Adaptive predistortion lineariser using polynomial functions

Linear modulation methods are highly sensitive to high power amplifier nonlinearities. An adaptive predistortion lineariser using polynomial functions is described. An important problem faced in optimisation of nonlinear circuits such as predistortion linearisers is that of convergence into a local minimum. This results from the nonquadratic shape of the objective functions. A solution to this problem is presented using a postdistorter with similar polynomial functions as the predistorter. A sample of the signals at the power amplifier output and predistorter input are demodulated. These demodulated signals are used for estimation of the postdistorter polynomial coefficients. The objective functions used in the estimation are quadratic functions of the coefficients being estimated resulting in a rapid convergence to the global minimum. The coefficients of the predistorter polynomials are set from those of the postdistorter. Computer simulation results for the proposed lineariser are presented. These results show 50 dB spectrum spreading improvement. Further, unlike previously reported linearisers, the proposed lineariser is insensitive to the demodulator gain and phase imperfections. The performance of the proposed lineariser structured with fifth order polynomial functions is also compared with that for earlier polynomial type linearisers