Deep Learning-based Frame and Timing Synchronization for End-to-End Communications

End-to-end learning-based communications systems are more likely to achieve the global optimal performance. In this work, we implement a complete communications system as an end-to-end deep neural network, including transmitter, channel model, synchronization and receiver. There are a lot of issues which can result to the out of synchronization between the transmitter and the receiver, including sampling timing offset, sampling frequency offset and so on. Therefore, we focus on frame and timing synchronization in the end-to-end learning-based communications systems. Our results show that such an end-to-end learning-based communications system is robust to the impairments that arise from sampling frequency offset. An extra synchronization model based on the convolutional neural networks is introduced to achieve frame synchronization and compensate for the impairments which arise from sampling time offset and sampling time error. The performances of the synchronization model are evaluated from the symbol error rate and the frame header detection results. It can correctly detect most of the actual position of frame header, and it is about 2dB better than that of the direct correlation detection in the end-to-end communications systems.

[1]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[2]  Timothy J. O'Shea,et al.  Radio transformer networks: Attention models for learning to synchronize in wireless systems , 2016, 2016 50th Asilomar Conference on Signals, Systems and Computers.

[3]  Stephan ten Brink,et al.  Deep Learning Based Communication Over the Air , 2017, IEEE Journal of Selected Topics in Signal Processing.

[4]  Ali Mousavi,et al.  Learning to invert: Signal recovery via Deep Convolutional Networks , 2017, 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[5]  Kiran Karra,et al.  Learning to communicate: Channel auto-encoders, domain specific regularizers, and attention , 2016, 2016 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT).

[6]  Philip Schniter,et al.  Onsager-corrected deep learning for sparse linear inverse problems , 2016, 2016 IEEE Global Conference on Signal and Information Processing (GlobalSIP).

[7]  Jakob Hoydis,et al.  An Introduction to Deep Learning for the Physical Layer , 2017, IEEE Transactions on Cognitive Communications and Networking.

[8]  Andrea J. Goldsmith,et al.  Detection Algorithms for Communication Systems Using Deep Learning , 2017, ArXiv.