不可靠链路下基于压缩感知的WSN数据收集算法

Compressive sensing based data gathering algorithm over unreliable links in WSN

为了解决WSN中基于压缩感知(CS,compressive sensing)的数据收集方法会受不可靠链路影响的问题,首先通过实验对基于CS的数据收集算法中数据重构信噪比与链路误码率的关系进行了定量研究,根据WSN链路分组丢失特性将分组丢失问题分为轻负载和重负载2种情况。针对轻负载下的链路不可靠,建立随机分组丢失模型,并提出了基于邻居拓扑空间相关预测的CS数据收集算法,利用数据空间相关性减小错传的影响。针对重负载下的链路不可靠,建立节点伪失效模型,并提出了基于稀疏调度的CS数据收集算法,通过改变观测矩阵稀疏度,避免观测出错数据,弱化不可靠链路的影响。仿真分析表明,在不增加能耗的前提下有效提高了数据重构质量,降低了不可靠链路对CS数据收集的影响。

To solve the problem that the ubiquitous unreliable links in the WSN influence the performance of the compressive sensing（CS） based data gathering, first the relationship between the reconstruction SNR of CS-based data gathering algorithm and the bit-error-ratio（BER） were simulated quantitatively. Then classify two cases were classified, namely light-payload and heavy-payload, relying on the analysis of wireless link packet loss characteristics. The random packet loss model was conceived to describe the packet loss under light-payload scenario. Further the neighbor topology spatial correlation prediction-based CS data gathering（CS-NTSC） algorithm was proposed, which utilized the nodes spatial correlation to reduce the impact of error. Additionally, the node pseudo-failure model was conceived to describe the packet loss occurred in network congestion, and then the sparse schedule-aided CS data gathering（CS-SSDG） algorithm were conceived, for the purpose of changing the sparsity of measurement matrix and avoiding measurements amongst the nodes affected by unreliable links, thus weakening the impact of error/loss on data reconstruction. Simulation analysis indicates that the proposed algorithms are not only capable of improving the accuracy of the data reconstruction without extra energy, but also effectively reducing the impact affected by the unreliable links imposed on CS-based data gathering.