D.R. Valiullin - Post-graduate Student, Faculty of Physics M.V. Lomonosov Moscow State University E-mail: vdr91@mail.ru P.N. Zakharov - Ph.D. (Phys.-Math.), Associate Professor, Faculty of Physics M.V. Lomonosov Moscow State University E-mail: zakharov1@mail.ru

Equalizers are used in communication systems operating in multipath environment in order to minimize the BER degradation caused by intersymbol interference. Existing equalizers such as DFE (decision feedback equalizer) and MLSE (maximum likelihood sequence estimation) prove to be either computationally complex or having large energy loss compared to Shannon limit. An HNN-equalizer in-troduced in [1] based on Hopfield-Tank neural network was claimed to achieve high energy efficiency with a moderate computational complexity. However, the claimed in [1] high performance is achievable in Rayleigh channels only on average. In a number of channels it demonstrates large energetic loss (10 dB and more) compared to MLSE [2]. The purpose of this paper was to improve the HNN equalizer energy efficiency in such channels. The learning process is used for neural network in HNN equalizer. The coefficients of neurons connections are altered to minimize MSE (mean-square error) with gradient descent. Simulation was used to estimate and compare the BER (bit-error rate) performance for: a) basic HNN equalizer; b) modified HNN equalizer with neural network learning; 3) DFE equalizer; 4) MLSE equalizer. The results of the simulation have shown that after the adaptation process the coefficients converge to the values which slightly differ from the coefficients of basic HNN equalizer. The energy efficiency of modified HNN equalizer was improved by 1 dB compared to the basic HNN equalizer.

 

1. Myburgh H.S., Olivier J.C. Near-optimal low complexity MLSE equalization // IEEE Wireless Communications and Networking Conference. 2008. P. 226-230.
2. Valiullin D.R., Zakharov P.N., Korolev A.F. Iterativnyjj ehkvalajjzer na osnove nejjronnykh setejj dlja mnogoluchevogo radiokanala // Trudy shkoly-seminara «Volny-2016». Sekcija 6. S. 18-21.
3. Prokis Dzh. Cifrovaja svjaz. Per. s angl. M.: Radio i svjaz. 2000.