Thesis description

The continuously increasing energy consumption in cellular communications is one of the remarkable concerns. 5G with large number of antennas at base stations, wider spectrum, larger number of cells and higher modulation places a significant burden on operators because it could easily double/triple the power requirements at a cell site. There is the potential of optimizing the energy consumption through dynamic cell zooming technique, where the coverage area of base stations can expand and contract as per the traffic volume. This is done by switching-off base stations having low traffic and compensating the coverage loss by expanding the neighboring base stations coverage through increasing transmit power.

The objective of the thesis is to study dynamic cell zooming problems in 5G networks with tools of optimization and machine learning. The student is expected to tackle the problem under the optimization framework (e.g., linear programming, interger programming, quadratic programming). Furthermore, the machine learning can be applied to simplify the solutions. The solutions will be implemented and evaluated (by Matlab or C). The work involves mathematical formulation, code implementations and results analysis.

Qualifications

• Master student in Wireless Communications, Electrical Engineering or equivalent.
• A solid theoretical background in areas such as signal processing or linear algebra.
• Experience in modeling and simulation.

• Knowledge of LTE/5G principles, e.g. MIMO, radio channel.

• Knowledge of statistis, machine learning.

Contact person

Stavroula Vassaki

stavroula.vassaki@huawei.com