提出了一种新的协作频谱感知方法,该方法采用DS证据理论。首先,在本地感知通过投影近似法分解感知结果以减少次级用户上报至融合中心的感知数据,从而降低协作带宽开销。其次,在融合中心根据次级用户证据矢量之间的距离计算可信度,以可信度为权重对每个次级用户的基本概率分配进行加权平均来减小高度冲突数据对检测性能的影响。仿真结果表明,该方法减少了协作感知的带宽开销,改善了数据高度冲突时的检测性能。
An improved cooperative spectrum sensing scheme based on dempster-shafer evidence theory in cognitive ra-dio networks was proposed. In the local sensing period, the projection approximation algorithm was introduced to reduce bandwidth for cooperation. In the fusion period, credibility was calculated by the distance between evidence vectors of different secondary users and adopted as the weights to combine different local sensing data. The new scheme reduces channel bandwidth and simulation results show that the new scheme improves the performance under the condition of high level conflict.