Integrating Future Smart Home Operation Platform With Demand Side Management via Deep Reinforcement Learning

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

14 Scopus Citations
View graph of relations



Original languageEnglish
Pages (from-to)921-933
Number of pages13
Journal / PublicationIEEE Transactions on Green Communications and Networking
Issue number2
Online published19 Apr 2021
Publication statusPublished - Jun 2021


Residential demand side management (DSM) is apromising technique in smart grids to improve the power systemrobustness and to reduce the energy cost. However, the ongoingparadigm shift of computation, such as mobile edge computingfor smart home, poses a big challenge to residential DSM.Therefore, it is important to schedule the new smart homecomputing tasks and traditional DSM in a smart way. In thispaper, we investigate an integrated home energy managementsystem (HEMS) that participates in a DSM program and implements smart home computation tasks by offloading tasks withthe help of a Smart Home Operation Platform (SHOP). Thegoal of HEMS is to maximize the user’s expected total reward,defined as the reward from completing computing tasks minus thecost of energy consumption, execution delay, running the SHOPservers, and the penalty of violating the DSM requirements. Wesolve this task scheduling based DSM problem using a deepreinforcement learning method. The DSM program consideredin this paper requires the household to reduce a certain amountof energy consumption within a specified time window, which,in stark contrast to the well-studied real-time pricing, resultsin a long-term temporal interdependence and thus a high-dimensional state space in our formulated problem. To addressthis challenge, we use the Deep Deterministic Policy Gradient(DDPG) method to characterize the high-dimensional state spaceand action space, which uses deep neural networks to estimatethe state and to generate the action. Experimental results showthat our proposed method achieves better performance gains overreasonable baselines.

Research Area(s)

  • Batteries, deep reinforcement learning., Demand side management, Edge computing, Energy consumption, Servers, Smart grids, Smart homes, Task analysis, task offloading