Intelligent Resource Allocation for Edge-IoT: Enhancing QoE with Deep Reinforcement Learning
B. Sreedhar, C. Sasikala, V T Ram Pavan Kumar M, Ruhisulthana Shaik, Syed Muqthadar Ali, Yadaiah Balagoni
Abstract
A resource allocation framework zeroed in on Quality of Experience (QoE) that is explicitly intended for the heterogeneous edge Internet of things (IoT) environment's dynamic and fluctuated situation. Applications for the Internet of Things (IoT) are confronting edge figures. Be that as it may, because of the heterogeneity of usages, the edge cloud finds it trying to relegate complex compelled resources (handling, memory, limit, information transmission, etc) with limitations in light of the Quality of Service (QoS) necessities of clients. In this exploration, we tackle the issue of resource allocation in Edge-IoT structures by supporting a canny framework called Deep Edge that assigns resources to the different IoT applications with the sole motivation behind working on the Quality of Experience (QoE) of the clients. To accomplish this objective, we create a complex QoE model that considers adjusting the different prerequisites of Internet of Things applications to the accessible edge resources. The arrangement is accomplished by choosing a QoS necessity arrive at that the accessible resources can meet. Moreover, we propose a brilliant two-stage deep reinforcement learning (DRL) move toward that successfully allots edge resources to help IoT applications and further develop QoE for people. Not at all like the famous DRL, our arrangement utilizes deep neural networks (DNN) to design the Edge-IoT state to joint resource allocation movement, which incorporates resource allocation and QoS class. Along these lines, DNN can be used to deal with exercises' exploration. The cooperative exertion not just builds the quality of experience (QoE) of clients and fulfills the prerequisites of assorted applications, yet it likewise adjusts the QoS necessities to the accessible resources.