Digital predistortion (DPD) has been widely adopted to keep RF power amplifier operating with high efficiency without losing linearity in the exiting 4G systems. It is expected that DPD will continue to be deployed in 5G systems. However, due to shifting from the single antenna to the multiple-input multiple-output (MIMO) phased array and continuously increased signal bandwidth, system designers face significant challenges in managing power consumption and meeting linearity requirement of wireless transmitters. In this talk, we will discuss how the recently developed machine learning techniques can be utilised to resolve some of the issues in linearizing 5G MIMO systems, including data-clustering assisted DPD for multiple dynamic configurations, model tree-based behavioural model construction and simplified model extraction techniques.
Speaker(s): Prof Anding Zhu,
Virtual: https://events.vtools.ieee.org/m/297808