Research on Link Quality Estimation Mechanism for Wireless Sensor Networks Based on Support Vector Machine

• DOI: 10.1049/cje.2017.01.013
• Publication Date: 2017-03-16T18:08:06Z
• Authors: Jian Shu; Song Liu; Linlan Liu; Liqin Zhan; Gang Hu
• Keywords: Link (geometry); Kernel (algebra); Tree (set theory)

In the application of Wireless sensor networks (WSNs), effective estimation for link quality is a basic issue in guarantying reliable data transmission and upper network protocol performance. A mechanism proposed, which based on Support vector machine (SVM) with multi-class classification. Under analysis wireless characteristics, two physical parameters communication, Receive signal strength indicator (RSSI) Link (LQI), are chosen as parameters. The divided into five levels according to Packet reception rate (PRR). model SVM decision tree established. built kernel functions radial basis polynomial respectively, RSSI, LQI input experimental results show that reasonable. Compared recent published models, our can estimate current accurately relative small number probe packets, so it costs less energy consumption than one caused by sending large packets. So this high efficiency saving prolong life.