In-band full-duplex (IBFD) is an attractive technology in telecommunications since it can improve spectral efficiency and data rates. However, IBFD is challenging since its implementation requires a complex architecture and an advanced algorithm for echo (self-interference, SI) cancellation. In addition, IBFD deployment is especially challenging in broadband power line communication (BB-PLC) due to channel frequency selectivity caused by multipath propagation, impedance mismatches, and time-variant behavior. Such PLC channel conditions require the ability of the system to track channels and adapt to their changes. In this paper, we propose a pilot-free channel tracking algorithm for IBFD 2-by-2 multiple-input multiple-output (MIMO) BB-PLC. We demonstrate the ability to obtain initial estimates of the channels and SI responses using a half-duplex training phase. Then, we use a data decision-driven approach for channel tracking. The tracking ability of the system is evaluated using the average error vector magnitude (AEVM), a metric that characterizes the equalized constellation quality at the receiver. Our data-driven approach shows good tracking ability with 75% of the subcarriers being updated at each data symbol.
