Four orders of magnitude cost reduction in the Viterbi algorithm for GSM MIMO demodulation

It is shown that multiple complementary techniques for trading BER performance for cost reduction in the Viterbi algorithm for multiple-input/multiple-output joint demodulation can be merged to obtain substantial and efficient cost reduction. Quantitative results for application to the separation of GSM and IS-136 cochannel signals are presented. It is shown that up to six signals can be separated at a GSM base station with BER of 10% before decoding gain with only a single dual-polarized antenna and E/sub b//N/sub o/=10 dB in real time on a Compaq 1 GHz ES-40 server, with cost savings of up to four orders of magnitude. It is also shown that for a specified BER and number of signals, the cost of MIMO demodulation decreases as the number of sensors increases. Finally, it is shown that for a fixed computational cost, BER decreases by a factor of eight, on average, for each sensor added.