Categorization and performance analysis of advanced avionics algorithms on parallel processing architectures

Some of the emerging candidate advanced avionics algorithms for next-generation aircraft and missions are discussed. Details of algorithm categorization and architecture classification and categorization are presented. In addition, a methodology for mapping the algorithms into the parallel-processing architectures is developed. In order to determine the algorithm performance on candidate architectures, a task graph is constructed for the critical processing areas, indicating data dependencies, memory accesses, throughputs, data rates, primitives, and timing requirements. Both the primitives and processing architectures are categorized, and a heuristic mapping is performed. For each task graph and primitive, candidate architectures are analyzed in order to determine performance metrics.<
>