For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
New Neural Network Based Nonlinear and Multipath Distortion Equalizer for FTTA Systems
Jun IDO Minoru OKADA Shozo KOMAKI
IEICE TRANSACTIONS on Communications
Publication Date: 1997/08/25
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Issue on Mobile Computing)
digital communication, nonlinear distortion, multipath, equalizer, neural network, QAM,
Full Text: PDF>>
A new Neural Network Equalizer (NNE), employing multilayer feedforward neural network, is proposed as a compensation method for nonlinear and multipath distortion that arises from FTTA (Fiber To The Air) system. If a signal in a channel is affected by nonlinear distortion, the conventional Decision Feedback Equalizer (DFE) finds difficulty in perfect compensation of it. To compensate for nonlinear distortion as well as multipath distortion, an equalizer, employing neural network, is investigated. A new neural network equalizer, yielding a cubic function as unit output function, is proposed in order to compensate the nonlinear distortion effectively. We also propose an initial weights of neural network for preventing from local minimum. Computer simulation results show that the compensation performance of the new NNE is superior to the conventional DFE and the conventional NNE.