An Experimental Study of Congestion Control in Wireless Sensor Networks Using a Combination of Resource and Traffic Control

Wireless Sensor Networks (WSNs) consist of a number of spatially distributed autonomous sensor nodes that are configured to gather physical or environmental information in an area. A sensor node can be subject to many constraints including limited power, low processing and sensing capabilities, small memory, small bandwidth, and low communication range. In WSNs, when an event occurs, a sudden burst of network traffic will be generated which may lead to congestion and hence to lost or delayed data packets, decreased reliability, and increased power consumption. One approach that controls the congestion in WSNs is the resource control method, in which congested nodes are avoided by sending the packets through alternative paths. To test the existing methods of congestion control e.g. the Hierarchical Tree Alternative Path (HTAP) algorithm and to develop a new method, an experimental facility consisting of 40 nodes (Waspmotes) was deployed. In order to accurately measure end to end delay and to ensure nodes wake up at the same time, the node clocks were synchronized to an accuracy of approximately 40μsec between nodes and to 5ms between nodes and external time, using a method applied experimentally for the first time. The experiments demonstrated that HTAP has better performance (Packet Delivery Ratio - PDR, throughput, and End to End delay) than a network without congestion control. However, in some network topologies where alternative paths do not exist, in the event of congestion, the network performance will be degraded. Therefore, traffic control, which reduces the packet rate, has also been employed to mitigate congestion. The location of the single path affects the network performance with poorer performance when it is close to the sink than to the source. Based on the results, using both resource control and traffic control leads to the best network performance.