Space-time adaptive processing on the Mesh Synchronous Processor

This paper presents an assessment of the applicability of the Mesh Synchronous Processor (MeshSP) architecture to a class of radar signal processing algorithms known as space-time adaptive processing (STAP). The MeshSP is a SIMD processor developed at MIT Lincoln Laboratory and marketed by Integrated Computing Engines, Inc. A high-level overview of the MeshSP architecture is presented. Mappings of a matrix decomposition algorithm to the MeshSP architecture are described. A performance model for estimating computation and communication latencies for the Householder algorithm on the MeshSP using different mapping approaches is presented. Results of the performance model are discussed.