Enhancing Quality of Experience of 5G Users Exploiting Deep Q-Learning

The Fifth Generation (5G) network aims to redesign the network service architecture so that it can offer an excellent Quality-of-Experience (QoE) to the users. However, the exponentially increasing user demands and large volume of required media services often make the 5G networks congested. Existing works in the literature are limited by focusing either on reducing congestion of the network or maintaining Quality of Service (QoS) expected by the users. However, in this paper, we emphasized on increasing the Quality of Experience (QoE) of the users by selecting an optimal network topology using Deep Q-Learning method, namely, OT-DQL system. The proposed OTDQL system exploits Reinforcement Learning (RL) technique to learn the existing organization of the network connectivity as well as the offered service quality to the users. Thereafter, it predicts the optimal topology for the network that can excel user service experiences and provision the resources accordingly. The performance analysis has been carried out in mininet environment and the experimental results depict significant performance improvement in terms of enhancing the users Quality of Experience (QoE) while reducing the percentage of Service Level Agreement (SLA) violation.