Abstract: Cleaning sensory data is an important problem in wireless sensor networks（WSNs）.Existing cleaning algorithms haven’t considered the factor of sensor density’s ununiformity and information congestion，so they will consume more energy.This paper takes these two factors into account，and proposes a density-based adaptive algorithm named DHA（density-based hybrid approach） for cleaning sensory data in WSNs.The DHA algorithm can do better decision for cleaning sensor data along with different node density，and it adopts adding weight to data in time dimension which makes it response fast to a data change.The experimental results show that DHA can conserve more energy than existing best algorithm WMA（weighted moving average-based） while cleaning effectively and offering quicker response time.

Key words: wireless sensor network, noise, sensor density, algorithm for cleaning sensory data

摘要： 在传感器网络研究领域中，去除感知数据含有的噪声是个重要的研究课题。现存的去噪算法没有考虑节点密度不均匀及信息拥塞的情况，从而过多地消耗了能量。考虑这两个因素，使用时间维加权的方法，提出了一个基于节点密度的网内自适应去噪算法－DHA（density-based hybrid approach）。DHA能够根据节点密度来进行算法决策，并且在时间维进行加权，能够对数据变化作出快速反应并且提高数据精度。实验结果表明，DHA方法能够在保证良好的去噪效果、快速响应时间的前提下，比目前最好的去噪算法WMA（weighted moving average-based）更节省能量。

关键词: 传感器网络, 噪声, 节点密度, 感知数据去噪算法