Enhancing the Beam Alignment in 6G Networks using Deep Learning

Abstract: The emergence of sixth generation (6G) wireless communication networks requires a huge amount of data rate, low latency, massive connectivity, and high reliability in communication. The use of millimeter-wave (mmWave) communication is seen as one of the key technologies that will satisfy these requirements because of the availability of a wide bandwidth in the mmWave bands. In contrast, however, the path loss for mmWave systems is very high, signals can be blocked by objects and there is a high overhead of beam alignment that makes access to the systems difficult in dynamic wireless environments. Traditional exhaustive beam sweeping techniques are very time consuming and complex since all the beam directions have to be swept prior to establishing communication. To overcome these challenges, a Deep Learning-based Initial Access (DeepIA) framework for fast and reliable beam alignment in AI-powered 6G mmWave Networks is introduced in this paper. The proposed approach is based on a Deep Neural Network (DNN) designed to predict the best beam direction based on Received Signal Strength (RSS) measurements instead of doing a beam sweeping. A novel beam selection method called Sequential Feature Selection (SFS) is used to select the most informative beam combinations to achieve accuracy in prediction while minimizing beam sweeping delay. Moreover, to further improve the system performance under Non-Line-of-Sight (NLoS) channels, a technique called RSS averaging is introduced as an approach to reduce the fluctuation and shadow fading effects of the channel. The simulation results show that the proposed DeepIA framework can accurately predict the beams with a very small number of beam sweeps, which can significantly shorten the delay of initial access and enhance the efficiency of communication. The proposed approach is scalable and intelligent that can be used in the future 6G mmWave communication systems with the use of AI.

Keywords: Deep Learning, 6G, mmWave Communication, Beam Alignment, Beamforming, Initial Access, Deep Neural Network (DNN), Artificial Intelligence (AI)