Research of Improved Particle Swarm Optimization Based on Genetic Algorithm for Hadoop Task Scheduling Problem

Scheduling is NP-hard problem in Hadoop, because scheduling algorithm must use available resources to complete assignments in the shortest time. This paper proposes an improved Genetic-Particle Swarm Optimization (IG-PSO) algorithm to solve scheduling problems. Traditional PSO algorithm is easy to fall into local optimum solution, so novel improved Genetic-Particle Swarm Optimization (IG-PSO) algorithm introduced GA's mutation and crossover to overcome the shortcoming and increase the ability of global optimization. Compared with traditional PSO and GA, the experiment simulation shows that IG-PSO algorithm can escape from local optimal solution and find a better global optimal solution. Because the position of PSO particle falls into local optimal solution, GA uses mutation and crossover to diversify particles, which make the particle escape out of local optima.