What question did this study set out to answer?

The central aim is to enhance adaptive signal processing in 6G networks by utilizing a deep reinforcement learning model.

April 1, 2026

Multi-Objective, Mobility-Aware, and Lightweight Deep Reinforcement Learning for Adaptive Signal Processing in Real-Time 6G Edge Networks

Key Points

The central aim is to enhance adaptive signal processing in 6G networks by utilizing a deep reinforcement learning model.
Developed a dual-agent actor-critic deep reinforcement learning model
Incorporated convolutional features and mobility forecasting
Applied in real-time for modulation and power allocation tasks
Compared performance against traditional methods
Reduced end-to-end latency by up to 27% compared to tabular RL methods
Achieved a 20% latency reduction relative to A3C algorithms
Improved energy efficiency by 13–16% over fixed-weight configurations
Maintained low bit error rates in vehicular and aerial simulations

Abstract

As sixth-generation (6G) wireless networks advance, real-time adaptive signal processing is critical for mobility and multi-objective constraints. Traditional methods fall short in dynamic edge environments. We propose a DRL model with dual-agent actor-critic, convolutional features, and mobility forecasting for tasks like modulation and power allocation. It learns in real-time, reducing end-to-end latency by up to 27% relative to tabular RL baselines and by 20% relative to A3C, improving energy efficiency by 13–16% compared to fixed-weight and mobility-unaware configurations, and maintaining low bit error rates in vehicular and aerial simulations. This enables applications in transportation, drones, and IoT.

Demander à l'IA

Bookmark