Xingliang Li, Guoqiang Xue
5G wireless networks have introduced new enabling technologies to improve performance and meet the ever-increasing user demand. Harnessing mmWave spectrum bands for usage in mobile systems and employing large-scale or massive MIMO are key enabling technologies for 5G networks. However, the use of large-scale antenna arrays and new frequency bands (24–100GHz) in 5G introduce big challenges to the acquisition of channel state information (CSI) for the employed frequency bands. As such, efficient channel reconstruction schemes are much needed. CSI is not only important for high system performance (in particular for downlink) but can also be utilized in many other aspects. CSI and accurate channel characteristics can be used to build channel models to be used for analyzing system performance such as delay and throughput. They can also be used to build emulators that can be used to test the performance of new products or compare the performance of different algorithms.
In this project, we propose developing a simulation tool to reconstruct temporal-spatial channels based on limited CSI feedback that has been collected from User Equipments (UEs) communicating over real networks. We will use machine learning and deep learning algorithms to develop models to reconstruct channel characteristics from the limited CSI feedback.