基于协作频谱预测的认知网络吞吐率分析

频谱预测是将预测结果传递给次级用户（Secondary User，SU），使SU有选择性地实施频谱感知，提高频谱感知的有效性。但是存在预测结果不准确的情况，影响整个网络的吞吐率。在基于遗传算法优化的神经网络预测模型基础上，提出了SU进行协作的频谱预测方法，提高了SU预测空闲信道的准确率。讨论了协作频谱预测条件下，在通信强度、协作用户数量、信道数量不同时的系统吞吐率。仿真结果表明协作频谱预测比传统非协作频谱预测系统吞吐率有较大提升。

Throughput performance of CRN based on cooperative spectrum prediction

Spectrum prediction provides its predicting results to sensing channels selectively for Secondary Users（SUs）. But the results are often inaccurately, which limits the throughput performance of the whole Cognitive Radio Network（CRN）. Based on the Genetic Algorithm optimized training for Neural Network（GA-NN） spectrum prediction model, a novel cooperative spectrum prediction scheme is proposed. The probability of the SU idle channels sensing is significantly enhanced. The impacts of traffic intensity, cooperative SU number and channel number on the CRN throughput are also investigated respectively in this paper. The simulation results indicate that the throughput with cooperative spectrum prediction is significantly improved compared with traditional spectrum prediction in CRN.