TABU SEARCH vs GENETIC/HEURISTIC ALGORITHMS TO SOLVE THE REAL-TIME MULTITASK/MULTIPROCESSOR ASSIGNMENT PROBLEM

摘要

The problem of mapping a set of preemptible real-time tasks over a set of heterogeneous processors is NP-complete. The assignment is performed taking into consideration five constraints: time, allocation, resource, communications and precedence. Two methods, Tabu Search and Genetic/Heuristic are applied to solve the problem. Results are analysed and a comparative evaluation performed.