Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Protocol) and RSVP [8] (Resource ReSerVation<br />
Protocol).<br />
The other approach has been termed Adaptive<br />
QoS Middleware. This consists of both an<br />
enhanced end terminal stack as well as QoS brokers,<br />
mobility gateways and media filters located<br />
within the network – forming a complete distributed<br />
architecture for QoS management (Figure<br />
3). Applications are presented with a standard<br />
API, rather than having to deal with session<br />
and QoS negotiation and violations themselves.<br />
The BRAIN project has recognised the merits of<br />
both approaches and developed a modular, component-based<br />
architecture that encompasses<br />
both. We have designed enhancements to the terminal<br />
stack (Figure 4) that provide interfaces to<br />
a number of different application types: legacy<br />
(type A); those that utilise session protocols<br />
(type B); those that can make use of a component<br />
API (providing frame grabbers, packetizers<br />
...) (type C); and those that can make use of a<br />
full blown QoS broker to deal with all connection<br />
issues (type D).<br />
The QoS enabled transport interface, also called<br />
Service Interface (SI), allows a complete separation<br />
of the BRENTA from a specific network<br />
implementation which could be provided by<br />
over-provisioning or by more complex protocols<br />
such as RSVP. Well-known transport primitives<br />
at the SI interface will be enhanced by additional<br />
primitives to give applications the facility to use<br />
QoS, with a specific focus on the needs in the<br />
wireless and mobile domain.<br />
The Access Network<br />
The BRAIN Access Network (BAN) (Figure 5)<br />
is intended to be deployed over a campus sized<br />
area – a city centre, Science Park or airport.<br />
Within this domain it will offer terminals Quality<br />
of Service (QoS), Terminal Mobility and<br />
Vertical Hand-over support. Personal mobility<br />
and service creation will largely reside outside<br />
of the BAN.<br />
Terminal mobility will be provided by a modified<br />
micro-mobility protocol such as MER TORA [9],<br />
[10] or HAWAII [11]. A mobile host terminal<br />
will acquire an IP address for at least the duration<br />
of a session and the routers of the BAN will create<br />
per-host entries to allow packets to reach the<br />
mobile, wherever it moves within the domain. At<br />
present the BRAIN project is evaluating the currently<br />
proposed protocols for micro-mobility<br />
against an evaluation framework [12] derived<br />
from the perceived requirements from the usage<br />
scenarios outlined above.<br />
The evaluation framework consists of three<br />
parts. Firstly a deconstruction of the protocol<br />
Telektronikk 1.2001<br />
Application<br />
Session<br />
layer<br />
Transport<br />
layer<br />
Network<br />
BRAIN<br />
Air Interface<br />
layers<br />
Type A Type B Type C Type D<br />
design issues involved: fast handover, paging,<br />
security, address management and so on. Secondly<br />
a classification of micro-mobility protocols<br />
against these design issues. Finally an evaluation<br />
against requirements such as scalability,<br />
robustness and efficiency.<br />
Results present so far ([12], [13], [14]) show that<br />
there are only three fundamental approaches to<br />
IP micro-mobility and that a per-host routing<br />
approach is probably optimal – for BRAIN scenarios.<br />
Another important area of BRAIN research is to<br />
produce an interface definition for the layer2/3<br />
User<br />
Equipment<br />
BRAIN<br />
ACCESS<br />
NETWORK<br />
BRAIN<br />
<strong>Wireless</strong><br />
Routers<br />
BRAIN QoS<br />
Broker GUI<br />
Broker BRAIN high level API<br />
BRAIN component level API<br />
Session layer protocols (H.323, SIP;...)<br />
QoS enabled transport interface<br />
QoS and mobility support<br />
Transport layer<br />
IP layer supporting QoS and mobility<br />
Service Specific Convergence Sublayer for IP<br />
HIPERLAN/2 DLC layer<br />
HIPERLAN/2 physical layer<br />
BRAIN Mobility<br />
Gateway<br />
Figure 4 BRAIN End Terminal<br />
Stack<br />
Figure 5 The BRAIN IP<br />
access network<br />
IP Network<br />
BRAIN ACCESS<br />
NETWORK<br />
QoS<br />
Mobility Management<br />
Control<br />
61