Proposal

Cellular networks (CNs) are high-speed, high-capacity, wireless telecommunication networks designed for voice and data. They are of great importance to modern societies since their use cases have gradually shifted from phone calls and entertainment to more critical applications [1]. However, the parameters of a typical CN are difficult to optimize since their optimal values depend on: (i) the CN topology (i.e. the distribution of the nodes in space), (ii) the consumption habits of the CN users and (iii) exogenous factors. For these reasons, the performance of a CN is very difficult to model and can be viewed as a black-box function that varies with time.

Gaussian Process-based Bayesian Optimization (BO) [2] has been very successful at optimizing black-boxes, including those occurring in wireless networks [3, 4]. However, common BO algorithms struggle at optimizing dynamic objective functions (i.e. functions of time) and Dynamic Bayesian Optimization (DBO) is still emerging as a research area.

In this project, the student will apply a new DBO algorithm [5] to the optimization of CN parameters. The resulting work may have a significant impact on the design and the performance of present and next-generation CNs.

Skills

• Excellent coding skills (Python)
• Familiarity with stochastic processes would be a plus.
• Familiarity with Bayesian optimization would be a plus.
• Familiarity with wireless networks would be a plus.

How to Apply

If interested, please send a resume and transcripts at anthony.bardou@epfl.ch.

References

[1] Vacca, J. Network and system security. Elsevier, 2014.

[2] Williams, C., & Rasmussen, C. (1995). Gaussian processes for regression. Advances in neural information processing systems, 8.

[3] Bardou, A., & Begin, T. (2022). INSPIRE: Distributed bayesian optimization for improving spatial reuse in dense WLANs. In Proceedings of the 25th International ACM Conference on Modeling Analysis and Simulation of Wireless and Mobile Systems (pp. 133-142).

[4] Dreifuerst, R. M., Daulton, S., Qian, Y., Varkey, P., Balandat, M., Kasturia, S., ... & Heath, R. W. (2021). Optimizing coverage and capacity in cellular networks using machine learning. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 8138-8142). IEEE.

[5] Bardou, A., Thiran, P., & Ranieri, G. (2024). This Too Shall Pass: Removing Stale Observations in Dynamic Bayesian Optimization. arXiv preprint arXiv:2405.14540.