Congestion Control In Wireless Sensor Networks

The nodes in WSN(Wireless Sensor Networks) have low computation capacity, con-strained energy and weak communication ability. WSN often experiences congestion dueto these constraints on resources and the communication with the many-to-one and multi-hop pattern. In WSN applications, effective congestion control is the key technique toprolong the network system lifetime and improve the network QoS performances. Al-though some work related to congestion control in WSN have been proposed, they oftenconsume more energy and can not guarantee some QoS metrics, such as transport fairnessand real time. This dissertation devotes to two fundamental problems in congestion issueof WSN, congestion causes and congestion control techniques. To be more specific,(1) This dissertation introduces the concept of congestion threshold to describe thecritical phase of packet generation rate, beyond which some nodes will experience bufferover?ow, leading to network congestion. Congestion threshold connects the transport ca-pacity of WSN and the congestion of WSN, and represents both the maximal transportcapacity of WSN and the maximal ability of resisting congestion. By using betweennessto describe the traffic load per node, the author studies the congestion threshold of WSNand presents the asymptotically upper bound, and discusses the impacts of wireless en-vironment, traffic load, and network scale on the transport capacity of WSN, providingsome insights for the designing of congestion control in WSN.(2) The author systematically designs a rate-based congestion control scheme forWSN, called RbCC. In RbCC, an energy-efficient congestion detection method is pre-sented, which can effectively detect congestion by referring the buffer change behavior.RbCC also involves a bi-directional congestion feedback to faster transfer the local con-gestion information toward the correlated source nodes. RbCC makes the source ratefaster converge into a stable level by a local rate adjustment scheme and a closed-loopsink-source rate adjustment, decreasing the number of dropped packets and improvingthe transport fairness.(3) The rate-based congestion control is more suitable to the WSN application withstreaming data, and not suitable to the applications requiring highly the real-time andreliable performance. The author designs a window-based congestion control scheme WbCC, which lets the upstream nodes send packets in only the case that the related down-stream nodes have some available buffer spaces, instead of detecting node congestion. Forguaranteeing the real time and transport fairness, WbCC involves a simple but effectivereal-time-oriented queue schedule method and a sending window assigning method. InWbCC, the ?ow information corresponding to each source node is not necessary to bemaintained, and only low computation load is needed to guarantee that the packet earlygenerated can reach the sink early.(4) The author investigates a general WSN applications in which the packet gener-ation is very low, the application fidelity is small and transient sudden source load occursoccasionally. He presents an algorithm STC to control the source traffic by referring toboth the diversity of remained energy of different source nodes and the fidelity require-ment of WSN application. STC improves the system lifetime. In addition, the authorconsiders the congestion control issue in the routing. He designs a congestion-sensitiveand energy-efficient algorithm CsEeR. With the above analyzing results of the transportcapacity of WSN, CsEeR aims at balancing the energy of network nodes and considersfully the impact of wireless interference and the energy level on the single node. CsEeRalmost avoids the congestion and prolongs effectively the system lifetime.