The Realize Of Matrix Operation Based On FPGA

Intensive matrix operations are widely used in signal processing and image processing.This operations always require the system for real-time. And the architecture with highthroughput rate is therefore highly sought for. So it's useful to find ways to realize matrixoperations for high speed. Compared with the other method, using FPGA has its ownunique advantages because of its high speed, and it's easy to realize parallel computationstructure by FPGA. This dissertation mainly discusses the realization of matrix operationbased on FPGA.First, the characteristic and the theoretic of matrix operations are introduced. And thedesign of floating-point operation unit using VHDL is described. On this basis, the matrixmultiplication cumulative module, the triangular matrix inversion module and the generalmatrix inversion module are designed. In order to solve the problems which brought by theexpansion of matrix order, the partitioned matrix algorithm to realize high order matrixoperation is discussed. Then the general matrix inversion module is simulated in Modelsim.The precision of result compared with Matlab, time and resources occupied by FPGA areanalyzed. Then the programmer is downloaded on Virtex-4 FPGA hardware platform torealize matrix inversion. And the result is transfered to PC by USB. Accordingly, thecorrectness and feasibility of matrix operation based on FPGA are validated. At last, theuse of matrix inversion in Radar Signal Processing is simply introduced.