Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19]

@

Journal of Information Science and Engineering, Vol. 22 No. 5, pp. 1163-1176 (September 2006)

A New ISF-FLANN Channel Equalizer for 4QAM Digital Communication Systems and Its FPGA Verification

Wan-De Weng, Chin-Tsu Yen and Rui-Chang Lin
Graduate School of Engineering Science and Technology
National Yunlin University of Science and Technology
Touliu, Yunlin, 640 Taiwan

A new improved soft-feedback functional link artificial neural-network (ISFFLANN) based nonlinear channel equalizer is proposed in this paper. By using the functional expansion utilities, the ISF-FLANN does not need the hidden layers, which are existed in most of the multilayer perceptron network (MLP)-based equalizers. So the ISF-FLANN exhibits much simpler structure and thus requires less amount of computation during the training mode. We find that the use of soft feedback can greatly improve the performance of our previous work on FLANN structure [13]. The comparison of the average transmission symbol error rates (SER) of the ISF-FLANN with the linear transversal filters (LTF) and the traditional FLANN based on FPGA verification are presented. Simulation results demonstrate that ISF-FLANN outperforms FLANN by about 2 to 3 dB, and is about 6 to 7 dB better than LTF. The learning curves (LC) show that our design well fits the real-time processing requirement for 4QAM modern digital communication systems.

Keywords: nonlinear channel equalizers, neural network, LTF, MLP, FLANN, ISFFLANN

Full Text () Retrieve PDF document (200609_11.pdf)

Received May 24, 2004; revised September 8, 2004 & March 7, 2005; accepted April 13, 2005.
Communicated by Liang-Gee Chen.