Adaptive IIR-based NEXT Canceller Using Modified PSO Algorithm

The multiple wireline communication system suffers some issues from echo and crosstalks. This work discusses the filter architectures of the near end crosstalk canceller and simulation that based on the IEEE 802.3bz™-2016 specification standard [1]. Near-end crosstalk (NEXT) can be suppressed in twisted-pair cable, but it has influence between different twisted-cable pairs. In high-speed circuits, there is a higher requirement for signal quality, so the near-end crosstalk problem needs to be solved. Generally, the adaptive filter used in the near-end crosstalk canceller is an adaptive finite impulse response filter (FIR filter), which has the characteristics of simple structure and easy algorithm. However, as the NEXT channel response becomes longer due to the higher data rate, the order of the finite impulse response filter increases in order to match it, which increases the cost significantly. In this work, the adaptive filter will be implemented in both of FIR and the IIR filter with suitable algorithms to simulate the NEXT channel in different lengths to confirm that the operation speed and function meet the specifications of IEEE 802.3bz™-2016 standard. To implement FIR filter, the LMS algorithm is a well-known method to mimic the NEXT channel impulse response, and the adaptive infinite impulse response filter adopts an improved particle swarm optimization algorithm to overcome the problem of multiple local minima in the error plane, to converge to the global minimum value and make the filter performance meet the specifications. Comparing the results, the conclusion shows that the infinite impulse response filter could be employed to replace the finite impulse response filter due to its impulse response properties that the infinite impulse response filter can model channels with fewer orders. [2] Keywords - NEXT canceller, FIR filter, IIR filter, least mean square (LMS), particle swarm optimization (PSO), Ethernet