A Novel Discrete Particle Swarm Optimization Algorithm for Job Scheduling in Grids

Collaborative multi-swarm PSO for task matching using graphics processing units

“…Clearly, this is not the case for task mapping as a task cannot be mapped to machine 4.32427, but, rather, must be mapped to machine 4 or 5. Unlike Kang et al [16] or Yan-Ping et al [4], we do not move to a modified discrete PSO algorithm, but maintain the use of the continuous domain in the solution space. We compared simple, single-swarm implementations of continuous versus discrete PSO and found that continuous provides improved results, as shown in Figure 1.…”

“…Opposite to continuous PSO, Kang et al [16] propose an implementation of discrete PSO for task mapping on the grid. They compare the results of their discrete PSO implementation to continuous PSO, the min-min algorithm, as well as a genetic algorithm.…”

Job Scheduling Algorithms on Grid Computing: State-of- the Art

“…Nevertheless, this may decrease the search space rate of convergence as successive generations may disappear. The limitations of GA and DE affected the performance of the job scheduling problem on computational grid as GA and DE produced long makespan and flowtime compared to other mechanisms [3,15,18,24,52].…”

“…A discrete particle swarm optimization method (DPSO) mechanism for job scheduling on computational grid was introduced in [18]. In DPSO, particles are represented as a vector of natural numbers.…”

“…This problem affects the PSO speeds and directions [61]. The disadvantages characteristics of PSO affected the performance of PSO in the process of scheduling jobs on computational grid as the standard PSO produces acceptable but not optimal schedules in terms of makespan and makespan times [18,20,35].…”