The Use of Linear Time Series for Prediction of Congestion Detection in Wireless Sensor Networks

A Wireless Sensor Network (WSN) is developed with large number of sensor nodes. Packet transfer in this network presents a range of challenges to protocol designers due to resource constrains, limited battery power, processing power, memory and storage capacity of sensor nodes in WSN. The applications that produce high volumes of data which require high transmission rates, may cause congestion in the sensor node and leading to packet loss and impairments in the quality of service (QOS) as well as throughput of networks. If data transmission to the network is not controlled, congestion status can arise and decrease network lifetime. Therefore, we need various congestion detection mechanisms to identify congestion. In this paper, we present a Prediction based Congestion Detection (P-CD) technique in order to identify congestion before congestion occurrence. These techniques use queue length as a parameter to recognition congestion. The introduced technique has better prediction accuracy.

Keywords:

Wireless Sensor Networks (WSNs) , Congestion detection , Queue length , Prediction , ARIMA model.