Traffic Prediction-Based Fast Rerouting Algorithm for Wireless Multimedia Sensor Networks

Abstract

Rerouting has become an important challenge to Wireless Multimedia Sensor Networks (WMSNs) due to the constraints on energy, bandwidth, and computational capabilities of sensor nodes and frequent node and link failures. In this paper, we propose a traffic prediction-based fast rerouting algorithm for use between the cluster heads and a sink node in WMSNs (TPFR). The proposed algorithm uses the autoregressive moving average (ARMA) model to predict a cluster head's network traffic. When the predicted value is greater than the predefined network traffic threshold, both adaptive retransmission trigger (ART) that contributes to switch to a better alternate path in time and trigger efficient retransmission behaviors are enabled. Performance comparison of TPFR with ant-based multi-QoS routing (AntSensNet) and power efficient multimedia routing (PEMuR) shows that they: (a) maximize the overall network lifespan by load balancing and not draining energy from some specific nodes, (b) provide high quality of service delivery for multimedia streams by switching to a better path towards a sink node in time, (c) reduce useless data retransmissions when node failures or link breaks occur, and (d) maintain lower routing overhead.