Traffic Protection via Bandwidth Scavenging in Heterogeneous Sensor Networks

This paper presents a history based statistical channel access mechanism for enabling traffic prioritization in wireless sensor networks. Prioritized access is realized such that low priority non-real-time sensors can access channel bandwidth that is unused by high priority real-time traffic. The key idea is for the low priority sensor nodes to first observe and statistically model the channel usage pattern by the high priority traffic. Then make probabilistic transmissions depending on the amount of time elapsed after the most recent high priority packet transmission ends. The objective is to dimension such probabilities based on the channel utilization statistics so that the non-priority traffic throughput is maximized while protecting the high-priority traffic from disruptions. The proposed access mechanism is implemented in a TelosB mote based sensor testbed in which the non-priority motes continually measures the RSSI to infer the channel usage pattern and probabilistically access the channel while different types of traffic is sent by high-priority TelosB motes. Experimental results from the testbed demonstrates that the proposed mechanism can improve non-priority traffic throughput by up to approximately 70% over compared protocols, while limiting the disruptions to high-priority traffic to pre-specified bounds of 3% to 5%.