Trainable Communication Systems: Concepts and Prototype

被引:112
作者
Cammerer, Sebastian [1 ]
Aoudia, Faycal Ait [2 ]
Doerner, Sebastian [1 ]
Stark, Maximilian [3 ]
Hoydis, Jakob [2 ]
ten Brink, Stephan [1 ]
机构
[1] Univ Stuttgart, Inst Telecommun, D-70659 Stuttgart, Germany
[2] Nokia Bell Labs, F-91620 Nozay, France
[3] Hamburg Univ Technol, Inst Commun, D-21073 Hamburg, Germany
关键词
Receivers; Training; Optical transmitters; Communication systems; Iterative decoding; Optimization; Autoencoder; end-to-end learning; iterative demapping and decoding; code design; geometric shaping; software-defined radio; PARITY-CHECK CODES; BANDWIDTH-EFFICIENT; MODULATION; CAPACITY; DESIGN;
D O I
10.1109/TCOMM.2020.3002915
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
We consider a trainable point-to-point communication system, where both transmitter and receiver are implemented as neural networks (NNs), and demonstrate that training on the bit-wise mutual information (BMI) allows seamless integration with practical bit-metric decoding (BMD) receivers, as well as joint optimization of constellation shaping and labeling. Moreover, we present a fully differentiable neural iterative demapping and decoding (IDD) structure which achieves significant gains on additive white Gaussian noise (AWGN) channels using a standard 802.11n low-density parity-check (LDPC) code. The strength of this approach is that it can be applied to arbitrary channels without any modifications. Going one step further, we show that careful code design can lead to further performance improvements. Lastly, we show the viability of the proposed system through implementation on software-defined radios (SDRs) and training of the end-to-end system on the actual wireless channel. Experimental results reveal that the proposed method enables significant gains compared to conventional techniques.
引用
收藏
页码:5489 / 5503
页数:15
相关论文
共 33 条
[1]   Deep Learning Constellation Design for the AWGN Channel With Additive Radar Interference [J].
Alberge, Florence .
IEEE TRANSACTIONS ON COMMUNICATIONS, 2019, 67 (02) :1413-1423
[2]  
[Anonymous], 2015, ACS SYM SER
[3]  
[Anonymous], 2008, Modern Coding Theory
[4]  
[Anonymous], 2011, IEEE Standard 802.15.7-2011, P1, DOI [DOI 10.1109/IEEESTD.2009.5307322, 10.1109/IEEESTD.2009.5251812, DOI 10.1109/IEEESTD.2009.5251812]
[5]   Model-Free Training of End-to-End Communication Systems [J].
Aoudia, Faycal Ait ;
Hoydis, Jakob .
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, 2019, 37 (11) :2503-2516
[6]   Bandwidth Efficient and Rate-Matched Low-Density Parity-Check Coded Modulation [J].
Boecherer, Georg ;
Steiner, Fabian ;
Schulte, Patrick .
IEEE TRANSACTIONS ON COMMUNICATIONS, 2015, 63 (12) :4651-4665
[7]   Bit-interleaved coded modulation [J].
Caire, G ;
Taricco, G ;
Biglieri, E .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1998, 44 (03) :927-946
[8]   On the design of low-density parity-check codes within 0.0045 dB of the Shannon limit [J].
Chung, SY ;
Forney, GD ;
Richardson, TJ ;
Urbanke, R .
IEEE COMMUNICATIONS LETTERS, 2001, 5 (02) :58-60
[9]  
Cover TM, 2012, Elements of information theory
[10]  
diaeresis>ocherer G. B<spacing, 2017, ARXIV170701134