Machine Learning-based Approaches for Channel Estimation and Spectrum Sharing

Using simplified language understandable to a layperson; provide a general, one-paragraph description of the proposed research project to be undertaken by the intern(s) as well as the expected benefit to the partner organization. (100–150 words)
The proposed research is focused on developing machine learning and deep learning-based approaches to improve the operation of 5G networks in two main sub-project areas: channel reconstruction and spectrum sharing. In the first sub-project, we propose a methodology to reconstruct the wireless channel based solely on limited downlink CSI metrics. The purpose of this study is to recreate channels experienced during over-the-air testing by extracting their key channel characteristics. These characteristics will be used to create a similar simulated channel for further testing. This will improve the simulated channel’s accuracy, making it resemble the real-world scenario in which the CSI was collected. In the second sub-project, i.e., spectrum sharing, we propose the use of different Deep Reinforcement Learning (DRL) algorithms to develop novel and innovative approaches for optimal and distributed spectrum sharing and optimizing the operation of RIS-assisted UAV networks and RIS-assisted m-MIMO networks for 5GB systems.
The outcomes of the project will include models for channel reconstruction as well as new approaches for spectrum sharing. The impact of this project is an increase in the data rates and significant reduction in delays over 5G networks, which will improve the services provided to end users and clients over such networks.

Faculty Supervisor:

Gabriel Wainer

Student:

Partner:

Ericsson Canada Inc (Ottawa, ON)

Discipline:

Engineering

Sector:

Information and cultural industries; Manufacturing; Professional, scientific and technical services

University:

Carleton University

Program: