Exploring parallel processing levels in turbo demodulation

Iterative processing in a wireless base-band receiver ensures promising error rate performance at the cost of high processing time for data retrieval. Exploration of parallelism in iterative processes, while maintaining the error rate performance, contributes toward fulfilling the high transmission rate requirements of emerging wireless communication standards. In this paper, parallelism levels available in the constituent components of turbo demodulation system have been classified. To compare the parallel execution performance to a serial, results of a high speed parallel turbo demodulation software model are presented. For a highly parallel turbo demodulation system, comprised of 16 demappers and 32 decoders running concurrently, the throughput can be enhanced 27 times as compared to a serial architecture for 16-QAM modulation. Parallelism efficiency is also tabulated for 16-QAM and 256-QAM, with different architectural configurations which can benefit the hardware designer about architectural efficiency to attain high throughput.