r indian Joural of Enginering & Material'Sciences Vol. 6, October 199 , pp.29-28 Integration of voice, video and data services in a Big 'a' network using a new access protocol K Gunavathi & A Shanmugam PSG College of Technology, Coimbatore 61 0, India Received 1 Ma 1 998; accepted 18 June 1999 A new high performance multiple access protocol has been proposed for bus LANs. The efciency of the proposed network has been analyzed under the new protocol for integration of voice, video and data services. It has been found that, for given network parameters, the protol yields high throughput and less packet delay which is also bounded compared to most of the known collision pne and collision-free rotocols. Also, the protocol is observed to be suitable for widespanned and high-speed networks, and real time applications. Recently, fast and wide-band packet switching has emerged s a promising long-term technology for integrated services. The network using the new access protocol is tested for voice, video and data applications. It is assumed that the service of stream trafc (voice and video) is exhaustive and data trac is nongated. Both interactive and bulk data services are considered. Performance of the network for integration of the above services under the new protocol is compared to that of a collision-prone protocol (the Fast Etheret), and a collision-free protocol (the Expressnet). t is obsered that the new protocol gives better results and is suitable for multimedia applications of wide spnned and high-speed networks. An event driven simulation of the network activities for integrated services of voice, video and data is carried out for wide range of network parameters and trac (stream and data parameters. A bus network is characterized by a parameter ' a' where 'a' is defined as the ratio of end-to-end propagation de lay to the packet transmission time. The Etheret net works are suitable only up to a < 0.5 1.2 . A Big 'a ' net work is one which performs well for values of a > 1. Such networks are suitable for high-speed applications and for wider span. 4 . The bus network using the new access protocol performs well for a > 1. and is classi fied as Big 'a' networkS It is reported that the protocol performs better than many of the known collision-free and collision-prone protacols 7• In this paper, performance of Big 'a ' network is analyzed for the application of voice, video and data services using the new access pro tocol and compared to that of Fast Etheret (collision prone protocol) and Expressnet (a collision-free proto col). Proposed Protocol In the proposed protocol for multi channel LANs., the network is assumed to consist of M' stations, evenly spaced on the bus. The distances between the adjacent stations are equal. The stations are uniquely indexed from left to right along the bus. The physical ordering of the stations over the bus is same as the logical order in which ' For corespondence they are served. The stations are divided into two gr Q oups namely group 1 and group 2. The stations of group 1 are indexed as 1 ,2, .... MI2 (M12 + 1 ) for M even (odd) the stations of group 2 are indexed as MI2 +2,(MI2 + 1) , . . . . M for M even (odd). The chanel bandwidth is divided into two sub-bands in the ratio p:q with the assumption that p :: q. Each group is allotted a sub-band. The sta tions of a group are entitled to transmit only through the sub-band allotted to it but are flexible to receive from both the bands. This ensures communication between any two stations of the network. Also, two simultaneous transmissions are possible through the two sub-bands of the network. This, indeed, increases the average utiliza tion of the channel more as compared to other protocols. The access of the shared medium is made collision free by adopting round-robin service nd a proper sched uling delay policy by the stations. Each station staggers its potential transmission by calculating a scheduling delay such that the station would be able to detect the beginning of transmission by a higher order station be fore its delay time expires. The scheduling delay (H is calculated by every station dynamically each time by hearing the end of transmission of the currently trans mitting station. Also, the protocol empoys a non-gated reversing seq R uential serice (NRSS). Every station s