Comparison of Diversity, Nondiversity, and Coding on a Parallel FSK HF Channel

Selected short block and convolutional code improvements using dual inband frequency diversity and nondiversity frequency-shift keyed experimental data obtained on a 2800-km HF path are compared. It is shown that time dispersion techniques (interleaving, diffusion, etc.) are necessary for the short codes to perform measurably better than uncoded dual inband diversity With interleaving, more than two orders of magnitude improvement resulted. The measured binary errors on a nondiversity FSK subchannel (one of 16) are shown to be statistically dependent (nonrandom) for transmission frequencies between the path lowest usable frequency and maximum usable frequency. The range of gap and burst distributions are given indicating the small difference in distribution for the set of frequencies and times on this particular mid-latitude HF path. Comparison of nondiversity and diversity burst distributions present the burst length reductions and change in distribution brought about by diversity.