Learning to optimize boosts ai-driven network control and decision making
Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks
Artificial Intelligence
Summary
Communication networks that use artificial intelligence often need to solve complex problems quickly, like deciding how to use resources or pick the best routes for data. This paper explains a new approach where optimization algorithms first generate expert solutions offline, which are then used to train AI models. These AI models can later make fast, smart decisions in real time, even in changing network conditions. The authors show how this method works by training a model for vehicle-to-everything communication that mimics near-best solutions quickly.
What this means in practice
- •For network schedulers: Use optimization-generated training data to build AI models enabling quick, near-optimal scheduling decisions without heavy online computation.
- •For vehicle communication engineers: Deploy AI models trained on optimization solutions to select relay nodes in vehicle-to-everything systems with real-time responsiveness.
A position paper. It proposes an approach and reports no results.
Authors
Giambattista Amati, Federica Mangiatordi, Pierpaolo Salvo, Emiliano Pallotti, Simone Angelini
Abstract
Artificial Intelligence (AI) is becoming a fundamental design principle of future AI-native communication networks, enabling autonomous resource management, adaptive control, and zero-touch network operation. While current AI-native architectures increasingly embed intelligence across network functions, they provide little guidance on how optimisation knowledge should be systematically generated, transferred, and exploited by AI models. This paper argues that the Learning-to-Optimize (L2O) represents the missing architectural layer between optimisation and AI-native intelligence. Rather than viewing optimisation merely as an online decision engine, the proposed paradigm redefines optimisation algorithms as offline knowledge generators that produce high-quality supervisory information for neural surrogate models. The resulting models inherit optimisation expertise while enabling low-latency runtime inference suitable for dynamic network environments. A generic four-stage L2O workflow is introduced, comprising optimisation, knowledge generation, surrogate learning, and runtime inference. Unlike existing Learning-to-Optimize approaches, which primarily focus on algorithm acceleration, the proposed framework establishes L2O as an architectural abstraction applicable across heterogeneous communication and computing systems. The proposed paradigm is illustrated by an NR-V2X relay-selection problem, in which optimisation-generated solutions from a Mixed-Integer Linear Programming (MILP) solver are used to train a Graph Neural Network that can reproduce near-optimal decisions in real time. The presented perspective positions Learning-to-Optimize as a key architectural enabler for future AI-native networks.