Integration Protocols for Voice and Data Traffic

Increasing demands for multimedia services offer integration of multimedia traffic as a hot issue in the future research areas. As a result, in the literature, many multiplexing schemes have been proposed. However, most of them have been implemented with a high complexity, others may be non-effective to satisfy the multiplexing performance criteria, while the rest are still not subjected to a wide range of analysis. Therefore, there is a critical need for comparing some of the recommended multiplexing schemes as well as developing a simple and effective integration protocol while still achieving reasonable bandwidth utilization. This thesis is intended to examine integration protocols for multimedia traffic, with primary focusing on voice-data integration. Firstly, a survey of the existing multiplexing schemes and related issues are presented. Next, an Adaptive Round Robin (ARR) protocol is proposed, as an alternative for voice-data integration, and extensively simulated. Finally, further comparisons, based on computer simulations, are carried out for various multiplexing schemes including Strictly Priority Servicing (SPS), Fixed Round Robin (FRR), Dynamic Bandwidth Allocation/(T1, T2) and Queue Length Threshold (QLT).As a contribution of the thesis, the proposed protocol tries to avoid the drawbacks of the previous multiplexing schemes besides satisfying the multiplexing performance criteria. The protocol differs from the others in that, it gives a limited priority for voice over data, it organizes the incoming packets to the single First-in First-out (FIFO) output buffer rather than the only outgoing scheduling, i.e., all data sources are polled in order according to the adaptation policy; however, before a data source can send a packet, all active voice sources are polled in order. Thus it provides an improvement in voice delay performance without significant effect on data delay performance over previous protocols. In addition, simulation comparisons between various multiplexing schemes have been discussed. In these simulations voice packets are assumed to be generated from on-off sources (talkspurt-silence calls), which is closer to reality and which is not considered in most of the performance analyses of previous schemes.

PDF FK_1997_9_A.pdf Download (778kB)

Actions (login required)

View Item