Deep-Reinforcement-Learning-Based Sustainable Energy Distribution for Wireless Communication

Abstract

Many countries and organizations have proposed smart city projects to address the exponential growth of the population by promoting and developing a new paradigm for maximizing electricity demand in cities. Since Internet of Things (IoT)-based systems are extensively used in smart cities where huge amounts of data are generated and distributed, it could be challenging to directly capture data from a composite environment and to offer precise control behavior in response. Proper scheduling of numerous energy devices to meet the need of users is a demand of the smart city. Deep reinforcement learning (DRL) is an emerging methodology that can yield successful control behavior for time-variant dynamic systems. This article proposes an efficient DRL-based energy scheduling approach that can effectively distribute the energy devices based on consumption and users' demand. First, a deep neural network classifies the energy devices currently available in a framework. The DRL then efficiently schedules the devices. Edge-cloud-coordinated DRL is shown to reduce the delay and cost of smart grid energy distribution.