A multi-rate tdma protocol and associated speech quality issues
Mathew Thomas, D.J. Goodman · 1997
Multi-rate PRMA is proposed in this thesis as an adaptive and distributed TDMA protocol to support multi-rate services on wireless systems. The PRMA protocol is extended to support multi-rate sources. The supported data rates are integral multiples of a minimum rate of one slot transmission per frame. A dual-rate voice only system, wherein active users can operate at either half- or full-rate, is presented as a design example. Relevant protocol and performance issues behind a variable rate TDMA protocol are explored using this system, and performance metrics are obtained. Results indicate that the system is capable of a wide range of operational modes. Wireless data services are becoming ubiquitous and they are also well-suited for multi-rate operation. Against this backdrop, the issue of voice data integration in the context of the dual-rate system is studied. The system design process is used to address protocol and performance issues behind multi-service integration under the general constraint of higher priority for voice users. Two extreme policies for data bandwidth allocation are investigated, and throughput and delay measures are obtained. The impact on the existing voice service is also monitored and these results yield feasible system operating modes. In wireless systems, voice call quality may be affected due to a variety of reasons, and obtaining real speech data from a working system for evaluation is difficult. Core issues underlying such quality evaluation problems are identified and a general framework to address these issues is presented. Any distortion artifact of interest is characterized by means of suitable statistical distribution(s) obtained via measurement, modeling or simulation. These distributions serve as constraints while introducing the distortions into a set of test speech segments. The test set with statistically representative distortion is used for subsequent quality measurement. This framework is applied to the dual-rate system studied before. A Markov model is proposed to characterize the distortion mechanism and the model parameters are estimated. Results indicate that the model is sufficiently accurate and it is incorporated within the evaluation procedure. Using this procedure, test speech segments were generated and used for subjective quality evaluation. These results provide operational and system design guidelines.