Deep Reinforcement Learning for Next-Generation IoT Networks

in Special Issue   Posted on July 20, 2020 

Information for the Special Issue

Submission Deadline: Sun 15 Nov 2020
Journal Impact Factor : 3.111
Journal Name : Computer Networks
Journal Publisher:
Website for the Special Issue:
Journal & Submission Website:

Special Issue Call for Papers:


Next Generation Internet-of-Things (NG-IoT) brings together 5G and beyond, Artificial Intelligence (AI), cloud-edge computing, virtual reality and augmented reality (VR/AR), and distributed ledger technologies (DLTs). Wireless sensors, actuators, smart devices, control systems, and communication networks form the infrastructure of an IoT that makes smart cities, smart factories, and wearable devices possible. Different advanced communication technologies and protocols can be used for IoT such as Low-power Wireless Personal Area Networks (LoWPAN), Low-power Wide Area Network (LPWAN), ZigBee, Near Field Communication (NFC). NG-IoT systems gather massive amounts of data that contains valuable information for detections, predictions, and decision-making. Thus, it’s important to research into new AI and Machine Learning-based methods to process and analyze such IoT data for achieving reliable and efficient communications in NG-IoT.

As a new research hotspot in AI, Deep Reinforcement Learning (DRL) has emerged as a promising solution for enhancing NG-IoT communications, where DRL agents intend to learn optimal decision-making by interacting with the environment. Through using Deep Neural Networks (DNNs) as powerful function approximators, DRL is able to handle high-dimensional and continuous problems. In recent years, DRL has achieved great success in many fields, such as robot control, computer vision, and natural language processing. Furthermore, the traditional model-based optimization for NG-IoT systems will be revolutionized by model-free DRL, which can better meet the varying requirements of NG-IoT applications. The excellent capability of DRL has attracted many researchers to tackle a wide range of challenging problems (e.g., scheduling, caching, and routing) in IoT systems by exploring different DRL methods (e.g., multi-agent DRL and hierarchical DRL). The application of DRL not only can reduce the costs of NG-IoT but also enhance its reliability. Although DRL is with great potential for solving complex problems in NG-IoT, many domain-specific issues are needed to be further researched. For example, the trade-off optimization between Quality-of-Service (QoS) and energy consumption in NG-IoT, the efficient design of DNNs in DRL for different NG-IoT problems, and the balance between exploration and exploitation in DRL when interacting with the NG-IoT environment.

The objective of this special issue (SI) is to assemble high-quality research papers on emerging theories, frameworks and architectures for solving the practical problems related to DRL in NG-IoT. Meanwhile, this SI will offer an open platform for scholars and engineers to exchange their recent novel ideas and explore the potential convergence of existing NG-IoT systems and advanced DRL technologies.

Topics of interest for this special issue, include, but are not limited to the following:

  • Novel deep reinforcement learning algorithms for NG-IoT systems and applications
  • Deep reinforcement learning for smart manufacturing and smart factories
  • Deep reinforcement learning for efficient communications in NG-IoT
  • Deep reinforcement learning for scheduling, caching, and virtualization in NG-IoT
  • Multi-agent/hierarchical deep reinforcement learning for NG-IoT systems and applications
  • Deep reinforcement learning for traffic prediction in NG-IoT
  • Hybrid deep reinforcement learning for NG-IoT systems and applications
  • Light-weight deep reinforcement learning for large-scale industrial IoT systems
  • Deep reinforcement learning for network security and reliability in NG-IoT
  • Deep reinforcement learning for energy and QoS management in NG-IoT
  • Testbeds, simulations, and evaluation tools for deep reinforcement learning in NG-IoT
  • Deep reinforcement learning for detection and automation in NG-IoT

Submission Guidelines

Submitted papers should present original, unpublished work, relevant to one of the topics of the Special Issue. All submitted papers will be evaluated on the basis of relevance, significance of contribution, technical quality, and quality of presentation. Manuscripts need to prepared according to Guide for Authors at according to the Guide for Authors as published in the Computer Networks Journal at We invite the prospective authors to submit their manuscript, via the online submission system in the main journal page and select “DRL-IoT” as the Article Type. Please make sure you mention in your cover letter that you are submitting to this special issue.

Important Dates

Manuscripts Due: 15 November 2020

Feedback to Authors: 15 February 2021

Revised Manuscripts Due: 15 March 2021

Second-Round Reviews to Authors: 15 February 2021

Second-Revised Manuscripts Due: 15 April 2021

Final Accepted Manuscript Due: August 2021

Lead Guest Editor

Guest Editors

Other Special Issues on this journal

Closed Special Issues