Performance of Noncoherent MFSK Channels with Coding

Abstract

Computer simulation of data transmission over a noncoherent channel with predetection signal-to-noise ratio ST/N₀ = 1 shows that convolutional coding can reduce the energy requirement by 4.5 dB at a bit error rate of 0.001. The effects of receiver quantization and choice of number of tones are analyzed; nearly optimum performance is attained with eight quantization levels and sixteen tones at ST/N₀ = 1. The effects of changing ST/N₀ are also analyzed; for lower ST/N₀, accurate extrapolations can be made from the data, but for higher ST/N₀ the results are more complicated. These analyses will be useful in designing telemetry systems when coherence is limited by turbulence in the signal propagation medium or oscillator instability.