Information Distribution System with Distributed Reinforcement Learning for Providing Local Information

In this paper, we propose an information distribution system with reinforcement learning so that the information can be distributed to users with a small number of distributions. In this system, the information distributor moves along the same route and distributes the information to users by using the wireless communication technology. Here, we consider a case where the distributor would like to send the information such as advertisement to users who pass a specified area related to the information. As long as the information is distributed at random, however, it is hard to distribute the information to users that pass the specified area effectively. To achieve this goal, in our proposed system, distributed reinforcement learning is utilized. With the distributed reinforcement learning, the distributor learns the optimal positions where the information can be distributed to a larger number of users who pass the specified area. Moreover, by considering road conditions, the information can be distributed to users effectively. We evaluate the performance of our proposed system with simulation. In numerical examples, we investigate the effectiveness of our proposed system by comparing with the conventional method where the information is distributed at random.