5 min read
Addressing the Infrastructure Condition Crisis
The condition of our civil infrastructure is approaching a crisis point. We need AI to help us reverse the damage. Every individual, community, and...
Case Study
Compressing data without having to compromise
Achieve unrivalled real-time intelligent compression of high-frequency
sensor data without extra computing power.
200
Mhz
speed of processor sufficient to run algorithms.
10
x
reduction in the size
of data recorded.
Problem:
Capturing, processing and storing data in resource-constrained environments can be extremely problematic.
Our client had a complex industrial process they wanted to replicate and analyse in a digital representation of the process. However, the data collection was locked behind a bandwidth bottleneck due to hardware limitations, and the process generated more data than required to summarise the state of operation. The client needed a real-time and intelligent compression technique to ensure sufficient data was logged for analysis.
Solution:
Mind Foundry developed a customised intelligent sampling solution that was able to reduce streams of continuous data, sampled at arbitrary frequencies, by an order of magnitude appropriate to the complexity of the underlying signal in the data.
The solution was to be deployed on an edge device, so the algorithm was required to process a data stream generating several gigabytes per hour using a single core of a single processor running at less than 200MHz. Nevertheless, the algorithm was able to run in real time under these extreme constraints on computational resources.
Results:
The technology enabled the client to develop advanced health monitoring checks for their complex machinery using the newly available data. Including:
-
Running the solution on a 200Mhz processor.
-
10x reduction recorded in the size of data.
-
Successful digital twin made possible by intelligent compression.
Contact us for more information.
Deepen your knowledge
5 min read
The Puzzle of Scaling AI in Insurance
by Mind Foundry
Building a performant model in insurance is one thing. Building, deploying, maintaining, and governing hundreds of models in production is another...
5 min read
Measuring What Really Matters
by Alistair Garfoot
Why end-user insight and operational context are the keys to understanding AI performance in the defence sector. In Defence and National Security,...
Stay connected
News, announcements, and blogs about AI in high-stakes applications.