A Node-Level Congestion Avoidance Algorithm for Reliable Transmission of Key Information in Wireless Sensor Networks

The communication ability of a wireless sensor network node is limited, when an event occurs, the data generation rate may increase sharply, which may result in network congestion. While the key information which characterizes the occurrence of an important event is critical for users, it is very important to transmit them timely and reliably for users’ high quality decision-making and effective response. Therefore, the paper puts forward a node-level congestion avoidance algorithm which can guarantee the reliable transmission of key information, named CAARTKI. The main idea of this algorithm is to divide the service by introducing differentiated services, and the packets are divided into different priority according to their importance. In this algorithm, a node sends data only when its next hop node has assigned a sending window for it, which may avoid the occurrence of node congestion, and congestion avoidance is not only considered in link layer, but also, in routing layer, a neighbor node with more buffer spaces is chosen as the next hop node. Thus key information can be transmitted timely and reliably to nodes with lighter load, and the possibility that important information can not be sent in time due to heavy load can be reduced significantly. When the key information is generated intensively, the algorithm adopts the active loss strategy and frees the cache for packets with higher priority through discarding some packets with lower priority. The NS2 simulation results show that this algorithm can alleviate congestion occurrence, the packet loss rate of the highest priority packets is extremely low and its average network delay is small, and key information can be transmitted reliably and quickly.