应用神经网络抑止OFDM系统载波干扰

Applying neural network to suppress OFDM system carriers interference

由于载波频偏移的存在,使得OFDM调制信号发生衰减,将这种衰减作为信道的传函,建立无线信道模型,在此模型的基础上分析用神经网络来抑止载波干扰。在应用神经网络抑止干扰时,需要考虑样本值和收敛算法的选取,提出了一种根据载波间相互干扰的程度来选取样本的方法,以拟牛顿优化算法作为收敛算法。仿真结果表明,该方法可获得更低的误码率。

Because of carrier frequency offset, OFDM modulating signal was faded. A wireless channel module was built with the channel translation function which adopted the fading as parameter. Then neural network was analysed to suppress the inter carrier interference（ICI） . There were two things should be considered ： sample and algorithm. A new way used to select samples according the interference degree was presented. A Quasi-Newton optimization algorithm was applied to it. The simulation result indicated the way has lower BER than other ways .