A Combined Push-Pull Access Framework for Digital Twin Alignment and Anomaly Reporting

TL;DR

This paper introduces a push-pull scheduler framework for digital twins that optimizes update communication, improving alignment accuracy and anomaly detection responsiveness while efficiently managing resources.

Contribution

It proposes a novel push-pull scheduler (PPS) framework that dynamically allocates communication resources for different update types in digital twins, balancing accuracy and anomaly reporting.

Findings

Reduced drift age of incorrect information by over 20%.

Lowered worst-case anomaly detection AoII from 70 ms to 20 ms.

Maintained anomaly detection guarantees while optimizing resource use.

Abstract

A digital twin (DT) contains a set of virtual models of real systems and processes that are synchronized to their physical counterparts. This enables experimentation and examination of counterfactuals, simulating the consequences of decisions in real time. However, the DT accuracy relies on timely updates that maintain alignment with the real system. We can distinguish between: (i) pull-updates, which follow a request from the DT to the sensors, to decrease its drift from the physical state; (ii) push-updates, which are sent directly by the sensors since they represent urgent information, such as anomalies. In this work, we devise a push-pull scheduler (PPS) medium access framework, which dynamically allocates the communication resources used for these two types of updates. Our scheme strikes a balance in the trade-off between DT alignment in normal conditions and anomaly reporting, optimizing resource usage and reducing the drift age of incorrect information (AoII) by over 20% with respect to state-of-the-art solutions, while maintaining the same anomaly detection guarantees, as well as reducing the worst-case anomaly detection AoII from 70 ms to 20 ms when considering a 1 ms average drift AoII constraint.