SIGNAL DETECTION AND PARAMETER IDENTIFICATION WITH APPLICATION TO UNDERWATER ACOUSTICS (PERIODIC RANDOM SEQUENCES, EXTENDED M-ORTHOGONAL).

摘要

A problem of passive detection of underwater moving acoustic sources is formulated. It is found that solving the exact problem, when a multipath acoustic propagation environment is assumed, is fairly difficult. Keeping the multipath assumption intact we develop two abstract problems that are solvable and yet are sufficiently related to reality to allow us to form guidelines and rules of thumb for the passive detection problem. Multipath propagation from a moving acoustic source is characterized by doppler and delay parameters on each path. We assume that sound is received at two remote receivers. If there are M paths to one receiver and N paths to the other receiver the normalized crosscorrelation of the receptions will exhibit M.N peaks.; We use the concept of Extended M-Orthogonal signals to unify the information on the ambiguity plane and to find how many of the peaks on the ambiguity plane need to be localized for acceptable detector performance.; The M.N peaks are generated by M + N independent sets of doppler and delay parameters. Periodic Random Sequences are used to address the question: Should the detector treat the M.N peaks as independent entities or as being generated from M + N independent parameter sets? i.e. should the detector have M.N degrees of freedom or M + N degrees of freedom?