中文摘要：

依据传感器网络面向应用的价值区分度特征，提出一种基于冗余价值滤波的传感器网络节能数据收集机制。所提机制采用预测模型在线评估采样数据价值，并映射为相应的价值因子，进而根据强化学习理论将价值因子引入区分服务的退避机制设计，驱动媒体介质访问层层竞争窗尺寸的自适应优化调整，在满足数据收集服务质量的前提下，有效地抑制网内价值冗余负荷传输量，实现价值区分性滤波的节能效果。仿真实验表明，所提机制能有效增加网络吞吐量和降低传输时延，且相对于一些传统的节能收集机制，能够从传感器网络数据内涵应用价值挖掘的角度，更有效地降低网络整体能耗。

英文摘要：

An energy-saving filtering mechanism （EFM） by mining the value redundant loads in networksaccording to distinctive valuable grade in application-oriented wireless sensor network （WSN） is proposed. The reinforcement learning theory, which relies on online estimation of the data value, is adopted to evaluate the da-ta value online and drive the adaptive optimization decision on contention window in medium access control. Fur- thermore, on the premise of quality of service （QoS） in data-gathering, the transmission of value redundancyloads can be effectively inhibited in networks to realize the energy-saving gathering mechanism based on mining the intension of value in transmission loads. Finally, the simulation results show that EFM can effectively reduce total energy cost in WSN via decreasing a large amount of redundant flow in network, enhance QoS of data gathering, and outperform some other classical data collection schemes in execution efficiency.