Analysis of asynchronous frequency-hop spread spectrum multiple-access networks.

Abstract

This thesis is concerned with analyzing the performance of Asynchronous Frequency-Hop Spread-Spectrum Multiple-Access (AFHSS-MA) networks sending one binary symbol using Binary Frequency Shift Keying (BFSK) modulation. Analytical expressions are derived for the probability of uncoded bit error using the theory of spherically symmetric r and om vectors. These expressions are exact when orthogonal BFSK is employed and are an approximation when nonorthogonal BFSK is employed. Using these expressions for the bit error probability, we show that a system which makes hard decisions performs better than a system with perfect side-information that erases the symbols that are hit. The reverse was thought to be true based on analyses assuming that the bit error probability is $1\\over 2$ (or 1) whenever a hop is hit. Using similar techniques, we consider Viterbi Ratio Thresholding (VRT) for generating information about the quality of the channel in an AFHSS-MA network and find that employing VRT results in substantial improvement in performance over hard-decisions. We also consider the performance of binary convolutional codes and find that the usual bounding techniques for the first event error probability and the packet error probability are very loose. Accordingly, we simulate the Viterbi decoder in order to obtain a more accurate estimate of the system performance. Finally, we consider slow AFHSS-MA networks. We compute the probability that l out of the n bits transmitted in a hop will actually be hit given that the hop is hit. Using this, we analyze the performance of a proposed concatenated coding scheme. We find that it performs significantly better than a system employing test patterns to obtain imperfect side-information and also better than a system using perfect side-information to erase the symbols transmitted in a hop that was hit.