Deep learning in wireless communications for physical layer

Current wireless communication faces challenges of spectrum congestion, interference, and accommodating Internet of Things and 5G demands. Artificial intelligence (AI) has recently been considered a powerful technique in many fields due to its excellent learning ability, such as image processing, speech recognition, and computer vision. It has also been applied to wireless communications to design communication modules at the transceivers. Communication transceivers integrated AI can optimize spectrum utilization, enhance interference management, and enable intelligent network adaptation for efficient and reliable wireless communication. This paper introduces deep learning (DL) in wireless communications for the physical layer. We investigate the DL techniques applied to the receiver design, modulation recognition, channel estimation, and signal detection. We mainly focus on the deep neural networks structure of the three communication modules and introduce the benefits of receiver-integrated DL. Lastly, we also conclude the limitation of current communication developments and envision a future where DL-based approaches hold the potential to address the deficiencies of existing wireless communication.