Analysis Of Microwave Nonlinear Scattering Function And Large-signal Modeling Based On Wavelet Neural Network

Along with the constant progress of modern communication, the design of communication system tends to apply larger power level and more sophisticated modulating program. The nonlinear characteristic of these microwave devices with large signal has become more obviously. And also these power amplifiers always work in the non-linear area, which deteriorates the performance of microwave system. Therefore, nonlinear modeling of large signals has become the main tough topic in design of microwave devices.In this research, a new nonlinear characterization method of microwave - nonlinear scattering function was studied, which can describe the character of large signals nonlinear network in essence. The essence of nonlinear system lies in generating harmonics. And nonlinear scattering function is defined based on this essence. So, nonlinear scattering function can describe the nonlinear characteristic of the large-signal modeling. Nonlinear scattering function is determined merely by the input signals of nonlinear network. So we can build the large-signal blackbox models in nonlinear large-signal networks based on this function.Whether nonlinear scattering function can accurately characterize nonlinear behavior of blackbox model is the key to choose the fitting technology. In this research, the Neural Network was studied. Neural Network as a nonlinear system in essence, can approximate any nonlinear function with any accuracy in theory. Wavelet Neural Network (WNN) has many advantages, such as excellent approximation ability and fault-tolerant capability, with the advantages of both neural network and wavelet theory.Wavelet neural network was used for modeling and analyzing in large signals model in our research. The advantages of this modeling method are as follow: the model parameters can be obtained through measuring large signal; the model can describe the nonlinear character exactly by means of studying the sample data, ignoring the internal components and structures.The nonlinear scattering function of a microwave FET was measured and the large-signal model was built in our research. Accurate data was obtained and the modeling theory was proved accurate by the proposed method.