Abstract: Parallel and distributed system plays a key role in the development of high performance systems. To achieve the high performance of a system, task scheduling is an important issue. At common, the problem of task scheduling has been considered to be NP-hard. Several algorithms put into practice to find the optimal schedule for task scheduling. Evolutionary algorithms are one of the best. In an evolutionary kind of algorithms, the time taken to find an efficient schedule is high. This study presents the implementation of Non-dominated Sorting Genetic Algorithm (NSGA-II) with MapReduce model. In a distributed system, most of the task scheduling problem is formulated as multi-objective optimization problem. For multi-objective problem formulation, minimization of makespan and flowtime is considered. MapReduce model can automatically parallelize the execution of NSGA-II. The algorithm is tested on a set of benchmark instances. Experimental results show that NSGA-II with MapReduce model minimizes the amount of time taken, makespan and flowtime than a Weighted Sum Genetic Algorithm (WSGA) with MapReduce model which is also implemented in this study for better comparison.
D. Rajeswari and V. Jawahar Senthil Kumar, 2016. Design and Implementation of Non-Dominated Sorting Genetic Algorithm Scheduler Using Mapreduce Model. Asian Journal of Information Technology, 15: 2584-2593.