| Proj No. | A2158-251 |
| Title | Reconstructing physical networks with reinforcement learning |
| Summary | This project addresses the challenging problem of network tomography: reconstructing the structure of a physical network from limited path data. Given a network with N observable end-point nodes, we are provided with the cumulative path lengths and hop counts for all simple paths between end-points. The objective of this project is to develop a Q-learning agent capable of inferring the network's topology and edge weights. This involves designing a suitable state representation, action space, and reward function to guide the agent's exploration and optimization of network configurations. The agent will iteratively refine its network model based on the provided path data, learning to accurately represent the underlying network structure. The project requires a rigorous implementation of the Q-learning algorithm and a thorough evaluation of its performance through comparison with known test networks. This research contributes to the development of efficient and robust network reconstruction techniques, with applications in areas such as infrastructure monitoring and sensor network analysis. Students will gain practical experience in reinforcement learning, algorithm design, and network analysis, developing skills applicable to a wide range of engineering and computer science fields. |
| Supervisor | Ast/P Matthew Foreman (Loc:S1 > S1 B1C > S1 B1C 77, Ext: ?) |
| Co-Supervisor | - |
| RI Co-Supervisor | - |
| Lab | Computer Engineering I (Loc: S2-B4c-15) |
| Single/Group: | Single |
| Area: | Wireless and Communications Engineering |
| ISP/RI/SMP/SCP?: |