Closed-loop digital pre-distortion with memory effects using genetic algorithms

This paper continues our investigation of genetic pre-distortion algorithms to linearize power amplifiers (PA) with strong memory effects. In previous work, we reported simulation results of an adaptive algorithm that requires only a measure of out-of-band emission to sequentially adapt polynomial functions. Compared to traditional algorithms that require wideband feedback, the proposed algorithm is implemented using only narrowband feedback, affording a large savings in analog to digital converters and high dynamic range downconverter components. The paper describes the hardware implementation of this algorithm on a typical basestation configuration. Hardware results indicate substantial reduction in adjacent channel interference when operating on RF power amplifiers with strong memory effects. The fast convergence rate suggests that this technique can be employed as a low-cost solution to meet regulatory requirements.