Towards Precise Synchronization Under Phase Distortion: Signal Design and Deep Learning

To achieve precise timing accuracy under phase distortion for 6G high-precision services, this paper presents two technological research approaches and proposes corresponding solutions, along with future prospects. The first approach involves signal design, where synchronization signals are crafted based on the distributed concatenation of a binary/complex sequence and its modification. The second approach focuses on deep learning, proposing a Convolutional Neural Network. Considering factors such as timing detection performance, computational complexity, memory usage, and the correlation between the wireless channel during offline training and the one during online application, performance evaluations provide insights into the challenges of deep-learning based approaches. These evaluations suggest that signal-design based approaches will be preferred at least until the advent of 6G. However, the paper also provides deep-learning approaches that aim to overcome these challenges beyond 6G.