Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Reinforcement Learning for Joint Time and Power Management in SWIPT-EH CIoT

Nadia Abdolkhani, Nada Abdel Khalek, Walaa Hamouda, Iyad Dayoub

TL;DR

The paper addresses joint time-switching and transmit power control for SWIPT-enabled CIoT in an underlay cognitive radio setting, modeled as a model-free MDP. It proposes a lightweight DDQN with UCB exploration to learn policies that adapt the TS ratio and power in real time. The method balances long-term throughput with energy and interference constraints, demonstrating superior performance over existing DRL baselines in simulations. This approach enhances the viability of energy-constrained CIoT deployments by improving throughput and reliability under realistic EH and spectrum-sharing conditions.

Abstract

This letter presents a novel deep reinforcement learning (DRL) approach for joint time allocation and power control in a cognitive Internet of Things (CIoT) system with simultaneous wireless information and power transfer (SWIPT). The CIoT transmitter autonomously manages energy harvesting (EH) and transmissions using a learnable time switching factor while optimizing power to enhance throughput and lifetime. The joint optimization is modeled as a Markov decision process under small-scale fading, realistic EH, and interference constraints. We develop a double deep Q-network (DDQN) enhanced with an upper confidence bound. Simulations benchmark our approach, showing superior performance over existing DRL methods.

Deep Reinforcement Learning for Joint Time and Power Management in SWIPT-EH CIoT

TL;DR

The paper addresses joint time-switching and transmit power control for SWIPT-enabled CIoT in an underlay cognitive radio setting, modeled as a model-free MDP. It proposes a lightweight DDQN with UCB exploration to learn policies that adapt the TS ratio and power in real time. The method balances long-term throughput with energy and interference constraints, demonstrating superior performance over existing DRL baselines in simulations. This approach enhances the viability of energy-constrained CIoT deployments by improving throughput and reliability under realistic EH and spectrum-sharing conditions.

Abstract

This letter presents a novel deep reinforcement learning (DRL) approach for joint time allocation and power control in a cognitive Internet of Things (CIoT) system with simultaneous wireless information and power transfer (SWIPT). The CIoT transmitter autonomously manages energy harvesting (EH) and transmissions using a learnable time switching factor while optimizing power to enhance throughput and lifetime. The joint optimization is modeled as a Markov decision process under small-scale fading, realistic EH, and interference constraints. We develop a double deep Q-network (DDQN) enhanced with an upper confidence bound. Simulations benchmark our approach, showing superior performance over existing DRL methods.

Paper Structure

This paper contains 8 sections, 6 equations, 2 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Scope and Contributions of the Work
  3. System Model and Problem Formulation
  4. Deep Reinforcement Learning for Joint Time and Power Management
  5. The Model-Free Markov Decision Process
  6. Proposed DDQN-UCB Strategy
  7. Simulation Model and Results
  8. Conclusion

Figures (2)

  • Figure 1: The proposed DDQN-UCB algorithm
  • Figure 2: (a) Benchmarking the ASR performance of our DDQN-UCB strategy in comparison to the existing strategies in the literaturetashma_drlIoT_2024_Nada_NadiaZarif_2022_dueling_vanilla_DQN, (b) Illustrating the effect of varying the number of slots occupied by PU $L$ and the number of time slots $T$ on our proposed DDQN-UCB strategy, and (c) Presenting the impact of varying the initial battery level $B_0$ and the duration of each time slot $\tau$ on our proposed DDQN-UCB strategy.