Deep Joint Source-Channel Coding for Wireless Image Transmission

• Published in: IEEE transactions on cognitive communications and networking Vol. 5; no. 3; pp. 567 - 579
• Main Authors: Bourtsoulatze, Eirina, Burth Kurka, David, Gunduz, Deniz
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.09.2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Artificial neural networks; Bandwidths; Channel coding; Coders; Codes; Communication channels; Decoding; deep neural networks; Digital transmission; Error correction; Fading channels; Image coding; image communications; Image compression; Image transmission; Joint source-channel coding; Noise; Performance degradation; Random noise; Signal to noise ratio; Transform coding; Wireless communication
• ISSN: 2332-7731; 2332-7731
• DOI: 10.1109/TCCN.2019.2919300

We propose a joint source and channel coding (JSCC) technique for wireless image transmission that does not rely on explicit codes for either compression or error correction; instead, it directly maps the image pixel values to the complex-valued channel input symbols. We parameterize the encoder and decoder functions by two convolutional neural networks (CNNs), which are trained jointly, and can be considered as an autoencoder with a non-trainable layer in the middle that represents the noisy communication channel. Our results show that the proposed deep JSCC scheme outperforms digital transmission concatenating JPEG or JPEG2000 compression with a capacity achieving channel code at low signal-to-noise ratio (SNR) and channel bandwidth values in the presence of additive white Gaussian noise (AWGN). More strikingly, deep JSCC does not suffer from the "cliff effect," and it provides a graceful performance degradation as the channel SNR varies with respect to the SNR value assumed during training. In the case of a slow Rayleigh fading channel, deep JSCC learns noise resilient coded representations and significantly outperforms separation-based digital communication at all SNR and channel bandwidth values.