SmartMove:Context-Aware Deep Reinforcement Learning for Service Discovery in Wireless Networks

New generation of Wireless technologies such as 5G, 6G and beyond 6G provide much better services in terms of throughput, network coverage or latency. These latest technologies expected to benefit or change many businesses involving Augmented Reality (AR), Virtual Reality (VR), smart cities, healthcare etc. In reality, voice and data services continue to be the critical requirements for the end user. However, in day-to-day usage, users continue to experience low/no coverage in certain areas more specifically inside buildings due to obstacles such as walls that block the signals. At the same time, there would exist some nearby locations that provide better services. Our proposal provides an opportunity for the user to explore such locations when requiring better service referred to as Service Discovery. This paper develops an approach for predicting the nearest location around the user which can provide better service using Reinforcement Learning (RL) model based on Deep Q-Network (DQN). This model is trained with features such as reception signals, time, and information of obstacles in different locations. Predicted signal conditions in the new location is used in an application guiding the user to a location which help provide better services.