Reinforcement-Learning-Based Resource Allocation for Energy-Harvesting-Aided D2D Communications in IoT Networks

This article proposes a novel approach to improve the energy efficiency (EE) of an energy-harvesting (EH)-enabled IoT network supported by simultaneous wireless information and power transfer (SWIPT). More specifically, the device-to-device (D2D) users harvest ambient energy throughout their communi...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE internet of things journal Jg. 9; H. 17; S. 16521 - 16531
Hauptverfasser: Omidkar, Atefeh, Khalili, Ata, Nguyen, Ha H., Shafiei, Hossein
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Piscataway IEEE 01.09.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
Algorithms
Computational geometry
Convexity
Device-to-device (D2D) communications
Device-to-device communication
Energy
Energy harvesting
Internet of Things
Internet of Things (IoT) networks
Learning
majorization–minimization (MM)
Mixed integer
mixed-integer nonlinear problem (MINLP)
Optimization
Optimization techniques
Power control
Power management
Power transfer
Quality of service
reinforcement learning (RL)
Resource allocation
Resource management
spectrum sharing
Wireless communication
ISSN:2327-4662, 2327-4662
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This article proposes a novel approach to improve the energy efficiency (EE) of an energy-harvesting (EH)-enabled IoT network supported by simultaneous wireless information and power transfer (SWIPT). More specifically, the device-to-device (D2D) users harvest ambient energy throughout their communication with the time switching (TS) technique, while the Internet of Things (IoT) users harvest energy from the base station (BS) based on the power splitting (PS) method. We study the EE optimization problem that takes into account transmit power feasibility conditions for D2D users and IoT users, the minimum data rate requirements for D2D and IoT users, joint spectrum sharing block, and time allocation for the D2D links. The underlying problem is a highly nonconvex mixed-integer nonlinear problem (MINLP) and the global optimal solution is intractable. To handle it, we decompose the original problem into three subproblems: 1) joint subchannel allocation and PS; 2) power control; and 3) time allocation. Since it is difficult to find an exact state model approach in a dynamic environment with a large state space, we exploit a <inline-formula> <tex-math notation="LaTeX">Q </tex-math></inline-formula>-learning method based on reinforcement learning (RL) to solve the first subproblem. To solve the second subproblem, we apply a conventional convex optimization technique based on the majorization-minimization (MM) approach and Dinkelbach method. Simulation results not only demonstrate the superiority of our proposed algorithm as compared to other methods in the literature but also confirm impressive EE gains through spectrum sharing and harvested energy from D2D and IoT users.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2327-4662
2327-4662
DOI:10.1109/JIOT.2022.3151001