Adaptive Digital Pre-distortion for Multiple Antenna Transmitters

This paper proposes adaptive algorithms to implement a cross-talk canceling pre-distorter (CTC-DPD), which accurately models the cross-talk and nonlinearity in multipleinput multiple-output (MIMO) radio systems, and compensates for these effects. The cross-talk introduced before the power amplifier (PA) in a multi-antenna transmitter, which is termed as non-linear cross-talk, degrades the performance of digital pre-distortion (DPD). A cross-over pre-distortion (CO-DPD) that compensates for the cross-talk and non-linearity in multi-antenna transmitters was proposed in [2]. However, CO-DPD ignores the higher order cross-terms between the input signals. As a result, the performance of CO-DPD degrades, as the crosstalk between the branches of the multi-antenna transmitter becomes stronger. A CTC-DPD structure which out-performs CO-DPD especially at high cross-talk levels, was proposed in [5]. In this paper, an indirect learning architecture for CTC-DPD with Recursive least square (RLS) pre-distortion and cross-talk cancellation, and three methods for cross-talk estimation are proposed. From simulations using a memory polynomial model of order 7 and memory length 3 for the PAs, it is found that CTC-DPD achieves a normalised mean square error (NMSE) less than -70 dB for cross-talk levels up to -15 dB, for a 2 antenna MIMO transmitter. This performance is similar to that of conventional DPDs in the case with no cross-talk.