In this paper, we proposed a new equalization scheme based on multi-layer perceptrons (MLPs) to mitigate two-dimensional (2D) inter symbol interference due to the nonlinear channel and misalignment. Two MLPs are used in the proposed equalization scheme. One MLP is used for the misalignment estimation and the other MLP is used for equalization. The MLPs learn the channel with misalignment in the training procedure. The MLP-based misalignment estimator estimates the horizontal and vertical misalignments from intensity detected by the charge coupled detector. The MLP-based equalizer considers the estimated misalignments as the input along with intensity, and then equalizes the channel. The simulation results show that the proposed equalization scheme is robust to the channel with misalignment and provides the improved BER performances compared to the conventional scheme using the 2D equalizer filter, one-dimensional generalized partial response target and Viterbi detector.
|Journal||Japanese journal of applied physics|
|Publication status||Published - 2019|
Bibliographical notePublisher Copyright:
© 2019 The Japan Society of Applied Physics.
All Science Journal Classification (ASJC) codes
- Physics and Astronomy(all)