A combined neural network and fuzzy systems based adaptive digital predistortion for RF power amplifier linearization

Linearization of nonlinear RF power amplifiers (PA) is an important issue when spectrally efficient modulation signals are used in mobile communications. Adaptive digital predistortion (ADP) is one promising linearization technique that can be employed. This paper presents a combined neural network and fuzzy systems based ADP for RF PA linearization. This hybrid approach employed is called adaptive neuro-fuzzy inference system (ANFIS). It has advantages of real-time processing of PA signals, offline adaptation, good IMD suppression, no convergence problems and flexible tuning capability. Experimental results show that a distorted WCDMA signal has been improved by around 12 dB, with noise floor well below -60 dBm. The adaptability of this ANFIS approach to instantaneous variation in PA response through time is also demonstrated, and results show that this ANFIS approach is capable of adapting to simulated environmental changes. Further testing also demonstrated that the tuning parameters involved in the training could be reduced by more than half for a fairly nonlinear PA without significantly degrading the distortion suppression capability, thus a tremendous amount of DSP processing burden can be saved.