Five Critical Steps to Enable Real-time Edge Processing in Your Enterprise

BrandPostBy Ken Durazzo, Nicole Reineke
Apr 21, 2021
Edge Computing IT Leadership

connected city featured 4 21 21
Credit: Dell Technologies

Edge intelligence solutions are no longer the future — they are here today. At Dell Technologies, our professionals are deploying prototypes of edge computing solutions driven by artificial intelligence in growing numbers of customer environments.

In one such example, we are working with an Olympic training team at the China National Aquatic Training Base, Qiandaohu, to enable the use of intelligent real-time processing at the edge to evaluate rowing techniques. This training program uses the Dell EMC Streaming Data Platform and AI on local edge environments, such as indoor gyms, as well as in far edge locations, such as drones that capture images of athletes rowing, for pose detection and biomechanical analysis. 

In this initiative, led by Dr. Sanping Li, a Senior Principal Research Scientist at Dell Technologies, the project team trained an algorithm to recognize signs of optimal athletic performance using images of many athletes. The team then put this model to work in a prototype environment that captures real-time images of athletes in action and uses the AI model running in an on-premises Dell Technologies system to compare the captured pictures with images of ideal biomechanical movements.

With the insight gleaned from this AI application, the Olympic training team can provide data-driven feedback on steps rowing athletes might take to optimize their movements and improve their overall performance. And while this AI-driven solution is targeted at rowing techniques, the training center’s kayak and canoe coaches have already expressed an interest in adapting the solution to their sports.

optimized rowing embeded 4 21 21

The bigger picture

There are opportunities to capitalize on real-time edge processing in all industries. From monitoring and controlling the movement of autonomous forklifts in a distribution center to using computer vision to categorizing images streaming into an endpoint, real-time edge processing can provide immediate and highly valuable business insights.

The initial challenge here is to determine the ways in which fast processing at the edge will impact your industry, and then make plans for incorporating the technology into your processes. With a brief analysis of multi-edge access computing technologies and a look at the AI solutions available today, you can project ways to include the new technology in your busines plans — and lay the foundation for a competitive edge.

With these thoughts in mind, we propose a five-step action plan for capitalizing on the promise of real-time edge processing.

Action 1: Understand multi-access edge computing.

Multi-access edge computing  (MEC) is a method of bringing processing power and other technology resources closer to end users and the points where data is generated and captured. The MEC approach couples an IT server environment with cloud computing capabilities to enable the real-time enterprise.

MEC takes place over a continuum of compute that spans from end-user devices to a centralized cloud. Here is a simplified example showing a basic workflow for multi-access edge computing.

Centralized Cloud

Latency-tolerant applications

5G Edge

Low-latency applications

Access Network

High-throughput bandwidth

5G Devices

Low-cost thin clients

← Data flow →

In this MEC continuum, edge data moves back and forth between 5G thin-client devices and a high-throughput network designed to handle massive data volumes. Further along the continuum, an edge cloud processes data for mission-critical low-latency applications, and beyond that a centralized cloud runs general-purpose, latency-tolerant applications. All along the continuum, data flows both east and west across the high-bandwidth network.

Action 2: Determine the types of end locations that matter to your business.

Your end locations might be stationary or mobile. Are you interested in end-user devices or connected equipment on the Internet of Things? Are you focused on homes, branch offices, stores or vehicles? The key here is to understand the types of end locations that are mission-critical. These locations are likely to be ideal candidates for collecting data and may be a location to perform real-time edge processing.

Action 3: Determine the types of information that you can gather at your end locations. 

This next step is closely tied to the specifics of your industry. In a wind-powered energy-generation system, the data you capture might pertain to mechanical dynamics, such as the vibration of components. On a manufacturing floor, this information might include visual identification of defective parts coming down the line. The key is to understand your data opportunities.

Action 4: Determine how you can use the information to improve your business.

At this point, the focus shifts to turning targeted data from select endpoints into immediate business value. What is it that you need to understand about the endpoint? What type of data-driven insights would allow you to better serve your customers? How can real-time edge processing help your people work smarter and accelerate the pace of your business? The answers to questions like these will help you gain a clear view of what edge processing will do for your business.

Action 5: Create your North Star and ‘backcast’ your edge-processing roadmap.

This next action establishes the vision for your edge-processing roadmap and then backcasts, or works backward, to identify the specific steps that will lead your organization to the ultimate destination. This process takes the vision for real-time edge processing and grounds it in reality.

One of these steps includes the identification of the products and solutions you will use for data capture, data processing and data storage at the edge, from data streaming platforms and rugged servers to edge management software. Other steps include determination of the processes and software tools you will use to develop and train AI models for your targeted tasks.

Ultimately, through this process, your organization will establish a clear vision for adoption of the AI and real-time edge processing workflows that will help your business thrive in the new digital economy.

Key takeaways

Real-time processing of edge data isn’t a vision for the future. It’s here today, and it is emerging as a critical success factor for organizations seeking to compete in the new digital economy. To move forward, IT leaders need to understand the big picture — the opportunity — and then shift the focus to the five concrete steps that lead to real-time edge processing and the corresponding business benefits.

To learn more

For a look at products and solutions for real-time edge-processing, explore the Dell Technologies Edge Portfolio. Explore more edge solutions from Dell Technologies and Intel.

Featuring the work of Dr. Sanping Li and his colleagues.