Practical considerations of using adaptive beamforming with real sonar data

Degradation of a sonar display may arise due to a variety of forms of unintentional acoustic directional interference such as, for example, ship propeller noise, engine noise, or maritime seismic activity etc. In contrast, and within a military role, interference in the form of acoustic countermeasures (ACMs) may be adopted deliberately as a defensive tactic to evade active sonar tracking. In the literature, for example Coates (1990) and Owsley (1992), it is shown that theoretically, in the context of active sonar analysis, standard adaptive beamforming (ABF) techniques may be used to decrease the effects of countermeasure activity. The difficulty, however, in applying these adaptive beamforming algorithms to real sonar data, stems from inherent practical limitations, particularly in terms of sample estimation errors and suppression of the target signal within the processing frame. In this paper, the performance of two such standard ABF algorithms, namely, minimum variance distortionless response (MVDR) and unbiased adaptive response (UAR), is investigated through their application to real sonar ACM trials data. In light of the practical limitations of such adaptive algorithms, two modifications are presented, to (a) combat the effects of sample estimation error, and (b) circumvent the partial nulling of the target signal