Research on Node Scheduling Algorithm of Sensor Network based on Support Vector Machine

Abstract

In the current Internet era with information explosion, SN (Sensor network) has important scientific research value and social and economic value in many fields because of its micro-intelligence and huge and basic perceptual data. Ensuring the coverage quality of the network cannot be at the expense of network connectivity. The decrease of network connectivity is equivalent to the failure of some nodes in the network, and the resulting data collection hole will affect the judgment of end users. In this paper, the scheduling algorithm of SN nodes based on SVM (Support vector machine) is studied. Considering the excellent performance of EN (ensemble learning), a scheduling algorithm of EN-LS-SVM is proposed by introducing bagging and random subspace method. The research results show that the network lifetime of the proposed algorithm is extended by 180s and 13.28% compared with the traditional algorithm on the premise that the network coverage is not less than 80%. The algorithm proposed in this paper can reduce network coverage vulnerabilities, keep network coverage at a high level and ensure network QoS.