Research Of Task-oriented Self-organization Mechanism In Wireless Sensor Networks

Wireless Sensor Networks (WSNs) is a task-oriented wireless ad hoc network system, usually composed of a large number of densely deployed sensor nodes in a certain monitoring area and one or more data gathering nodes located in or near the area. Due to the sensor nodes have limitations in power energy, communication ability, computing ability and storage capacity etc., so how to make a number of sensor nodes work together and accomplish certain tasks and optimize the network topology, is an important and difficult problem of designing a self-organizing wireless sensor networks.It can be seen from the existing methods of self-organization, self-organizing wireless sensor networks is mainly implemented by routing protocol or topology control technology, and these two methods mainly focus on how to improve the network energy efficiency and prolong the lifetime of the network. But in the practical application of wireless sensor networks, usually tasks have higher requirements of quality, especially when tasks have a high real-time requirement, it need to sacrifice part of the node’s energy, through synergy and self-organizing process, to complete the task with high quality and efficiency. Since the sensor nodes are small and inexpensive, so design a task-oriented self-organization mechanism is reasonable and desperately needed.Based on distributed task allocation scenario, this paper proposes a self-organizing wireless sensor networks mechanism based on Q-learning, the mechanism mainly focus on how to complete tasks with high quality and efficiency through the self-organizing process among sensors. The mechanism statistics the situation of task completion, choose non neighbor nodes who complete a large number of tasks and neighbor nodes who accomplish few tasks or never complete tasks by setting the frequency threshold, and add these nodes into the source node’s candidate neighbor set, then use the Q-learning algorithm to learn the optimal path to reach the node who contribute resource, and adjust the connection between the two nodes; at the same time, delete the source node’s neighbors who complete few tasks. By doing this, the mechanism can optimize the network topology and ensure that in the premise of efficient allocation of tasks to reduce the average delivery latency, and automatically adapt to changes in the external environment.In addition, on the basis of comprehensively considering the changeable situation of topology structure in wireless sensor networks, we proposes a self-organization mechanism in open environment, the mechanism analyzes the dynamic changes of topology in wireless sensor networks, while new nodes join the network or existing nodes leave the network, the topology of wireless sensor networks will have dynamically change, and we also puts forward the corresponding solutions. This framework enables each sensor node to build a cooperative neighbor set based on TDR and past experience. Such cooperative neighbors, in turn, can help the sensor to efficiently allocate tasks in the future.Experimental results demonstrate that the proposed mechanism outperforms three other approaches in terms of average delivery latency, successful delivery ratio, and the time consumption to finish a simulation run in the same scale of deployment of the same number of nodes in the network. And the two mechanisms proposed in the paper can optimize the network topology of wireless sensor networks, reduce the average delivery latency, and efficiently accomplish tasks. And it also shows the potential application of two mechanisms.