Cluster Computing Architectures and Algorithms for Passive Sonar Arrays

This paper summarizes ongoing research and development efforts involving the design, modeling, simulation, experimentation, and analysis of distributed, parallel computing algorithms and architectures for autonomous sonar arrays. Several fine-grain interconnection network models have been developed including unidirectional linear array, bidirectional linear array, and unidirectional register-insertion ring. Basic time-and frequency-domain algorithms for parallel beamforming have been designed and implemented, and promising performance results have been achieved on experimental Ethernet-, ATM-, and SCI-based cluster testbeds in the HCS Research Lab.