Adaptive coding for wireless data networks.
Tingfang Ji · Deep Blue (University of Michigan) · 2001
Current cellular systems typically have spectral efficiency in the range of 0.03--0.05 b/s/Hz/sector and the talk time is about 3 hours. The dramatic increase in demand for high speed wireless data services requires the next generation wireless networks to significantly improve the throughput and energy efficiency while maintaining a reasonable delay. In this thesis, we investigate rate adaptive coding and automatic repeat request (ARQ) schemes to meet these new challenges. An analytical framework is investigated for performance evaluation of general type-I and type-II hybrid ARQ protocols based on rate compatible error correcting codes over finite-state channels. As an example, an analysis of Reed-Solomon code based systems is shown to yield insight into the effect of the protocols, channel parameters, design parameters, and decoder implementations on the system performance. A family of powerful rate compatible multiple product codes are constructed and shown to have superior performance compared with S-random interleaved turbo codes at moderate signal-to-noise ratio. A few practical rate adaptive ARQ protocols are then applied to an asynchronous direct-sequence code division multiple access network and compared in terms of the total network throughput, average delay, and throughput under energy constraints.