WSEAS Transactions on Computers
Print ISSN: 1109-2750, E-ISSN: 2224-2880
Volume 14, 2015
A Genetic Task Migration Algorithm for Fault Recovery in NoC-Based Manycore Systems
Authors: , ,
Abstract: Recovery from permanent core faults in NoC-based manycore systems usually requires migrating tasks from faulty cores to fault-free cores, where balanced workloads are commonly desired. Finding optimal migration destinations for tasks, however, is a challenging issue due to the time complexity of the search process. To cope with this, in this paper, a genetic algorithm based task migration algorithm is proposed, where the adaptive crossover (AC) scheme and the An chaotic mapping disturbance (AD) scheme are incorporated to improve the search efficiency. Experiments show that the modifications to the standard genetic algorithm (SGA) are effective, and the proposed task migration algorithm outperforms two PSO-based algorithms, the SGA and a deterministic algorithm (DA) in finding more balanced migration solutions. The average improvements of the proposed algorithm over the DA, PSO, DPSO, and SGA algorithms are 86%, 88%, 37%, and 16%, respectively.
Search Articles
Pages: 487-501
WSEAS Transactions on Computers, ISSN / E-ISSN: 1109-2750 / 2224-2880, Volume 14, 2015, Art. #49