Evolutionary Computation Approach for Spatial Workload Balancing

The growing demand for Geographic Information Systems (GIS) calls for high computation reliability to handle vast and complex spatial data processing tasks. A better parallel computing scheme should ensure balanced workload at different data processors to ensure optimal use of computing resources and minimise execution times, which poses more challenges with spatial data due to the nature of having spatial correlations and uneven distributions. In this paper, we propose a spatial clustering approach for workload balance, by using an evolutionary computation method that considers the nature of spatial data, to increase the computation performance for processing GIS polygon-based maps with massive number of vertices and complex shapes. To evaluate our proposed approach, We proposed two different experimental approaches for comparing our results: (i) Non–merging based experiment, and (ii) merging based experiment. The results demonstrated the advantage of the proposed spatial clustering approach in real GIS map based partitioning scenarios. The advantages and limitations of the proposed approach are discussed and further research directions are highlighted toward a development work by the research community.