Cascaded Models for Cancellation of Air-Induced Passive Intermodulation in MIMO Transceivers

This paper describes and proposes cascaded models for the identification and cancellation of air-induced passive intermodulation (PIM) in frequency division duplex (FDD) multiple-input multiple-output (MIMO) systems. The air-induced PIM is interference which stems from nearby metallic objects outside the transceiver system. Previous works on the topic have been targeting a limited subset of the possible scenarios present in such systems, namely 2x2 MIMO operation with two parallel streams and two component carriers (CCs), totaling four distinct signals. We demonstrate that the recently proposed cascaded model utilizing spline-interpolated lookup tables for nonlinearity modeling is capable of handling the aforementioned case with low computational cost, even when channels with memory effects are considered. Furthermore, we expand the 2x2 case to include more transmit streams and CCs and develop a multi-input Hammerstein-type model with polynomial basis functions and a global filter, and accompany it with a low-complexity gradient descent-based parameter adaptation to cancel PIM in such more evolved scenarios. We evaluate the performance of the developed canceller numerically, where favorable cancellation results are shown.