Different Types of Edge Computing

The types of edge computing and examples of use cases for each.

Many organizations have started their journey towards edge computing to take advantage of data produced at the edge. The definition of edge computing is quite broad. Simply stated, it is moving compute power physically closer to where data is generated, usually an edge device or IoT sensor.

This encompasses far edge scenarios like mobile devices and smart sensors, as well as more near edge use cases like micro-data centers and remote office computing. In fact, this definition is so broad that it is often talked about as anything outside of the cloud or main data center. 

With such a wide variety of use cases, it is important to understand the different types of edge computing and how they are being used by organizations today. 

Provider edge

The provider edge is a network of computing resources accessed by the Internet. It is mainly used for delivering services from telcos, service providers, media companies, or other content delivery network (CDN) operators. Examples of use cases include content delivery, online gaming​, and AI as a service (AIaaS). 

One key example of the provider edge that is expected to grow rapidly is augmented reality (AR) and virtual reality (VR). Service providers want to find ways to deliver these use cases, commonly known as eXtended Reality (XR), from the cloud to end user edge systems. 

In late 2021, Google partnered with NVIDIA to deliver high-quality XR streaming from Google Cloud NVIDIA RTX powered servers, to lightweight mobile XR displays. By using NVIDIA CloudXR to stream from the provider edge, users can securely access data from the cloud at any time and easily share high-fidelity, full graphics immersive XR experiences with other teams or customers.

Enterprise edge

The enterprise edge is an extension of the enterprise data center, consisting of things like data centers at remote office sites, micro-data centers, or even racks of servers sitting in a compute closet on a factory floor. This environment is generally owned and operated by IT as they would a traditional centralized data center, though there may be space or power limitations at the enterprise edge that change the design of these environments.

Retailers can use edge AI across their business for frictionless shopping, in store analytics as well as supply chain optimization.
Figure 1. Enterprises across all industries use edge AI to drive more intelligent use cases on site.

Looking at examples of the enterprise edge, you can see workloads like intelligent warehouses and fulfillment centers. Improved efficiency and automation of these environments requires robust information, data, and operational technologies to enable AI solutions like real-time product recognition.

Kinetic Vision helps customers build AI for these enterprise edge environments using a digital twin, or photorealistic virtual version, of a fulfillment or distribution center to train and optimize a classification model that is then deployed in the real world. This powers faster, more agile product inspections, and order fulfillments.

Industrial edge

The industrial edge, sometimes called the far edge, generally has smaller compute instances that can be one or two small, ruggedized servers or even embedded systems deployed outside of any sort of data center environment.

Industrial edge use cases include robotics, autonomous checkout, smart city capabilities like traffic control, and intelligent devices. These use cases run entirely outside of the normal data center structure, which means there are a number of unique challenges for space, cooling, security, and management.

BMW is leading the way with industrial edge by adopting robotics to redefine their factory logistics. Using different robots for parts of the process, these robots take boxes of raw parts on the line and transport them to shelves to await production. They are then taken to manufacturing, and finally returned back to the supply area when empty.

Robotics use cases require compute power both in the autonomous machine itself, as well as compute systems that sit on the factory floor. To optimize the efficiency and accelerate deployment of these solutions, NVIDIA introduced the NVIDIA Isaac Autonomous Mobile Robot (AMR) platform.

Accelerating edge computing

Each of these edge computing scenarios has different requirements, benefits, and deployment challenges. To understand if your use case would benefit from edge computing, download the Considerations for Deploying AI at the Edge whitepaper.

Sign up for Edge AI News to stay up to date with the latest trends, customers use cases, and technical walkthroughs.

Leave a Reply

Your email address will not be published. Required fields are marked *