DIFFRACT: Neuralized Utility Maximization for Wireless Networks by Differentiable Programming

2026-06-05

AI Summary

This work addresses the challenge of resource management under dynamic multi-user interference in next-generation wireless systems, such as satellite and open radio access networks, by proposing the DIFFRACT framework. DIFFRACT enables efficient utility maximization subject to stochastic quality-of-service constraints. It establishes, for the first time, a duality theory for standard interference functions and integrates algorithmic unfolding with differentiable programming to construct a mathematically principled differentiable neural network. This architecture supports end-to-end gradient-based learning and facilitates distributed deployment at the network edge. Experimental results demonstrate that DIFFRACT significantly enhances scalability, robustness, and real-time adaptability of wireless resource management across diverse space-air-ground scenarios while preserving theoretical rigor.

Abstract

Next-generation wireless networks, including satellite-to-Open RAN systems, demand agile and intelligent resource management capable of handling dynamic multi-user interference under stochastic quality of service constraints. This paper introduces DIFFRACT, a neuralized utility maximization framework that leverages differentiable programming to integrate deep learning with optimization in wireless networks. Central to our approach is the exploitation of the mathematical structure of standard interference functions, which are foundational in wireless power control. By developing a duality theory for these functions, we map iterative interference management algorithms into differentiable neural network architectures via algorithm unrolling. This enables distributed, end-to-end gradient-based learning at the network edge, supporting real-time adaptation to interference in both terrestrial and non-terrestrial environments. DIFFRACT allows for scalable and robust utility maximization by modeling complex channel dynamics and leveraging the expressiveness of differentiable models. Experimental results confirm the framework's theoretical soundness and practical effectiveness for next-generation wireless systems.

Problem

• wireless networks
• resource management
• multi-user interference
• quality of service
• utility maximization

Innovation

• differentiable programming
• algorithm unrolling
• interference management
• neuralized optimization
• utility maximization