Resource Allocation Scheme for Wireless Sensor Networks: Machine Learning Approach
Gururaj S. Kori, Mahabaleshwar S. Kakkasageri · 2022
Wireless Sensor Network is a distributed and decentralized adhoc network comprising of powerful sensing, computing and processing nodes. Sensor nodes are restricted in terms of energy i.e. battery power, communication range, bandwidth, computational latency, and storage. Effective usage of Wireless Sensor Networks (WSN) resources is a challenging task, to enhance network lifespan, increase throughput, reduce computational delay and minimize the control overheads. Several intelligent strategies are proposed by adopting intelligent resource management schemes. In WSN, intelligent resource management involves resource discovery, resource scheduling, and resource (bandwidth) allocation. In this paper, a Classification and Regression Tree (CART), Machine Learning (ML) protocol is applied to deal with incomplete information about the WSN i.e., uncertainty during effective bandwidth allocation. The scheme operates as follows: 1) k-means clustering algorithm is applied to the network, clusters are formed & cluster head is selected. 2) k-NN algorithm is applied to find the number of neighbor nodes in the cluster. 3) The attributes of the Cluster Members (CM) % Cluster Head (CH) like distance from base station, their degree of connectivity, congestion rate of the network, type of data & size of data aggregated at CH after performing task, and channel quality etc are calculated. 4) Aggregation and classification of CM & CH attributes (data sets) using intelligent search and feature selection algorithm. 5) Data set are further processed for training phase, predication phase and decision tree model is built to achieve target attributes i.e. bandwidth allocation. 6) Heat matrix & confusion matrix are generated and performance evaluation of proposed scheme is done. From simulation results we observe that the proposed CART scheme enhances the performance of WSN in terms of resource allocation accuracy, allocation computational delay, and data transmission efficiency etc.