A Development of Spatial Skyline Query Based on Surrounding Environment for Data Streaming Using Apache- Spark

Abstract

Previous studies of spatial skyline queries based on the surrounding environment posed a challenge in handling skyline searches that supported mobile users. This study introduces a method which allows users to search for spatial objects dynamically. Online streaming data services are currently available to support user movements. With this condition, streaming data requires a longer execution time in processing. This work examines the effectiveness and efficiency of the Apache-Spark framework in developing spatial skyline queries based on the surrounding environment algorithm for data streaming. Further implementation of this algorithm on mobile devices can provide better location access for users. Comparative analysis of processing time was carried by comparing single processing, parallel distributed computing, and cluster computing algorithms with various measurement evaluation parameters. Result of the test on "number of types in surrounding environment" parameter shows that the execution time of parallel and cluster computing is faster than single computing, 63.694 and 72.772 times faster, respectively. Test on the "raw data size" parameter indicates that parallel computing execution time is 46.553 times faster than single computing, while cluster computing is 74.187 times faster than single computing. In the "main facility search radius" parameter, cluster computing is 45.224 times faster than single computing, and parallel computing is 77.022 times faster than single computing. Lastly, the "number grid" parameter indicates that the execution time for both parallel and cluster computing is faster than single computing, 157.944 and 276.251 times faster, respectively.