Dynamic Demand

What is it

Dynamic Demand is an AI-based application developed by One Big Circle to manage the flow of data from its sensors in the railway environment. It ensures that the most important data and information is rapidly made available to users of the AIVR Platform.

Dynamic Demand provides AIVR engineers a clear, up‑to‑date picture of the railway, while helping to reduce the volume of data that must be transmitted, stored, processed and analysed.

Where it’s used

AIVR sensors include video and thermal imaging. They are located on both moving trains and static trackside locations around the UK. Known collectively as Edge devices, they transmit data to AIVR’s data centre in the Cloud via the internet, using 4G and 5G networks.

Why it’s needed

AIVR’s Edge devices generate vast data sets simultaneously from multiple positions across the network. Large data sets are costly to transmit and store, and complex to process. If the most useful information within large data sets at the Edge is to be unlocked and shared quickly and efficiently on the AIVR Platform, these data sets must be effectively managed.

This problem required a solution that orchestrated the many data streams flowing from hundreds of Edge devices to the Cloud, intelligently filtering data so that only the most relevant, high‑quality data was uploaded to the Cloud.

When and how it is used

First deployed in AIVR systems in 2023, Dynamic Demand establishes a continual dialogue between the Edge and the Cloud.

Dynamic Demand is executed by the AIVR Cloud Agent, an intelligent program that automates decision-making. The Cloud Agent prioritises the transmission of useful data from Edge devices to the Cloud, while identifying and holding back data it deems unlikely to be helpful. (For example, AIVR sensors may be mounted on a train that runs on the same route 10 times a day. Much of the data generated is redundant, so the Cloud Agent will choose to upload only the best data from these runs.)

The Cloud Agent operates dynamically, automatically scaling the demand for data up or down to reflect the Cloud’s current capacity.

Simultaneously, the Cloud Agent assembles data from the Edge to create a ‘picture’ of the rail network. It assesses the quality of this data using Machine Learning models, identifies parts of the data set that require updating or improving, and automatically ‘demands’ the best available data from Edge devices to fill these gaps.

Dynamic Demand enhances the effectiveness of the whole AIVR product suite.

READ MORE in our white paper Automation in overdrive: engineering a revolution.