分布式混合压缩感知无线传感器网络数据收集

在传感器网络数据收集过程中，降低网络传输量对于网络传输效率和生命周期的延长具有重要意义。结合压缩感知思想，设计了一种分布式混合压缩感知的无线传感器网络数据收集方法。首先通过基于k-means++的方法均匀聚类形成簇，各簇进行基于混合压缩感知的分布式数据收集，完成后通过建立骨干树将数据传输至sink节点。仿真结果表明，在给定的仿真工况下（压缩率为10，节点数为800），与最短路径树混合压缩感知和最优树混合压缩感知算法相比，分别能减少40%和10%以上的传输量，与不使用混合压缩感知的收集方法相比减少70%以上的传输量；同时，节点传输量标准差由14.07和14.37和降低至11.85，置信区间大小由322.66和131.75降低至39.12，证明网络鲁棒性和负载均衡度均有提升。

Distributed Hybrid Compressive Sensing for Wireless Sensor Network Data Collection

During the process of Wireless Sensor Network（WSN） data collection, reducing the number of data transmissions is vital for improving transmission efficiency and extending the lifetime of wireless sensor networks. Drawing on the experience of Compressive Sensing（CS）, a distributed hybrid compressive sensing data collection method for WSN is presented. Sensor nodes are divided uniformly into several clusters by using k-means++. Then distributed hybrid CS algorithm is implemented to gather data inside these clusters. After that, a backbone tree is built to gather data to the sink node. Simulation results show that, under the designated working condition （compressive ratio is equal to 10 and number of nodes is equal to 800）, compared with Shortest Path Tree Hybrid CS and optimal tree hybrid CS method, the proposed method can reduce 40% and 10% total number of transmissions correspondingly. Compared with the method without using CS, the proposed method can reduce over 70% of transmissions. Meanwhile, the standard deviation reduces from 14.07 and 14.37 to 11.85, and the length of confidence interval reduce from 322.66 and 131.75 to 39.12, which indicates that there are great improvements in aspects of robustness and load balance of the network.