the Path Layer network.These Sub-Network and Link Connectionsare served by three D1 Trails in the SignalLayer network. The D1 Trail in the middleconsists of Trail Terminations and two LinkConnections.The D1 Trail is served by two PhysicalTrails, i.e. two cable sections separated bya regenerator in the physical layer.Fig. 13 shows the Functional Model of adigital multiplexer converting 64x2 Mbit/ssignals into one 140 Mbit/s signal. Terminala is connected to a digital multiplexera, and the multiplexer is connected to a140 Mbit/s line terminal a. Similarly, thecorresponding b-side contains line terminalb, digital multiplexer b and terminal b.The multiplexers and line terminals containresources for management. TransportFunctions 1 to 4 are explained in Table 1and functions 10 to 19 in Table 3.Synchronous Digital Hierarchy (SDH)The functions of SDH systems, as definedin CCITT Recommendations G.781, 782and 783 for digital multiplexers and for digitalcross connects, are much more complicatedthan PDH systems that conformto existing Series G Recommendations (-Blue Book). This is compensated by the increasedflexibility and manageability ofSDH systems, however. Ample overheadinformation, the corresponding resourcesand management interfaces suitable forFig. 13Functional Model of PDH multiplexer (64x2 Mbit/sto 140 Mbit/s)ERICSSON REVIEW No. 1-2, 1992

44No1PPI2PPI3SLTTransport Function(sending)No functionAdaptation totransmission mediaNo functionR---MO---Transport Function(receiving)No functionAdaptation toequipmentLoss of signal checkR--LOSMO——D1TTP4SLABinary to HDB-3conversionHDB-3 to binary codeconversionHDB-3 violation checkBERSINK5PPT6PPA10PPI11PPI12SLT13SLA2 Mbit/s overheadinsertion0.064/2 Mbit/smultiplexingNo functionAdaptationto cableNo functionBinary to CMIcode conversion----2 Mbit/s framealignment2 Mbit/s overheadextraction2/0.064 Mbit/sdemultiplexingNo functionAdaptation toequipmentLoss of signal checkCMI to binarycode conversionCMI code violationcheckLOF, AIS, BERRAI--'-LOS-BERDS1 TTPSINK--D4TTPSINK14PPT140 Mbit/s deframing140 Mbit/s overheadinsertion140 Mbit/s framealignment140 Mbit/s overheadextractionLOF, AIS, BERRAIDS4TTP15PPA34/140 Mbit/smultiplexing140/34 Mbit/sdemultiplexingSINK16PPT34 Mbit/s deframing34 Mbit/s overheadinsertion34 Mbit/s framealignment34 Mbit/s overheadextractionLOF, AIS, BERRAIDS3TTP17PPA8/34 Mbit/smultiplexing34/8 Mbit/sdemultiplexingSINKTable 3PDH Functions, Resources and Managed ObjectsLOSLOFBERAISRAIRMOLoss Of SignalLoss Of FrameBit Error RatioAlarm Indication SignalRemote Alarm IndicationResourceManaged Object18PPT19PPA8 Mbit/s deframing8 Mbit/s overheadinsertion2/8 Mbit/smultiplexing8 Mbit/s framealignment8 MBit/s overheadextraction8/2 Mbit/sdemultiplexingLOF, AIS, BERRAIDS2TTPSINKFig. 14SDH-based Transport Network Layered ModelERICSSON REVIEW No. 1-2, 1992

Fig. 15SDH Transport Functions and SignalsFig. 16Functional Model of SDH multiplexer (63x2 Mbit/sto 155 Mbit/s)TMN will permit SDH-based telecommuni- Fig. 16 shows the Layered Functionalcation networks to be monitored and con- Model of a synchronous digital multiplextrolled.Fig. 14 shows the Transport Net- er converting 63x2 Mbit/s framed signalswork Layered Model based on the SDH -connected via pair cables-into one 155multiplexing rules, and Fig. 15 shows the Mbit/s signal transmitted via a fibre-opticalTransport Functions (as defined in G.783) cable. The figure shows the way in whichand related signals. 2the functional blocks (as defined in G.783)ERICSSON REVIEW No. 1-2,1992

46Table 4SDH functions, Resources and Managed ObjectsFALFEBEFERFBELOPLOSMMPTMMSLMMSRKOOFLTUMFSFSDHMOFrame Alignment LossFar End Block ErrorFar End Remote FailureBlock ErrorLoss of PointerLoss of SignalMisMatch, Path TraceMisMatch, Signal LabelMisMatch between Sent and ReceivedK-bytesOut Of FrameLoss of TU MultiframeSignal FailSignal DegradeResourceManaged Objectof a synchronous digital multiplexer arematched to the transport functions definedin G.snal. Transport Functions 20 to 29are described in Table 4.SDH networks contain the following TransportFunctions (one direction):PI Physical Interface, required foradapting signals for transmissionover cables and vice versa. PI containsthe transport functions PPI,SLT and SLALPA Lower-order Path Adaptation,maps the payload into the containerLPT Lower-order Path Termination,adds VC Path overheadLPC Lower-order Path Connection, allowsflexible assignment betweenlower- and higher-order VCsHPAHigher-order Path Adaptation, processespointer to indicate phasebetween lower- and higher-orderVCs, assembles complete higherorderVCHPT Higher-order Path Termination,adds higher-order path overheadHPC Higher-order Path Connection, allowsflexible assignment betweenSAhigher-order VC and STM-NSection Adaptation, processespointer to indicate phase betweenhigher-order VCs and STM-N, andassembles complete STM-NMSP Multiplex Section Protection, providesswitching to another linesystem for protectionMST Multiplex Section Termination,adds/extracts rows 5 to 9 of theSection overheadRST Regenerator Section Termination,adds/extracts rows 1 to 3 of theSection overheadSPISynchronous Physical Interface,converts to/from in-station or interstationsignals.Networks with digital cross-connectionequipment can be modelled in a similarway. As shown in Fig. 17, N incoming andERICSSON REVIEW No. 1-2, 1992

47Fig. 17Functional model of digital crossconnectDigital cross connect between 155 Mbit s signals,switching of VC4 signalsM outgoing 155 Mbit/s signals areswitched in a digital cross connect whichconnects VC4 signals.The examples show that every detail of anetwork - such as cables, line terminals,regenerators, distribution frames, multiplexersand cross connects - can be modelled.The model shows the ConnectionPoints which might be required for performance,alarm and configuration management.The degree of detail shown in themodel depends on its purpose, e.g. whetherit is intended to model the managementof Network Elements or the managementof Networks.Information modellingManaged Object definitionsThe execution of a management task (aTMN Management service) in a certainarea of application (Management Layer)requires communication between the managingsystem (the Operation System)and the managed system (Networks orNetwork Elements) across managementinterfaces. The Information Model definesthe types and contents of messagesacross the interface which are suitable formanaging a given Managed Object. A ManagedObject is the data representation ofa physical or logical Resource in the managedsystem as seen by the managingsystem across the management interface.In accordance with CCITT RecommendationX.722 (Guidelines for the definition ofManaged Objects), a Managed Object isdefined by the following characteristics:- The attributes visible at its boundary. Attributeshave one or more associatedvalues- The management operations that maybe applied to a Managed Object. Operationscan manipulate the value of aManaged Object attribute (GET, SET,ADD, REMOVE) or the Managed Objectitself (CREATE, DELETE)-The notifications emitted by the ManagedObject- The behaviour exhibited by a ManagedObject in response to management operations- The Conditional Packages that can beencapsulated in the Managed Object. AConditional Package is a collection ofoptional attributes, notifications, operationsand behaviours which are either allpresent or all absent in a Managed Object.Managed Objects with similar attributesand behaviours may be grouped into aspecific object class. An object is characterisedby its object class and object in-REVIEW No. 1-2, 1992

The Managed Objects and their propertiesare contained in Network Information Models.The Generic Network InformationModel defines Managed Objects that arecommon to many different technologies.Specific Network Information Models havebeen developed - or are being developed- for SDH, PDH and ATM networks.Fig. 18Entity Relationship diagram derived from theTransport Functions in Fig. 6TTPCTPn>lTrail Termination Point MOConnection Termination Point MOAdaptation including multiplexingNamingConnectivity pointerManaged ObjectFig. 19Identification of termination point Managed ObjectsNumbers in parentheses are related to PDH functionsin Fig. 13BID Bidirectionalstance and may possess multiple attributetypes and associated values. An objectclass may be a subclass from another objectclass. A subclass inherits attributetypes and behaviours of its superclass, inaddition to possessing its own specific attributesand properties. Each object instanceis identified by a unique name.Such a unique name is composed hierarchicallyand the most common way of organisinga naming hierarchy is through acontainment relationship. The relationshipbetween Managed Object classes can beexpressed with Entity-Relationship diagrams.Inheritance hierarchy diagramsshow the subclassing of Managed Objects.3Generic and specific Managed Objects arelisted in a standardised form in ManagementInformation Bases (MIB). The processof defining "all" new Managed Objectsfor the existing telecommunicationnetworks will take many years, and newManaged Objects have to be foreseen asnew technologies are introduced.Use of FragmentsA number of views, referred to as Fragments,are used as a practical means ofdividing the Information Model into well definedparts. These Fragments can be appliedto modelling - according to the relevantpurpose - more or less independentlyof each other. A Fragment can be illustratedby means of Entity-Relationship diagrams.The following Fragments are derived fromFunctional Models:- Network Fragment- Managed Element Fragment- Termination Point Fragment- Transmission Fragment- Cross-connection Fragment.This article focuses on the TerminationPoint and Cross-connection Fragments.These Fragments assemble the ManagedObjects required for the management ofcross connects, muldexes and linesystems, i.e the main components of transmissionnetworks.The Managed Objects for the three otherFragments will be described on a technology-independentbasis in a future article.Information ModelThe functional modelling, which identifiesManageable Resources, provides the basisfor the information modelling leading toManaged Objects. The various InformationModels are introduced with referenceto the relevant Functional Models.ERICSSON REVIEW No. 1-2, 1992

49TERMINATION POINT FRAGMENTThe Termination Point Fragment decribesManaged Objects which are closely relatedto Resources in the vicinity of endpoints of Trails and Connections. Theseclose relationships are illustrated by comparingFig. 6, showing the Functional Model,with Fig. 18, showing the Entity Relationshipdiagram of the corresponding InformationModel. Table 2 lists TransportFunctions, Transport Reference Points,and relationships between Transport ReferencePoints and related ManagedObjects.Fig. 20Inheritance Hierarchy for Generic TerminationPointsSINK, SOURCE and BIDIRECTIONALTPTTPCTPTermination PointTrail Termination PointConnection Termination PointThe process that leads from the FunctionalModel of the transport network, Fig. 6,to the corresponding Information Model isdescribed below. The Functional Model ofthe transport network describes ConnectionPoints that are related to accessible ornon-accessible points in transmission networkimplementations. It also describesTransport Functions performed by theseimplementations, in which the Resourcesrequired for management are found too.The Network Element aspect provides adetailed description of the NEs, while theNetwork aspect uses only those detailsthat are required for network management.The Functional Model provides an abstractview of telecommunications equipment,based on the essential functions that suchequipment must perform. These functionalcomponents are modelled by ManagedObjects in the Information Model. ManagedObjects represent implementationindependentaspects of equipment andnetworks.Several Resources can be combined intoa Managed Object. The particular Resourcescombined to form a Managed Objectcan be chosen so that they permit- the use of appropriate facilities for themonitoring and controlling of functions- iterative usage of management informationleading to subclasses in hierarchies-the name and the naming structure ofManaged Objects leading to namingtrees.A relationship between Managed Objectscan be expressed either as a Managed Objector as an attribute of one or both of theManaged Objects associated by the relationship.In Fig. 18 two types of relationshipare shown:- Connectivity Pointer relationship, definingthe one-to-one association betweenTermination Points, appears as attributesin both Termination Points- Naming (containing) relationship, definingthe association between n TerminationPoints in the client layer and one TerminationPoint in the server layer,appears as an attribute in the TerminationPoint of the server layer.Fig. 19 illustrates signal transport to andfrom functional units in a network. Resourcesinvolved in signal receiving leadto "SINK" type Managed Objects. Resourcesinvolved in signal sending lead to"SOURCE" type Managed Objects. Forsignal transmission 2 to 140 Mbit/s acrossthe PDH multiplexer shown in Fig. 13,Functions 1-4 lead to "SINK" type andFunctions 10-19 to "SOURCE" type ManagedObjects. In the reverse direction,Functions 10-19 are related to "SINK" andFunctions 1-4 to "SOURCE". Resourcesfor signal receiving and sending lead to "BI­DIRECTIONAL" type Managed Objects."SINK" and "SOURCE" characterisingManaged Objects should not be confusedwith "in" and "out" characterising TransportFunctions.Fig. 20 shows the Inheritance Hierarchy ofTermination Points (TP). Bidirectional TPsinherit the properties from SINK andSOURCE TPs through a multiple inheritancemechanism. This is valid, in principle,for all PDH and SDH Inheritance Hierarchies(but not shown in all the figures).PDH Information modelContrary to the SDH case, for which RecommendationG.774 specifies the InformationModel, no CCITT recommendationfor the PDH Information Model exists atpresent. The Model described below is basedon the principles used in the SDH case.ERICSSON REVIEW No. 1-2, 1992

50Fig. 21Information Model of PDH demultiplexerAll Managed Objects in the figure are of SINKtvoeFig. 22Information Model of PDH multiplexer64x2 Mbit/s signals transmitted via a 140 Mbit/ssignal and vice versaIManaged ObjectERICSSON REVIEW No. 1-2, 1992

51Fig. 23Inheritance hierarchy for PDH Termination PointsTTPCTPTrail Termination PointConnection Termination PointFig. 24PDH naming treeThe choice of transport Managed Objectsis illustrated in Fig. 21, which shows the InformationModel of a PDH 2/0.064 Mbit/sdemultiplexer. The figure is based on thereceive side of Fig. 11 and shows howManaged Objects and their possible locationsare derived from Transport Functions.The Resources in the vicinity of anend point of a Trail or a Connection can beassembled into a Managed Object. Thename of the Managed Object is related tothis end point.Fig. 22, which is based on Fig. 13, showsthe Information Model of a PDH 2/140Mbit/s multiplexer, multiplexing 64 separate2 Mbit/s signals into a 140 Mbit/s signaland vice versa. The multiplexer is connectedto adjacent equipment by means ofinterface cables.However, only those Managed Objectswhich are based on Resources are usablefor management. Table 3 lists Functions,Resources and Managed Objects relatedto the examples in Figs. 21 and 22. The tabletakes into account both transmissiondirections, whereas the figures are concernedwith only one direction.In PDH equipment, Resources normallyappear only in the receive side, leading to"SINK" type Managed Objects.The PDH Information Model can be usedto model, in a general way, all the differentkinds of PDH networks. This leads to theinheritance hierarchy shown in Fig. 23and to the Naming Tree shown in Fig. 24.The inheritance hierarchy consists of a genericpart and a PDH-specific part. EachERICSSON REVIEW No. 1-2, 1992

Fig. 25Information Model of SDH multiplexer63x2 Mbit s signals are transmitted via a 155Mbit s signalManaged ObjectAdaptation FunctionTrail Termination FunctionSub-Network Connection Functionof the managed TPs may be eitherSOURCE, SINK or BIDIRECTIONAL.SDH Information ModelThe SDH Information Model is based onthe Functional Model already presented.Fig. 15 shows the SDH multiplexing structurewith Transport Functions and relatedsignals. Fig. 25 exemplifies the derivationof Managed Objects related to the multiplexingof 63x2 Mbit/s signals into an STM-1 signal. Fig. 26 shows a correspondingexample of an SDH Regenerator. Table 4lists functions, Resources and those ManagedObjects which are based on Resourcesfor the send and receive parts involvedin multiplexing and demultiplexing.SDH multiplexers offer - in addition to themultiplexing function - cross-connectfunctions, which are treated in detail in thenext chapter. (See also Table 5.) The comparisonof SDH and PDH multiplexers clearlyshows the increased management capabilitiesof the SDH concept.The SDH Information Model can be usedto represent, in a general way, all the differentkinds of SDH networks. The correspondinginheritance hierarchy is shown inFig. 27 and the naming tree in Fig. 28.As a general principle, the containment relationshipbetween TTPs and CTPs (TTPcontains CTP) is used for naming. The associationsbetween CTPs and TTPs areeither fixed links represented by Connec-MultiplexerSection LayerFig. 26Information Model of SDH regeneratorERICSSON REVIEW No. 1-2, 1992

Fig. 27, aboveSDH Termination Point inheritance hierarchy* The name of this MO is not included inRec G.774ANSI standardFig. 28, belowSDH naming treeERICSSON REVIEW No. 1-2, 1992

54Table 5Cross-connect functions, Resources and ManagedObjects related to the example in Fig. 25No30LPC31HPCFunction (sending, receiving)Assignment of VC12s within VC4ResourceCross-connectioncapabilityAssignment of VC4s within STM-1 Cross-connection(STN-N in general case)capability* The name of this Managed Object is not included in Rec. G.774Managed Object *VC12TTP-TU12CTPVCTU12XCVC4TTP-AU4CTPVCAU4XCFig. 29Point-to-Point cross connectionsIn the Information Model (Figs. 29-31)Cross Connection Pointers pointing from CTP toCX are not shownManaged ObjectTo/from TP pointer between a CTP and an XCConnectivity pointer between CTPsTo/From TP pointer between a GTP and an XCCross Connection object pointertivity pointers (see Table 2) or flexibilitypoints represented by Managed Objectsbelonging to the cross-connection Fragment(see below).Cross connectionThe cross-connection capability referredto in Table 5 deals with the possibility (withinthe equipment) of freely assigning VCsto available VC-n capacity in a higher orderPath (LPC) or in a multiplex section(HPC). The association can either be anassignment (preparation) or an actualcross connection.In the Functional Model (the left side of figures29,30 and 31) cross connection is thepossible association between a TCP anda CP or between CPs. It is by controllingthis association that networks can be configuredto meet changing demands. Certainrequirements have to be met to makethe control function easier. Examples offunctional requirements are:- It should be possible only to assign (prepare)a cross connection without activatingit (Figs. 29 and 30)- It should be possible to perform point-topointand point-to-multipoint (broadcast)cross connections (Figs. 29 and 31)- It should be possible to cross-connectseveral Termination/Connection Pointsas a single unit without the need to makeindividual cross connections (Fig. 30)- It should be possible to group Termination/ConnectionPoints for managementpurposes like routing, etc.Figs. 29,30 and 31 illustrate how the functionalrequirements listed above are usedto derive both the Functional and InformationModels.Fig. 29 shows a point-to-point cross-connectioncase. The Information Model (theright side of Fig. 29) consists of:-A cross-connection Managed Objectclass (XC) which models the associationbetween the Termination Points involvedNamingIn the Functional Model (Figs. 29-31)ERICSSON REVIEW No. 1-2, 1992

55Fig. 30Group Termination Point cross connection- To/from TP pointers, which are relationshipattributes belonging to the XC ManagedObject and pointing at the CTPs (orGTPs)-Connectivity pointers, which are relationshipattributes belonging to the CTPsand indicating which CTP is connectedto which CTP.In the example, four instances of this Objectclass represent the four point-to-pointcross connections. Three of the four to/-from TP pointers correspond to the threeactivated cross connections. In the fourth- only assigned - cross connection, theConnectivity pointers will be NULL to indicatea non-activated state.Fig. 30 shows cross connections betweendifferent groups of Termination/ConnectionPoints. In addition to the point-to-pointcase, the Information Model consists of:- A Group Termination Point ManagedObject class (GTP) which represents agroup of CTPs to be cross-connected asa single unit- A Cross Connection object pointer,which is a relationship attribute belongingto the GTP and pointing at the XCManaged Object- A "naming", which is a relationship attributebelonging to the GTP and pointing atthe CTPs that constitute the GTP.In the example, the three cross-connect-Fig. 31Broadcast cross connectionERICSSON REVIEW No. 1-2, 1992

Fig. 32, leftCross-connect Fragment, Inheritance hierarchyFig. 33, rightCross-connect Fragment, Naming hierarchyed groups require six instances of GTPsand three instances of XCs. When GTPsare cross-connected, the nth CTP of oneGTP is cross-connected to the nth CTP ofthe other GTP. The assigned (not activated)group 3 in Fig. 30 will be indicated bythose CTPs (which constitute the GTPs)and which have NULL Connectivity pointers.Fig. 31 shows a broadcast (point-to-multipoint)cross-connection case. In additionto the point-to-point case, the InformationModel consists of the Multipoint cross-connectionManaged Object class (MPXC)which represents the assignment relationshipbetween the common CTP and thebroadcast CTPs. One instance of this Objectclass and three instances of the XCObject class will be required to model thecase.The Managed Objects for the cross-connectionFragment are defined in Box 1. Theinheritance hierarchy and naming tree areshown in Figs. 32 and 33 respectively.Box 1Object definitions for crossconnectFragmentThe Fabric Object class represents the set of logicalmatrices that are co-ordinated to provide connectivityof a range of characteristic informationtypes. It manages the establishment and releaseof cross connections. It also manages the assignmentof Termination Points to TP pools and GTPsThe Group Termination Point Managed Objectclass (GTP) represents a group of TerminationPoints that can be treated as a single unit for crossconnection and other administration purposes'TP pool" represents a set of Termination Pointsor GTPs that are equally handled in a number ofmanagement applications, e.g. routing. All theTPs in a TP pool have the same signal typeThe "Multipoint cross-connection" Managed Objectclass (MPXC) represents an assignment relationshipbetween a TP or GTP and the correspondingTPs or GTPs.Reference1 CCITT RecommendationsM.60M3010(M30)M.3020(M.meth)M.3100(M.gnm)M.3180(M.cat)M.3200(M.app)M.3300(M.f)M.3400(M.func)Maintenance Terminologyand DefinitionsTDSG.IV June 1992Principles for a TelecommunicationsManagementNetwork (TMN)COM IV-R28E, part 2, 991TMN interface specificationmethodologyCOM IV-R28E, part 2, 1991Generic Network InformationModelCOM IV-R28E, part 2, 1991Catalogue of TMN ManagedObjectsCOM IV-R28E, part 2, 1991TMN management services,introductionCOM IV-R28E, part 2, 1991F-lnterface managementcapabilitiesCOM IV-R28E, part 2, 1991TMN management functionsCOM IV-R28E, part 2, 1991G.773 Protocol profiles for the Qx-Interfaces for management oftransmission systemsG.774 SDH Network InformationModel for TMNQ.811 Lower Layer Protocol ProfilefortheQ3-lnterfaceQ.812 Upper Layer Protocol Profilefor the Q3-lnterfaceWidl, W.: Telecommunications NetworkArchitecture. Ericsson Review 67(1990):4, pp. 148-162.Widl, W.: CCITT standardisation of TMN.Ericsson Review 60(1991 ):2, pp. 34-51.Rec. Architecture of transport net-G.803 works based on SDHCCITT RecommendationsG.702G.707G.781G.782G.783Digital hierarchy bit ratesSynchronous Digital Hierarchybit ratesStructure of Recommendationson MultiplexingEquipment for SDHTypes and general characteristicsof SDH MultiplexequipmentCharacteristics of SDH multiplexingequipment functionalblocksERICSSON REVIEW No. 1-2, 1992

ERICSSONISSN 0014-0171 Telefonaktiebolaget LM Ericsson 92044 Ljungforetagen, Orebro 1992