Coding for Spread-Spectrum Communication Networks.
Bal Gi Kim · Deep Blue (University of Michigan) · 1987
The multiple-access capability of a frequency-hop packet radio network is investigated from a coding point of view. The achievable region of code rate and channel traffic and the normalized throughput are considered as performance measures. We model the communication system from the modulator input to the demodulator output as an I-user interference channel, and evaluate the asymptotic performance of various coding schemes for channels with perfect side information, no side information, and imperfect side information. The coding schemes being considered are Reed-Solomon codes, concatenated codes, and parallel decoding schemes. We derive the optimal code rate and the optimal channel traffic at which the normalized throughput is maximized, and from these optimum values the asymptotic maximum normalized throughput is derived. The results are then compared with channel capacities. It is shown that the capacity of interference channels with perfect side information and no side information can be achieved by Reed-Solomon codes and concatenated code respectively. For channels with imperfect side information, it is found that the parallel decoder performs better than the errors- and -erasures decoder. The performance improvement becomes more significant as the side information becomes less reliable.