Aiming at the problems of severe inter-symbol interference and high symbol error rate in short-wave time-varying channels, convolutional recurrent neural network (CRNN), which combines convolutional neural network (CNN) and recurrent neural network (RNN), is used as blind equalization algorithm for the short-wave time-varying channel. A CRNN blind equalizer (CRNNBE) is designed for short-wave time-varying channels (such as Rayleigh flat fading channels and frequency selective fading channels). The blind equalizer is based on the fast convergence speed of CNN and the ease of processing sequence signals by RNN, which overcomes the problem of inter-symbol interference and effectively improves the communication quality. The simulation experiment results show that compared with the blind equalizer based on RNN and CNN, the CRNNBE after training has higher accuracy, lower cross-entropy loss, and the convergence speed is significantly higher than that of the RNN blind equalizer. The model can complete convergence in about 20 times. In the short-wave time-varying channel, as a whole, compared with other equalizers, under the same signal-to-noise ratio, CRNNBE has the lowest symbol error rate and the highest communication reliability.
本文提出基于CRNN的短波时变信道盲均衡器,结合了CNN与RNN的优点,利用神经网络的训练与分类功能解决盲均衡问题。图2为本文所提出的基于CRNN的短波时变信道盲均衡器结构示意图。该均衡器主要由Fold层、卷积层、Unfold层、Flatten层、长短时记忆(long short term memory,LSTM)层、全连接层、Softmax层和分类层构成。
AHMEDS, KHANY, WAHABA. A review on training and blind equalization algorithms for wireless communications[J]. Wireless Personal Communications, 2019, 108(3):1759-1783. DOI: 10.1007/s11277-019-06495-8 .
[2]
LIX H. Blind channel estimation and equalization in wireless sensor networks based on correlations among sensors[J]. IEEE Transactions on Signal Processing, 2005, 53(4):1511-1519. DOI: 10.1109/TSP.2005.843744 .
GODARDD. Self-recovering equalization and carrier tracking in two-dimensional data communication systems[J]. IEEE Transactions on Communications, 1980, 28(11):1867-1875. DOI: 10.1109/TCOM.1980.1094608 .
LIR, ZHANGT, ZHUQ Y. Study on variable step blind equalization algorithm based on CMA[J]. Modern Electronics Technique, 2018, 41(3): 10-13. DOI:10.16652/j.issn.1004-373x.2018.03.003(Ch ).
[7]
BURSEK, YADAVR N, SHRIVASTAVAS C. Channel equalization using neural networks: A review[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 2010, 40(3):352-357. DOI: 10.1109/TSMCC.2009.2038279 .
[8]
GENGY X, ZHANGL Y, SUNY S, et al. Research on ant colony algorithm optimization neural network weights blind equalization algorithm[J]. International Journal of Security and Its Applications, 2016, 10(2):95-104. DOI: 10.14257/ijsia.2016.10.2.09 .
[9]
LAVANIAS, KUMAMB, MATEYP S, et al. Adaptive channel equalization using recurrent neural network under SUI channel model [C]// 2015 International Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS). New York: IEEE Press,2015:1-6. DOI:10.1109/ICIIECS.2015.7193035 .
ZHOUC. Research on digital equalization technology of BP neural network algorithm [J]. Communication Power Technology, 2019, 36(9):50-51. DOI:10.19399/j.cnki.tpt.2019.09.015(Ch ).
[12]
HINTONG E, SALAKHUTDINOVR R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786):504-507. DOI:10.1126/science.1127647 .
GREFFK, SRIVASTAVAR K, KOUTNÍKJ, et al. LSTM: A search space odyssey [J]. IEEE Transactions on Neural Networks and Learning Systems, 2017, 28(10): 2222-2232. DOI:10.1109/TNNLS.2016.2582924 .
[15]
ZHANGA G, ZHUW, LIJ. Spiking echo state convolutional neural network for robust time series classification[J]. IEEE Access, 2019, 7: 4927-4935. DOI:10.1109/ACCESS.2018.2887354 .