Anasol Pena Rios

Umer Iqbal, Ashiq Anjum, Anthony S Conway et al.

Telecom networks scale with growing users and data-intensive applications, generating heavy traffic that causes congestion, reducing throughput, increasing delay, and raising computational costs. Traditional routing protocols act only after performance degradation, making them unsuitable for dynamic traffic and topological changes. Addressing these challenges requires a routing approach that adapts in real time, scales with network growth, operates without disrupting active services, and provides continuous feedback for congestion-aware traffic optimisation. The Network Digital Twin (NDT) addresses these needs by mirroring global network behaviour using Message Passing Neural Networks (MPNNs) through bidirectional communication with the physical network. To align the NDT with physical network behaviour, synthetic traffic is generated with increasing load across topological structures that incrementally scale as routers are added. These topologies are created by graph-generating models such as Erdos-Renyi, Barabasi-Albert, and Watts-Strogatz, customised with vertex degree limitations. The NDT collects performance metrics from routers and links, and MPNNs classify edges based on local vertex and global network behaviours. Based on these classifications, feedback is sent as Policy-Based Routing (PBR) protocol commands to each router, enabling optimal traffic distribution across links of the physical network.

Muhammad Saad Zia, Ashiq Anjum, Lu Liu et al.

Physics Informed Machine Learning has emerged as a popular approach for modeling and simulation in digital twins, enabling the generation of accurate models of processes and behaviors in real-world systems. However, existing methods either rely on simple loss regularizations that offer limited physics integration or employ highly specialized architectures that are difficult to generalize across diverse physical systems. This paper presents a generic approach based on a novel physics-encoded residual neural network (PERNN) architecture that seamlessly combines data-driven and physics-based analytical models to overcome these limitations. Our method integrates differentiable physics blocks-implementing mathematical operators from physics-based models with feed-forward learning blocks, while intermediate residual blocks ensure stable gradient flow during training. Consequently, the model naturally adheres to the underlying physical principles even when prior physics knowledge is incomplete, thereby improving generalizability with low data requirements and reduced model complexity. We investigate our approach in two application domains. The first is a steering model for autonomous vehicles in a simulation environment, and the second is a digital twin for climate modeling using an ordinary differential equation (ODE)-based model of Net Ecosystem Exchange (NEE) to enable gap-filling in flux tower data. In both cases, our method outperforms conventional neural network approaches as well as state-of-the-art Physics Informed Machine Learning methods.