Edge and IoT: key partners in enabling the real-time enterprise

Edge computing and IoT are a perfect marriage to enable real-time data to drive objects in the physical world. They allow fast data analysis to be closer to end users – particularly for safety-critical applications – while helping to minimize network traffic and operational costs.

Do you ever get the feeling the world is speeding up? You might be right. IDC predicts that by 2025, nearly 30% of the global “datasphere” will be analyzed and used in real time, up from just 15% in 2015, with a mind bending 175 zettabytes of data being produced by 80 billion things.

For example, smart cities need real-time data to analyze current congestion levels on the road so they can sequence the traffic lights in the most effective way. In factories, real-time data is vital to control automated robotic production lines, helping to ensure the safety of workers walking around a site and maintain high product quality and efficiency levels.

The datasphere encompasses all the data that is generated, collected and stored in any given year across the world. Over the last 10 years, enterprises have been rapidly migrating data workloads from on-premises mainframe servers to the cloud to benefit from highly scalable and cost-effective processing power. IDC estimates that by 2020 the average person will generate 1.5 GB worth of data per day. Cloud computing won’t be able to handle it all.

As Bask Iyer, CIO and GM, Edge Computing and IoT at Dell Technologies and VMware, points out: “The cloud is not free. You can’t backhaul every bit of traffic to the cloud as it’s cost prohibitive. Secondly, the latency just won’t work for every application. You don’t want your Tesla car to ask the cloud if it should stop if it sees a deer or a child – it needs to make that call right away. For these reasons, enterprises will increasingly need edge computing.”

Today, data is generally sent from a “thing” to a cloud data center for processing and analysis. With the rapid rise in the number of connected objects in the world and the massive volumes of data that are being collected, this is no longer tenable. For example, AI and machine learning can now be applied to CCTV video footage at source for face, behavior and object recognition to identify shop-lifters or overcrowding in public areas. It would be cost prohibitive to send all the video footage to the cloud for analysis.

By 2025, nearly 30% of the global “datasphere” will be analyzed and used in real time.

Compute closer to the edge

Instead, basic processing can be performed near the cameras – on-campus or even within the devices themselves. Sections of video that are flagged as suspicious are sent to an aggregation point, which has the compute and storage resources available to compare images and produce valuable insights in the form of improved detection of objects or people. This is edge computing, which places high-performance compute, storage and network resources as close as possible to end users and devices.

Bask highlighted this concept in action in the city of Las Vegas in the U.S. The city wanted to leverage video, microphones and other sensors to generate real-time alerts for incidents so the city could respond more rapidly and accurately. Working with Dell Technologies and VMware, the city deployed cameras and other sensors, as well as gateways for edge processing, to sense and capture incident alerts as they happen – and respond with the right people and right equipment, and in the right way.

Advisory firm ABI Research has predicted that AI inference on the edge will grow to 43% in 2023, up from just 6% in 2017.

The second big driver for edge computing are applications that are hyper-sensitive to latency. Automotive is a case in point. When cars can share data insights with each other and roadside infrastructure, tasks like automated overtaking, cooperative collision avoidance and “platooning” (with vehicles forming high-density road trains) become possible. This is all dependent on the ability to react to unpredictable and dynamic events in the real world to make driving safer.

The camera, radar, LiDAR and ultrasonic sensors on-board each car and sensors on roadside infrastructure use deep learning to interpret events as they happen, powering “see-through services” – augmented reality displays that enable people to see hazards around the corner. Orange has been testing these capabilities on autonomous cars with Ericsson. The sub-millisecond responses delivered by edge analytics solutions will also be required for factory automation applications where workers and robots need to work together safely.

Advanced on-device processing and analytics

“Edge computing is growing in importance in IoT environments where there may be intermittent access to connectivity, low tolerance for network latency or high security demands,” according to Stacy Crook, Research Director, IoT at IDC. “As a result, IoT platform vendors are expanding their cloud offerings to support data ingestion, processing and analytics at the edge.”

Wind farms are a great example here. Using AI and edge-computing capabilities on each turbine, it’s possible to analyze the shifts in wind direction in real time. The front turbine can tilt the angle of its blades to allow some of the breeze through. This enables energy companies to capture 20% more power across all the turbines that act almost like a “flock” of birds.

Meanwhile, in the oil and gas industry, pressure and humidity sensors in pipelines need to be closely monitored and cannot afford a lapse in connectivity for environmental and safety reasons, especially as they’re mostly located in remote areas. Meanwhile, India's largest dairy – Chitale – is using IoT and edge-computing technology from Dell Technologies and VMware to help farmers keep their cows healthier and more productive, creating new economic opportunities that transform communities. From monitoring dairy cow habits and health using IoT sensors, to automating and improving milk production through a high-speed, high-availability network, Chitale is the top of the Indian dairy industry. Everyone wins: the farmers, the company, the community — and, yes, even the cow.

From a cyberdefense point of view, edge computing enables you to carry out authentication tasks closer to end users, which is inherently more secure. And by decentralizing data and distributing it among devices, it’s far more difficult to bring down a whole network with a single attack. By processing and refining data locally, you can also cut down on the amount of sensitive data sent through networks, addressing privacy, governance and security issues. For example, you can zoom in on suspected criminals in CCTV footage and anonymize the rest of the video footage.

Currently, around 10% of enterprise-generated data is created and processed outside traditional centralized data centers or clouds. Gartner predicts that by 2025, this figure will have hit 75%.

The edge is blurred

Edge computing will come in many guises. It can be co-located with motors, pumps, generators or other sensors. It can be mobile (as in airplanes) or static on production lines or even a combination of the two (as is the case with autonomous vehicles which use on-board and roadside AI). In complex scenarios, enterprises will use micro clouds or clusters of edge servers. According to Iyer, “Instead of thinking about the network edge, the compute edge, the device edge, we look at edge as a complete solution – all software defined for maximum flexibility and control.”

The other major trend is edge computing co-located with mobile base stations and points of presence. This allows the network operator to host a range of applications and services that benefit from low latency and reduce the amount of data traffic that needs to be sent back to the core network.

The strongly forecast demand for edge computing is one of the reasons why Dell Technologies has partnered with Orange to work jointly on a wide set of solutions that fall under the edge-computing and acceleration technology umbrella and ensure 5G delivers on its promises. The two companies will cooperate on defining and developing promising use cases, pilots and business models and seek out open source consortia and partnerships for the edge ecosystem.

But, critically, edge computing doesn’t have to wait for 5G networks – edge computing can be, and is being, deployed today. The radio access from the edge-computing platform to the device can be Wi-Fi, LTE or LPWAN, depending on the customer need.

“When it comes to edge and IoT, I don’t think any vendor or any partner can do all the solutions all by themselves. Partnerships are going to be critical,” explains Iyer. “Edge computing is an area of co-innovation. People don’t come and ask for just IoT or edge solutions – they’re looking for applications, business cases and value propositions. We have to explore these together.”

People who said all data would remain on mainframe computers lost out.

A new perspective

Creating an insight-driven enterprise around edge and IoT requires strong business and technical consulting. This is in addition to enhanced infrastructure and new technologies – such as AI, machine learning algorithms and virtualization tools – to harvest useful insight from the data in real time and over time.

“AI will not work without data. You need as much data as you can collect. Throwing data out means you lose insight. Enterprises need a lot more ways of keeping data,” explains Iyer. “5G means more data, so you need an efficient way to analyze it and work out what’s important, so you know what to store. And that capability is provided by edge computing.”

The future combines the edge and the cloud

A lot of data evaporates after use – such as IoT sensor signals that don't trigger alarms and surveillance images that get recorded over. Other types of data are vital to powering decision making over the long term. This data powers comparative analysis – where the correlations between different data sets are assessed to identify potential causality. It is also used for historic analysis to spot trends over time. For this very reason, the cloud isn’t going away any time soon. It will continue to rise in importance in tandem with the forecasted growth of the datasphere, which is anticipated to be ten times larger by 2025 than it is today.

“People who said all data would remain on mainframe computers lost out. While the cloud will continue to grow, we’re now seeing another radical architectural shift. But we won’t have one thing or the other – edge computing or cloud. Enterprises need both. We’ll live in a hybrid world,” explains Iyer.

Edge computing and IoT have enormous potential to create new business value, but as with any new technology, it can be complex. “The IoT wave is here. Working with partners to build the right edge computing and security architecture is now the next step towards the real-time enterprise,” concludes Iyer.

Read about our work at the Roland Garros tennis tournament to analyze CCTV footage for real-time crowd guidance using edge computing and AI.