Scheduling multiprocessor job with resource and timing constraintsusing neural networks

摘要

The Hopfield neural network is extensively applied to obtaining annoptimal/feasible solution in many different applications such as thentraveling salesman problem (TSP), a typical discrete combinatorialnproblem. Although providing rapid convergence to the solution, TSPnfrequently converges to a local minimum. Stochastic simulated annealingnis a highly effective means of obtaining an optimal solution capable ofnpreventing the local minimum. This important feature is embedded into anHopfield neural network to derive a new technique, i.e., mean fieldnannealing. This work applies the Hopfield neural network and thennormalized mean field annealing technique, respectively, to resolve anmultiprocessor problem (known to be a NP-hard problem) with no processnmigration, constrained times (execution time and deadline) and limitednresources. Simulation results demonstrate that the derived energynfunction works effectively for this class of problems