TG-DIN: Theory-Guided Demand Inference Network for Generalizable QoS Measurement and Prediction

2026-05-15

AI Summary

Existing methods struggle to accurately infer latent user demand from observed network QoS data, leading to limited generalization under distribution shifts. This work proposes a theory-guided differentiable architecture that integrates scheduling and queuing theory into deep learning models to construct an interpretable demand inference framework. By introducing a differentiable theory layer, user demand is modeled as a latent variable, enabling stochastic training without access to ground-truth demand labels and facilitating deployment across diverse scenarios. Experimental results demonstrate that the proposed method, trained solely on synthetic data, accurately recovers user allocation structures over shared links in real-world traffic and significantly outperforms purely data-driven approaches under varying link capacities, demand levels, and traffic patterns.

Abstract

In this paper, we introduce TG-DIN, a theory-guided demand inference network that infers latent user demand from observable network quality-of-service (QoS) measurements. Rather than directly predicting QoS outcomes using black-box models, TG-DIN explicitly models latent demand as an intermediate variable and links it to observable behavior through a differentiable theory layer grounded in scheduling and queuing principles. This design yields an interpretable, mechanism-consistent representation of user demand that is directly applicable to downstream tasks such as congestion diagnosis, resource allocation, capacity planning, and policy evaluation. The theory layer further enables a principled randomized training regime that exposes the model to diverse yet physically meaningful operating conditions without requiring labeled demand data. Extensive synthetic experiments show that TG-DIN generalizes robustly across capacities, demand levels, and traffic patterns, substantially outperforming purely data-driven baselines under distribution shift. Moreover, when trained exclusively on synthetic data and applied directly to real packet traces, TG-DIN accurately recovers per-user allocation structure in shared-link scenarios. Together, these results demonstrate the effectiveness of theory-guided inductive biases for achieving transferable, deployment-ready inference in dynamic network environments.

Problem

• user demand inference
• QoS prediction
• network generalization
• latent variable modeling
• theory-guided learning

Innovation

• theory-guided learning
• demand inference
• differentiable theory layer
• QoS prediction
• generalizable network modeling