Evolutionary Computing Assisted K-Means Clustering based MapReduce Distributed Computing Environment for IoT-Driven Smart City

In the last few years, the exponential rise in urban population and allied demands have alarmed governing agencies as well as industries to achieve more quality-of-service (QoS) oriented solutions to meet up-surging demands, especially towards real-time decision making, information exchange and knowledge-driven decisions. To achieve it, smart city concept which employs Internet- of-Things (IoT), distributed software computing, and BigData analytics has gained widespread attention. Though, inclusion of QoS-sensitive routing has helped enabling better and efficient sensory or node's data collection and dissemination; however, ensuring optimal query-driven knowledge mining and information exchange has remained a challenge. Considering it as motivation, in this paper an evolutionary computing assisted K-Means clustering algorithm is developed for MapReduce computation in Hadoop distributed framework. The proposed method employs genetic algorithm to enhance centroid estimation as well as clustering, which as a result helped in achieving better clustering to support MapReduce. The proposed GA based K-Means clustering has been applied over Hadoop-MapReduce, where to achieve aforesaid centroid estimation and clustering enhancement Silhouette coefficient was used as the objective function. Here, GA-K Means was applied in such manner that it estimates optimized centroid and clusters simultaneously over Mapper and Reducer, which makes overall computation faster and more accurate.