Edge Processing Design: Intelligent Vision


Products incorporating technologies with artificial intelligence (AI) at the network edge are a growing market trend. However, supporting their functionality from the cloud has run into problems due to latency, safety concerns and the cost of transferring data.

This case study is an example of how you can architect a chip to support AI within an edge device.

Design Requirements & Chosen Architecture

The client required a SoC to support high end visual data capture and analysis. It needed to be able to monitor a perimeter, track specific classifications of objects and recognise vehicle license plates for traffic control.

This system on chip has a heterogenous multicore architecture in TSMC 28nm technology, enabling real-time analytics and the ability to send data to the cloud or to a remote server for further processing and storage.

Accelerated SoC Construction

To facilitate accelerated construction of the design solution for this device, our architects started with a reference design, as we have an existing modelling environment that we can configure according to project needs. This approach benefits the client with:

  • lower project risk and uncertainty with better budget control
  • It enables us to automate many of the integration steps, boosting productivity
  • it allows us to maximise reuse of subsystems
  • faster time to market

For more information please contact us