Journal - Biweekly Meeting #week 9-10 07.-20. Mar 2022
Problem Encountered: I dwelled in building the simulation, didn't realize that I need to read more in order to identified a concrete problem. I can't draw solutions from an obscure problem.
Things Been doing: Read papers extensively in three areas: network building in disaster scenarios, multi-agent reinforcement learning, reinforcement learning in flying ad-hoc networks. Separate different functionalities of different components of the simulation, so that I know exactly what functionalities I want to bring by using multi-agent reinforcement learning. Then formulate a problem.
What I have learned: One background setting, possible locations of routers should be provided information. The drone swarm has two functionalities, one is to bridge communication between isolated router groups, two is to bring Internet to routers from outside base stations. network functionality: form groups of routers, spread Internet connection if exists.
possible next steps: direction 1: use RL in trajectory design upgrade - more realistic: placement of the broken down base station, routers and charing points matches to the layout of Darmstadt? where routers match housing areas, charging points match police stations and ?