Channel Agnostic End-to-End Learning Based Communication Systems with Conditional GAN

Abstract

In this article, we use deep neural networks (DNNs) to develop a wireless end-to-end communication system, in which DNNs are employed for all signal-related functionalities, such as encoding, decoding, modulation, and equalization. However, accurate instantaneous channel transfer function, i.e., the channel state information (CSI), is necessary to compute the gradient of the DNN representing. In many communication systems, the channel transfer function is hard to obtain in advance and varies with time and location. In this article, this constraint is released by developing a channel agnostic end-to-end system that does not rely on any prior information about the channel. We use a conditional generative adversarial net (GAN) to represent the channel effects, where the encoded signal of the transmitter will serve as the conditioning information. In addition, in order to deal with the time-varying channel, the received signal corresponding to the pilot data can also be added as a part of the conditioning information. From the simulation results, the proposed method is effective on additive white Gaussian noise (AWGN) and Rayleigh fading channels, which opens a new door for building data-driven communication systems.

Conditional Mutual Information Neural Estimator

“…Learning the end-to-end communication system is challenging though, and requires the knowledge of the channel model, which it is not available in many applications. One solution to this problem is to train a GAN model to mimic the channel, which can later be used to learn the whole system [13,14]. Alternatively, one can design encoders and optimize them to achieve the maximum rate (capacity), which is characterized by the MI.…”

Deep Energy Autoencoder for Noncoherent Multicarrier MU-SIMO Systems

“…For example, in actual systems, the channel model and statistics as described in (2) may be completely unknown, and this obviously hinders the channel gradient computation to update the transmitter. To overcome this issue, reinforcement learning [25] or generative adversarial networks [26], which has recently been used to learn the channel model of end-to-end learning-based communication systems, can be applied to NC-EA. Yet, such extensions are far beyond the scope of this work and will be part of our future work.…”

“…The AE concept was also applied to OFDM and noncoherent MU-SIMO systems in [23] and [24], respectively. Some end-to-end AE-based schemes under unknown channel models were proposed in [25], [26], which aim to eliminate the need of a differentiable channel model. Note that under fading channels, these learning-based schemes have to employ pilot transmissions to estimate the CSI for signal detection.…”

Machine Learning Tips and Tricks for Power Line Communications

“…For the same objective, [56] and [57] utilized stochastic perturbation and policy gradients techniques, respectively. Generative models can be used to generate samples of a given communication channel, in particular GANs have been exploited in [50], [58]. • Link and Medium Access Control Layer: Spectrum sensing exploiting GANs and resource management for LTE were studied in [59] and [60], respectively.…”