Number-Theoretic Net-Based Particle Filtering for Linear Phase Noise Tracking in CO-OFDM Systems

In this paper, we propose a number-theoretic net-based particle filter (NT- PF) to dynamically estimate the linear phase noise (LPN) for CO-OFDM systems in the time domain. Inspired by the concept of uniform design (UD), we recursively draw the particles from the NT -net at each time index, which leads to a significant improvement on the algorithm efficiency compared with the conventional Gaussian particle filter (GPF). In addition, we propose a signal detection approach to estimate the LPN on the entire received signal first, then recover the transmitted signal from the received ones. The dynamic tracking performance, efficiency and robustness of the proposed NT-PF is verified in an 89 Gb/s 16-QAM CO-OFDM system.