Edge AI and the urban digital twin

This article examines three uses of data and analytics in urban systems: energy metering, traffic flows and development scenario modelling. Each area requires consideration of information sources, integration, operator authority and operation with limited connectivity.

Intelligent metering and consumption analysis

Consumption anomalies may indicate losses, equipment issues or unmetered load. In this scenario, models help identify such signals in a meter data management system, or MDMS.

Federated learning is considered as an option for working across data sources without centralising them. The information exchanged, protection measures and access policies require separate design work.

The aim is to help identify the causes of deviations and select follow-up actions. Any reduction in commercial losses would need to be evaluated in use; this article does not report measurements of that outcome.

Traffic management with Edge AI

Cameras and local compute nodes can estimate vehicle, pedestrian and public-transport flows. In the proposed design, operators receive analytical results; whether source video is transmitted depends on the use case and data-handling policies.

The results can support signal-plan preparation and traffic assessment. Automatic adjustments, permitted actions and operator intervention are defined separately. Project validation covers latency, detection quality, network load and behaviour during connectivity loss.

A digital twin of the urban environment

A digital twin combines geospatial data, transport, energy, environmental indicators and development plans. It allows planners to compare scenarios before construction begins: accessibility, network demand, emissions and resilience to extreme events.

Data and operational requirements

privacy and data minimisation;
transparent decision criteria;
meaningful human intervention;
resilience to connectivity and data-source failures;
documented interfaces and assessment of supplier dependencies;
regular assessment of outcomes for residents and the urban economy.

Design work should identify data sources and owners, update frequency, model limitations, operator actions and connectivity-loss scenarios. For the digital twin, it should also record assumptions and the method used to validate scenario calculations.