Implementation/acceleration of vectorized acoustic source localization algorithms on the CELL multi-core processor

This paper describes the optimized design and implementation of sonar signal processing algorithms (matched filter) on the CELL multi-core processor. The algorithm is modified to achieve maximum parallelism to enhance time performance on the multi-core platform. In this proof-of-concept effort, we show a factor of 10 speedup by parallelizing the matched filter code on 16 separate computing nodes on the CELL processor for broadband matched filter implementation. This demonstration of considerably faster signal processing capability should be of substantial interest to the signal processing community in general, and the oceanic engineering community in particular.