Fully static processor optimal assignment of data-flow graphs

Abstract

The data-flow graph (DFG) is an important graph-theoretic model for multiprocessor implementation of real-time digital signal processing (DSP) algorithms. Given a time schedule for a DFG,, we consider the problem of the processor-optimal assignment for a fully static schedule. Previously, the solution of this problem was found by solving an integer programming problem. In this letter, we propose a linear programming approach to solving the problem.