Efficient Learning-based Parameter Configuration of Cellular Networks

Network parameter configuration is crucial for optimizing performance in a cellular network. Often, such parameters are too numerous and their interdependence too complicated for them to be efficiently configured by human experts. Therefore, it is of great interest to study network parameter configuration as a machine learning problem. We aim to expand on the machine learning approaches in recent work from industry leaders such as Huawei. Recent solutions make use of “transfer learning”, that is, sharing information learned between cells in the cellular network to improve the overall performance. In this project, we will consider additional constraints for each cell that must be satisfied and may conflict with the primary goal of network performance optimization.

Faculty Supervisor:

Ning Lu

Student:

Partner:

Pennsylvania State University

Discipline:

Computer science

Sector:

Education

University:

Queen's University

Program: