Workflow Scheduling Using Hybrid GA-PSO Algorithm in Cloud Computing

Table 1

Literature review summary.

Author	Name of Algorithm	Objective	Advantages	Limitation
Braun et al. [16]	min-min algorithm	Time	12% better than GA	Delayed large tasks for long time
Kumar and Verma [20]	Combination of min-min and max–min strategies in Genetic Algorithm	Time	Faster than the GA	Time consuming
Guo et al. [21]	Particle Swarm Optimization (PSO) algorithm	Execution and transfer time	Faster than the M-PSO and L-PSO algorithms in a large scale	Stuck in local optimal solution
Pandey et al. [23]	Heuristic algorithm based on particle swarm optimization	Time and cost	Three times better cost compared to BRS, good load distribution over resources	Stuck in local optimal solution
Arabnejad and Barbosa [24]	Heterogeneous Budget-Constrained Scheduling (HBCS) algorithm	Execution time and cost	Reduction of 30% in execution time while; maintaining the same budget	Not considering the load over resources
Verma and Kaushal [6]	Bicriteria Priority Based Particle Swarm; Optimization (BPSO) algorithm	Time and execution cost	Decreasing the execution cost; compared to BHEFT and PSO	Not considering the load over resources
Xu et al. [25]	Heuristic algorithm based on the min-min algorithm	The fault recovery, the time, and the cost	Fault recovery has a significant impact on; the two performance criteria	Better choice only when both cost and makespan are considered
Chitra et al. [26]	The PSO algorithm	Load balance and the makespan	Better than GA and PSO	Time consuming
Ge and Wei [27]	The Genetic Algorithm	Load balance and makespan	Better than FIFO	Time consuming to reach to optimal solution
Fard et al. [28]	The heuristic algorithm	Makespan, economic cost, energy consumption, and reliability	Improve all four objectives	Not efficient with small number of tasks and processors
Wu et al. [29]	The Revised Discrete Particle Swarm Optimization (RDPSO) algorithm	Makespan, communication costs, and computation costs	Better than the standard PSO and BRS (Best Resource Selection) algorithm	Not efficient with large search space
The proposed algorithm	Genetic and particle swarm optimization algorithm	Makespan, communication costs, load balance, and execution and transfer time	Faster convergence to the solution in comparison with other approaches	Supports one data center without considering the dynamic workflow