A jointly designed source encoding-multiple access protocol for wireless communications

The problem of designing an efficient network protocol that maintains the quality of service (QoS) for different types of traffic has both theoretical and practical significance. From a network point of view, admitting more customers and as signing the network resources more effectively among them is an important issue. On the other hand and from a customer's point of view, quality of the delivered service is the most important matter. However, there is a trade off in satisfying both of these two requirements. More specifically, a higher number of customers admitted to the network brings about a higher network loading that results in network congestion to occur more often; and network congestion causes degradation in QoS. An intelligent network protocol tries to employ an efficient way of packing the source information and to provide the promised QoS to the customers with assigning the network resources among them as needed. This requires the network layer to be aware of some characteristics of the application layer, or in other words, the network layer and the application layer should be designed jointly.; This dissertation presents SEAMA, a source encoding assisted multiple access protocol, to integrate voice and data in a micro-cellular wireless network. SEAMA employs a multi-state, embedded voice encoding scheme for encoding and packing speech information. A multi-state voice encoder reduces the average bandwidth that a voice call needs. Thus, SEAMA exploits the time variations of the speech coding rate, through statistical multiplexing of voice packets, to efficiently use the available bandwidth and to increase network utilization. In each frame, SEAMA allocates bandwidth among voice calls as needed. The use of the embedded encoder makes it possible for the network to resolve congestion by dropping low priority packets, when congestion occurs. The flow control mechanism of SEAMA takes full advantage of the embedded nature of the source encoding scheme to handle rate assignment among voice calls in each frame for an optimal QoS. Since during congestion periods, the network selectively drops some of the less significance packets, the voice quality degrades gracefully. Our analyses demonstrate that SEAMA provides soft capacity for TDMA communication networks through an optimal congestion control mechanism. Therefore, voice quality is less sensitive to network congestion. The admission control policy of SEAMA meets a QoS requirement based on the delivered signal-to-noise ratio (SNR) or average distortion per call. We show that SEAMA exploits the characteristics of the source encoder to provide an acceptable quality for voice traffic and achieves a significant gain in the network utilization.; In this thesis, a new methodology for increasing the network utilization is proposed. Performance analyses of both the voice section and the data section of SEAMA are presented. For both voice and data, new approaches for analyzing the system are presented. For voice a new way of evaluating the performance is presented. For data, unlike previously reported results, a dam theoretic technique and also a diffusion approximation are outlined. From a theoretical point of view and for a given level of QoS, the achievable network capacity is obtained. From a practical point of view, the performance results presented in this dissertation illustrate that SEAMA improves the network capacity about 100% compared to a circuit switched network and about 20% compared to its competitor, the PRMA (packet reservation multiple access) protocol.