面向宽频段频谱感知的WSN架构及其分簇算法

针对动态频谱管理、频谱监测等领域对宽频带频谱感知的需求，提出了一种面向宽频段频谱感知的无线传感器网络系统架构，采用分簇的网络结构、分频段协作感知和分级的感知结果处理机制，实现了对宽频段频谱的有效、高精确度感知。重点研究了该架构下的分簇算法，提出了一种面向感知任务频段特性的分布式分簇算法。算法基于近邻传播模型，通过分布式消息交互积累网络结构信息，实现将信噪比相似的节点分在较少簇中的分簇目标。仿真表明，提出的分簇算法生成的簇数目随着节点数量的增加线性增长，簇结构在簇首与簇成员信噪比标准差和能量差等指标上均优于现有算法，说明了算法在可扩展性、支持感知结果融合和提高网络工作时间等方面的优越性。

Framework of WSN for wideband spectrum sensing and clustering algorithm

For the wide band spectrum sensing requirement in dynamic spectrum management and spectrum monitoring, a wireless sensor network framework for wide band spectrum sensing was proposed. Through clustered network structure, frequency division based cooperative sensing and cooperative sensing result processing scheme, the framework aims to achieve efficient and precise spectrum sensing. Specifically, the clustering algorithm following this framework was researched and a Mission Spectrum Oriented Clustering algorithm based on Distributed Affinity Propagation (MSOC DAP) was proposed. Through distributed message exchange and network structure information accumulation, the nodes sensing the same band were clustered into fewer clusters with fewer inter cluster SNR difference. Extensive simulation result shows that the generated cluster numbers of the proposed algorithm increase linearly with nodes numbers in the network and the generated cluster structure performs better than the existing algorithms on SNR mean standard deviation and mean energy difference. These demonstrate the preference of MSOC DAP on expansibility, supporting sensing result fusion, and prolonging the network longevity.