Channel estimation using an adaptive neuro fuzzy inference system in the OFDM-IDMA system

Authors : Necmi TAŞPINAR, Şakir ŞİMŞİR

Pages : 352-364

View : 21 | Download : 8

Publication Date : 0000-00-00

Article Type : Research Paper

Abstract :In this paper, a channel estimator based on an adaptive neuro fuzzy inference system insert ignore into journalissuearticles values(ANFIS); is proposed for the purpose of estimating channel frequency responses in orthogonal frequency division multiplexing-interleave division multiple access insert ignore into journalissuearticles values(OFDM-IDMA); systems. To see the performance of our proposed channel estimation method, five different techniques including well-known pilot-based estimation algorithms such as least squares insert ignore into journalissuearticles values(LS); and minimum mean square error insert ignore into journalissuearticles values(MMSE); with other heuristic methods like multilayered perceptron insert ignore into journalissuearticles values(MLP); trained by a backpropagation insert ignore into journalissuearticles values(BP); algorithm insert ignore into journalissuearticles values(MLP-BP);, MLP trained by the Levenberg--Marquardt insert ignore into journalissuearticles values(LM); algorithm insert ignore into journalissuearticles values(MLP-LM);, and radial basis function neural network insert ignore into journalissuearticles values(RBFNN); are compared with our proposed method by computer simulations. The comparisons are made with the aid of bit error rate and mean square error graphs. According to the simulation results, our proposed channel estimator based on ANFIS shows better performance than both the LS algorithm and the other considered heuristic methods like MLP-BP, MLP-LM, and RBFNN, whereas the MMSE algorithm still shows the best performance as expected because of exploiting channel statistics and noise information, which makes it very complex to be used in any system. As well as being less complex compared to the MMSE algorithm, the estimator based on ANFIS does not need pilot tones for channel estimation. These properties bring our proposed method to an advantageous position among the other estimation techniques.
Keywords : Channel estimation, adaptive neuro fuzzy inference system, multilayered perceptron, radial basis function neural network, orthogonal frequency division multiplexing interleave division multiple access