基于变分贝叶斯推断的新型全局频谱协作感知算法Novelcooperative.PDF

第37 卷第2 期 通 信 学 报 Vol.37 No.2 2016 年2 月 Journa l on Communications February 2016 doi:10.11959/j .issn.1000-436x.2016037 基于变分贝叶斯推断的新型全局频谱协作感知算法 吴名，宋铁成，胡静，沈连丰 (东南大学移动通信国家重点实验室，江苏 南京 2 10096 ） 摘 要：为了实现多维动态频谱接入，首先给出了主用户的全局功率谱近似模型，并构建了新型全局频谱协作感 知算法的总体流程，以获得主用户网络中占用频段、功率及位置等全局信息。接着利用变分贝叶斯推断技术，设 计了相应的模型系数向量估计器。仿真结果表明，该方法采用的近似模型具有较好的准确性，相应的系数向量估 计算法具有较高的有效性和收敛稳定性，同时指明了信噪比和泄漏总虚假功率的关系以及两者对均方误差性能的 影响。此外，还证明了该方法通过利用系数向量? 的稀疏性，而在均方误差性能上具有较大优势。 关键词：认知无线电；全局频谱协作感知；变分贝叶斯推断；稀疏性 中图分类号：TN914 文献标识码：A Novel cooperative global spectrum sensing algorithm based on v ariational Bayesian inference WU Ming, SON G Tie-cheng, HU Jing, SHEN Lian-feng (N ationa l Mob ile Communications Research Laboratory, Southeast University, N anj ing 2 10096, China ） Abstract: To realize mu lti-dimensional dynamic sp ectrum access, an approximat e model w as prop osed for the glob al p ower sp ectral dens ity (P SD) of prim ary users (PU). Based on the proposed model, a novel cooperative spectrum sensing algorithm w as prop osed, and its overall flow w as also built to obtain glob al information in the n etwork of PU. The glob al information inc luded locations, occup ied frequency b ands and transmitting p owers of the PU. Then, an estim ator of mod- el coefficient vector was designed by utilizing the th of variational Bay esian inference (VBI). Simu lation resu lts show that the proposed approximate model has good accuracy, and the corresponding estim ation algorithm of model coefficient v ector has good convergence and stab ility. Meanwhile,