Genetic algorithm-aided fixed-point design of E-UTRA PRACH detector on multi-core DSP

This paper presents a new genetic algorithm (GA)-aided methodology of software implementation in Digital Signal Processors (DSPs) under both the computational accuracy and bus bandwidth constraints. The design issue is firstly stated as two classes of fixed-point design problems, one of which is then formulated to a constrained integer programming (CIP) problem. And the genetic algorithm is proposed to treat with such CIP problem for the sake of efficiency. Then the fixed-point evolved (E)-UTRA PRACH detector is presented, which further underlines the feasibility and convenience of applying this methodology to practice. Finally, the numeric results justify the proposed GA-aided approach and demonstrate that a speedup by a factor of 33 can be achieved compared to the exhaustive search for the solution of E-UTRA PRACH detector design problem.