Joint Task Offloading and Resource Allocation for Intelligent Reflecting Surface-Aided Integrated Sensing and...

Area of Science:

• Wireless communication
• Signal processing
• Artificial intelligence

Background:

• Spectrum scarcity and poor wireless environments hinder integrated sensing and communication (ISAC) systems.
• Optimizing resource allocation is crucial for balancing sensing, communication, and computational offloading performance.

Purpose of the Study:

• To propose an intelligent reflecting surface (IRS)-aided ISAC framework.
• To optimize system performance by jointly designing beamforming, phase shifts, and resource allocation.
• To address the non-convex and NP-hard optimization problem using advanced AI techniques.

Main Methods:

• Formulation as a Markov decision process for dynamic channel conditions.
• Development of two deep reinforcement learning (DRL) schemes: Deep Deterministic Policy Gradient (DDPG) and Twin Delayed DDPG.
• Inclusion of prioritized experience replay to accelerate DRL convergence.

Main Results:

• The proposed DRL-based schemes effectively optimize spectrum and computing resource allocation.
• Enhanced radar sensing quality and increased communication data rates were achieved.
• Improved energy efficiency and reduced latency in computational offloading.

Conclusions:

• The IRS-aided ISAC framework with DRL significantly outperforms benchmark methods.
• The proposed DRL approaches provide a viable solution for complex optimization in dynamic wireless environments.
• This work demonstrates the potential of AI in advancing ISAC systems.