Analysis of a Computer Simulation of a DIMUS System

Abstract

In the usual shift-register/delay-line arrangement, as detailed by Anderson (1960, 1961) and Rudnick (1960), geometry of the array, in conjunction with sound velocity in the medium, establishes a time delay for each wavefront arrival at a particular phone. The pressure waveform is then stored in each shift register, and is then clocked at a rate sufficiently above the Nyquist frequency, such that degradation of zero crossover information is avoided. Sampling the shift-register matrix, at points appropriate to the array geometry, will result in a gain for the signal emanating from a given direction. One characteristic feature of the DIMUS system is an amplitude normalization, which devolves from the binary nature of quantization. Thus, from the maximum signal-to-noise ratio to the minimum, from +∞ to −∞ dB, a response ratio corresponding to the directivity index is achieved. For 100 hydrophones, this range would be 20 dB (re power); for 10 phones, 10 dB, etc. The levels are thus predictable, regardless of pressure level of the source as seen at the array. The paper discusses the analytic and computer simulation aspects of these phenomena. Correlation functions are derived analytically and compared with simulation results for several cases.