Study of Ds/ssma Receivers in the Low Frequency Band (Communications, Detection, Direct-Sequence, Multiple-Access, Spread-Spectrum)

107 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1986.Receivers for asynchronous direct-sequence spread-spectrum multiple-access (DS/SSMA) communications over the low frequency (LF) band are studied. We consider a system which uses binary phase-shift-keyed modulation and a system in which each user's signature sequence set consists of M orthogonal sequences and the signature sequences use r-phase modulation. The LF atmospheric noise samples are modeled as an (epsilon)-mixture whose first-order probability density function is a mixture of a Gaussian density representing thermal and background noises and a Cauchy density representing impulsive atmospheric noise.Assuming that the signal strength is small, the locally optimum Bayes detector (LOBD) is derived. The LOBD for the binary signaling system consists of a chip correlation filter followed by a sampler, a zero-memory nonlinearity, and a discrete-time matched filter. The LOBD for the r-phase signaling system consists of inphase and quadrature chip correlation filters followed by samplers, a nonlinearity, and 2M discrete-time matched filters. The shape of the locally optimum nonlinearity depends on the noise distribution. The average probability of error of the LOBD is evaluated. The receiver performance for long sequences is analyzed via a Gaussian approximation.For easier implementation, some suboptimum but simpler nonlinearities are considered as replacements for the locally optimum nonlinearity. In particular, the clipper and the blanker, which depend only on the clip level, are considered. The performance of the receiver, when the locally optimum nonlinearity is replaced by these suboptimum nonlinearities, is analyzed, and a performance comparision with the LOBD is made. The performance analysis shows that the clipper and the blanker perform nearly as well as the LOBD. It is also found that the clipper and the blanker are quite insensitive to errors in estimating the optimum clip level.The suboptimum nonlinearities considered for the r-phase signaling system are functions of the envelope detector output. Thus, there is a joint nonlinear processing of both the inphase and quadrature components. We introduce a further simplified receiver structure in which the sampled outputs from inphase and quadrature channel filters go through separate nonlinearities. We analyze the performance of this receiver and make a performance comparision. (Abstract shortened with permission of author.)U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD