Power Spectrum Estimates of Sampled Pseudo-Random Sequences

摘要

The report concerns power spectrum estimates of sampled pseudo-random signals, and shows that for these signals the spectral power is more relevant than, say, the autocorrelation function. The power spectrum estimates considered here assume stationarity and zero means. The signal under investigation is a pseudo-random noise signal obtained from maximal length sequences. Pseudo-random signals are popular because (a) they can be fashioned with relative ease from linear shift registers, (b) their deterministic waveforms have uniform power throughout their bandwidths, and (c) their autocorrelations are approximately zero outside a narrow peak at zero tau. This study is concerned with some applied methods and results of these applications to bandlimited signals that have been corrupted by additive-noise, multiple-path, and Doppler effects and that have been processed digitally to obtain correlations and power spectra. The object was to observe the effects on the power spectrum estimates due to these perturbations. The analysis was performed exclusively on filtered maximal length pseudo-random sequences. These filtered sequences were tailored to a bandwidth and sample size to yield optimal results for a given computer memory capability. (Author)