Shifting AI Computation from Cloud to Edge

Giuseppe Amato, Technical Manager & Business Development Europe, VIA Technologies

The Internet Of Things is collecting information from the nodes but to use the data efficiently the latency have to be reduced. This mean moving from centralized data (cloud) to decentralized data (edge). With Edge computing the data are processed near the source (iOT nodes) reducing data traffic while improving the response time of the iOT node.

Edge Computing brings the following advantages:
1. Reduction of data traffic from the iOT node to the cloud.
2. Increase security because less datas are transferred to the cloud.
3. Reduce the response time.
4. Scalability.
5. Predictive maintenance of the automation equipment through analysis of reference model with the real model.
6. Deviation of produced element with “golden” model through image recognition.
7. Image recognition for access control.

While Edge Computing requires more computation at “node” level, the “value” brought by this computation, reduce the Total Cost Of Ownership bringing all the advantages listed above.

VIA has developed Edge Computing Platform and added related algorithms and SW Framework to enable Artificial Intelligence running at the Edge.

With VIA platform, certified to work with Microsoft Azure, together with EdgeML SW Framework from our partner FogHorn we will provide to our customers an Intelligent data aggregation unit having the capabilities to analyze data at Edge level while forwarding non critical data to the cloud.

During the speech VIA will highlight:
1. Edge Computing Platform with Machine Learning SW Framework to run data analytics at the Edge.
2. Smart Camera with Image Processing.