Bachelor and Master Theses
To apply for conducting this thesis, please contact the thesis supervisor(s).
| Title: | Reinforcement Learning based topology control for SDN-IoT systems |
| Subject: | Computer network engineering, Computer science |
| Level: | Advanced |
| Description: |
Low-power and Lossy Networks (LLN) are composed of a mesh of constrained devices that run a stack of communication protocols that are designed to minimise power consumption. Routing Protocol for Low-power and lossy networks (RPL) as the de-facto standard for routing in this type of networks is designed to support static networks so no wonder it has a poor performance in case topology of the network keeps changing. However, frequent evaluation of the topology consumes a big part of the battery-operated devices. Learning the more frequent handovers can help optimize this trade-off. There are some recent efforts that consider offloading the control processes to a centralised entity. Generally, the nodes find the initial routes using the traditional RPL but the SDN controller has the ability to install routes in the nodes which have a higher probability. In scenarios that the mobile nodes move randomly, it is harder to learn a mobility pattern. Using reinforcement learning the controller can install routes to ensure an specific throughput minimizing the power consumption due to handovers. Supervisor(s): Iliar Rabet Examiner: |
| Start date: | |
| End date: | |
| Prerequisites: | |
| IDT supervisors: | |
| Examiner: | |
| Comments: | |
| Company contact: |