Fast prediction of the performance of wireless links by simulation-trained neural networks

The paper illustrates a technique for fast and accurate prediction of the performance of wireless links while taking into account the nonlinear distortions in the RF/microwave sections. Each simulation is partly carried out by modulation-oriented harmonic balance, and the results are used to generate a specialized neural network which efficiently performs the major part of the task. High accuracy is obtained, and dynamic behavior of wideband modulating signals can easily be included.