基于神经网络合并技术的Rake接收机性能仿真

Simulation on Performance of Rake Receiver with Neural Network-Based Combiner

传统Rake接收机的线性合并技术不能适应无线信道的非线性特性.针对这一问题,开展了基于神经网络合并技术(NN)的研究.采用多层感知机神经网络模型,通过仿真实验确定了模型结构.在不同的仿真环境下,分别对采用等增益合并技术、最大比合并技术和神经网络合并技术的Rake接收机进行了性能仿真.结果表明:在理想信道下,基于NN技术的Rake接收机性能略好于传统的Rake接收机,性能改善不明显;在复杂信道环境下,基于NN技术的Rake接收机性能好于传统的Rake接收机.

Linear combination technique for conventional Rake receiver is inadaptable to the nonlinear characteristics of wireless channel. So, the NN （neural network ）-based combination technique is studied, which introduces the NN model with multilayer perceptron to determine model structure through simulation, Under different simulation conditions, the performance of Rake receiver using the equal gain, maximum ratio or NN-based combination technique was simulated separately. Simulation results indicated that the performance of Rake receiver with NN- based combiner is slightly better than that of conventional Rake receiver in ideal channel but obviously better in complicated channel environment.