A fast start-up RBF equalizer with a useful nonlinear function and its evaluation

✍ Scribed by Motoharu Miyake; Kunio Oishi; Shoichiro Yamaguchi

• Book ID: 102661833
• Publisher: John Wiley and Sons
• Year: 2000
• Tongue: English
• Weight: 565 KB
• Volume: 83
• Category: Article
• ISSN: 1042-0967
• DOI: 10.1002/(sici)1520-6440(200004)83:4<9::aid-ecjc2>3.0.co;2-2

✦ Synopsis

In this paper, a nonlinear function is proposed that is effective for RBF expansion based on hardware realization by a digital signal processor. The RBF equalizer that uses the proposed function consists of four arithmetic operations, which are mainly product-sum operations. Therefore, compared to a Gaussian RBF automatic equalizer, the processor time needed for estimating the transmitting signal can be reduced substantially, and the operation can be accelerated. Also, it is determined from a comparison of the error probability by Monte Carlo simulation that the decision boundary of the RBF automatic equalizer using the proposed function can accurately approximate the Bayesian decision boundary. When the Bayesian decision boundary is accurately approximated, a bit error rate equivalent to that of the Bayesian equalizer can be attained. A computer simulation shows that the RBF automatic equalizer using the proposed function can realize a bit error rate equivalent to that of a Bayesian equalizer and that the convergence is faster than the transversal automatic equalizer and DFE. Further, an RBF equalizer using the proposed function can realize a smaller bit error rate than the others.