Energy-Aware Pruning for Federated Deep Reinforcement Learning in Multi-Interface IoT Networks

Future wireless networks are expected to support ever increasing amounts of Internet of Things (IoT) data traffic, while satisfying heterogeneous and stringent quality of service (QoS) constraints. Although AI-based resource allocation approaches have shown great potential, current methods are unable to cope with the severe energy limitations of IoT devices. In this work, we investigate the IoT device-to-multi-access points (APs) association problem in heterogeneous sub-6-GHz/mmWave IoT networks. To reduce the complexity and latency issues inherent to centralized deep reinforcement learning (DRL) methods, we design a solution based on multiagent DRL (MADRL), where each IoT device selects its AP and band association, according to its local environment. This method is empowered by a federated learning (FL)-based aggregation process, enabling cooperation among agents with limited signaling costs. Unlike previous works, the proposed method fully adapts to the heterogeneous QoS demands and energy constraints of each IoT device. In particular, our approach is specifically designed to reduce energy consumption by exploiting DRL-tailored pruning, while handling the devices’ diverse requirements. Numerical results show that the proposed method outperforms benchmarks in terms of rate outage probabilities, while considerably reducing AI energy consumption.

• Publication year

  2026

• Venue

  IEEE Internet of Things Journal

• Publication date

  2026-02-15

• Fields of study

  Computer Science, Engineering, Environmental Science

• Identifiers
  DOI 10.1109/JIOT.2025.3639564
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Deep Reinforcement Learning-Empowered Federated Learning for Wireless Clients with Energy and Bandwidth Constraints

  2024cited by this paper
• Smart Band Association for Wireless IoT Networks: a Personalized Federated Multi-Agent Deep Reinforcement Learning Approach

  2024cited by this paper
• Federated Multiagent Deep Reinforcement Learning for Intelligent IoT Wireless Communications: Overview and Challenges

  2024cited by this paper
• Federated Multi-Agent Deep Reinforcement Learning (Fed-MADRL) for Dynamic Spectrum Access

  2023cited by this paper
• Pruning for Power: Optimizing Energy Efficiency in IoT with Neural Network Pruning

  2023influential reference
• Wireless Multi-Interface Connectivity with Deep Learning-Enabled User Devices: An Energy Efficiency Perspective

  2023cited by this paper
• Towards Personalized Federated Learning

  2021cited by this paper
• Towards an Energy-Efficient DQN-based User Association in Sub6GHz/mmWave Integrated Networks

  2021cited by this paper
• Dynamic Channel Access and Power Control in Wireless Interference Networks via Multi-Agent Deep Reinforcement Learning

  2021cited by this paper
• 6G Internet of Things: A Comprehensive Survey

  2021cited by this paper
• Deep Reinforcement Learning-based User Association in Sub6GHz/mmWave Integrated Networks

  2021influential reference
• Single and Multi-Agent Deep Reinforcement Learning for AI-Enabled Wireless Networks: A Tutorial

  2020cited by this paper
• Deep-Reinforcement-Learning-Based Mode Selection and Resource Allocation for Cellular V2X Communications

  2020cited by this paper
• Layer-adaptive Sparsity for the Magnitude-based Pruning

  2020cited by this paper
• Power and Beam Optimization for Uplink Millimeter-Wave Hotspot Communication Systems

  2019cited by this paper
• Federated Learning with Personalization Layers

  2019influential reference
• Energy-Constrained Compression for Deep Neural Networks via Weighted Sparse Projection and Layer Input Masking

  2018cited by this paper
• A method to estimate the energy consumption of deep neural networks

  2017cited by this paper
• A distributed learning–based user association for heterogeneous networks

  2017cited by this paper
• Communication-Efficient Learning of Deep Networks from Decentralized Data

  2016cited by this paper
• Free Space Optics Vs Radio Frequency Wireless Communication

  2016cited by this paper
• Learning both Weights and Connections for Efficient Neural Network

  2015cited by this paper
• Human-level control through deep reinforcement learning

  2015cited by this paper
• Compressing Deep Convolutional Networks using Vector Quantization

  2014cited by this paper
• Networks

  2007cited by this paper

• No citing papers are available for this paper.