ECE Seminar Series: Oluwaseun T. Ajayi
The Department of Electrical and Computer Engineering presents its seminar series featuring guest speaker Oluwaseun T. Ajayi, a Ph.D. candidate at Illinois Tech, who will present “Combinatorial Optimization Learning Approach for Energy-Efficient Backhauling in UAV-Assisted B5G/6G Wireless Networks.” This seminar will take place on Friday, April 3, from 12:45–1:45 p.m. in Siegel Hall, room 118.
Abstract
Beyond 5G (B5G)/6G networks rely on ultra-dense small cells to meet heterogeneous user traffic demands. Unmanned aerial vehicle base stations (UBSs) offer flexible deployment on tall structures, enabling relay-based multi-hop wireless backhauling for ultra-reliable low-latency communications. The main challenge is jointly optimizing UBS deployment and multi-hop backhaul to minimize overall energy consumption, an NP-hard mixed integer linear programming problem. To address this, a Combinatorial Optimization Learning Approach (COLA) is proposed, combining neural networks with combinatorial algorithms. The neural network predicts “magic” link weights from application-level inputs, which are then used in Dijkstra’s algorithm to find energy-efficient shortest paths for multi-commodity flows. COLA decomposes the original problem into learnable single-commodity shortest path subproblems, facilitates training data generation, and enables backpropagation through the combinatorial component. Evaluations show COLA drastically reduces computation time while achieving near-optimal energy efficiency in both small- and large-scale wireless networks.
Biography
Oluwaseun T. Ajayi received his B.S. degree in telecommunication science from the University of Ilorin, Nigeria, in 2018, and his master’s degree in electrical and computer engineering from Illinois Tech in 2024. He is currently pursuing his Ph.D. in electrical engineering at Illinois Tech. His research interests include machine learning, wireless network optimization, information freshness, cloud/edge computing, and artificial intelligence for science.