A close up of a woman's hand using a mobile phone to check the price at the store - edge and fog computing shopping Concept
PHOTO: Shutterstock

According to a new IDC Spending Guide, worldwide spending on edge computing will reach $250 billion by 2024. Services will make up nearly half (46.2%) of all spending, with hardware next at 32.2%, and 21.6% to be spent on edge-related software.

Looking even a bit further, Gartner estimates that by 2025, 75% of data will be processed at the edge, outside of traditional, centralized data centers and the cloud, up from less than 10% today.

If these statistics are any indication, edge computing will likely play a huge role in improving the customer experience and could change the playing field in not just marketing, but in manufacturing, security, healthcare and retail.

Defining Edge Computing

Edge computing refers to systems in which the computing resources are placed closer to a device or user, in other words, at the edge of the network. Typically systems reach out to cloud services that are located at a distance from a user or device. By moving the data processing out to the edge, latency is reduced and the majority of data is now processed near the user or device.

Potential applications include self-driving cars, augmented reality (AR), in-store personalized marketing, artificial intelligence (AI), big data analytics, manufacturing, and IoT devices.

Here is one example, where edge computing is being used to bring computing power and data closer to the consumer and create immersive shopping experiences via AR and VR.

"AR- and VR-based shopping experiences are an example where edge computing is used. Combined with 5G and mobile cloud edge, data processing will move closer to the consumers instead of a remote cloud-based data center," said Janakiram MSV, principal analyst at Janakiram & Associates.

How Lean Data Processing, Fog Nodes and the Cloud Work Together

Cloud computing works well for applications that aren’t dependent on low latency like SaaS. For those applications that do require low latency like IoT devices, it’s more efficient to analyze the data in a location closer to the device itself. This real-time functionality must be relevant and reliable, according to Rohinee Mohindroo, head of technology solutions at Genpact.

“Edge computing uses lean data processing to make the most relevant decision at the edge, and then uses fog and cloud services for reliability,” she said.

Fog computing brings the cloud closer to the things that create and act on the data from IoT devices. The physical fog devices, called fog nodes, can be located anywhere with a network connection, such as along the aisle in retail stores, in the ceilings of factories, on telephone poles, in airports, and even in cars. The only requirement is storage, computing, and network connectivity.

The data collected by fog nodes is dispersed differently based on how quickly it needs to be analyzed. Data that is the most time-sensitive (i.e. within a second) is analyzed on the closest fog node itself. Other data that is able to wait longer, anywhere from seconds to minutes, is sent to an aggregation node. Data that is the least time-sensitive is passed to the cloud.

Edge Computing Challenges

One of the largest challenges for enterprises looking to use edge computing is the logistics of deployment and management. “Edge systems have to be plug-and-play. For example, think about that grocery store. They did not have an IT person to plug it in and turn it on, so deployments needed to be simple enough that a non-technical store manager can plug the thing in and turn it on,” said Collier.

Another challenge, according to Mohindroo, is training. “Moving AI capabilities to the edge requires moving not just data collection and processing to the edge, but also moving training and learning to the edge,” Mohindroo said. “But doing so will facilitate the necessary real-time processing for self-healing, self-optimized intelligent devices and also make voice-based conversations, and mixed-reality experiences more intuitive and secure.”

Edge Computing Examples

Jason Collier, analyst at Gigaom, said the autopilot feature in Tesla vehicles is an example of how edge computing works using IoT data and fog nodes. “Autopilot will drive you from point A to point B with no user input, but let’s say that an 18-wheeler starts to merge into your lane, and you slam the brakes and jerk the steering wheel,” he said.

“The 30 seconds before and after that user input, the edge device takes all of that IoT sensor data and uploads it to the fog. It then does further processing on the data and uploads it to the cloud where it is used on the next training iteration of the neural network that runs autopilot.”

By the time you get home and plug in your car, the data has propagated back down the stack and your car is now smarter as a result. “Now multiply that by the other 1 million user inputs that happened on that day,” Collier said “IoT could not exist on that scale without edge as a mitigator.”

In retail, the temperature sensors in refrigeration units offer another example where IoT and edge work in conjunction. According to Collier, there are typically 50 - 100 IP addresses for IoT devices in a store reporting temperature data. “That data is then processed at the edge to determine if it is in line with the norms, and when it is not in line, it is reported up the stack to the core/cloud. A prime example of this is tiered data processing, with only the data that is important or relevant being reported up the stack,” he said.

Edge is already being used to run the legacy applications that keep many retail establishments running and has allowed them to push boundaries of what they can do, Collier said.

“A prime example of this is a grocery store chain that was an early investor in edge. They use edge computing to run the store’s inventory control, point of sale, and security systems. This has also allowed them to develop applications that really push the envelope on customer experience (CX)."

The Benefits of Edge Computing for Customer Experience

There are many ways edge computing can be used to improve customer experience, especially when combined with other technologies like AI, machine learning and natural language processing. “Customer experiences can be enhanced by incorporating sentiment analysis, facial recognition, voice analysis, and location based contextual targeting,” said Mohindroo.

One customer experience application is a grocery list app created for mobile devices. “While each one of the stores is laid out differently, the advantage with edge computing is that the app can rearrange the items on the list to give the customer the best path to get all of those items,” Collier said, adding the store can use that capability to route the customers by promoted items.

Edge Computing Plus AI Equals Real-Time Personalization

Customer expectations are higher than ever due to the hyper-personalization they routinely experience with the familiar tech giants. Using traditional approaches, it’s difficult to provide real-time interactions however by combining edge computing with artificial intelligence, brands will be able to provide real-time personalization and access and analyze in-depth customer data and behavior analytics, providing actionable insights for up-to-the-minute personalization.

According to Janakiram MSV, edge computing and AI are a marriage made in heaven because machine learning and AI are all about real-time decision making. “By moving AI to the edge, the results are delivered much faster. AI models have two phases — training and inference. While training continues to happen in the public cloud, inference — the mechanism of making predictions from unseen data — is moving to the edge.”

He said edge computing is a way to ensure exceptional customer experience extends to the brick-and-mortar world. It brings the level of personalization we see in online shopping to physical shopping. “In a multi-branded store or a mall, edge computing can calculate how much time you spend at each brand’s outlet and provide a recommendation through a contextual online ad. The combination of AR/VR and edge computing is set to revolutionize the consumer experience,” he said.

The amount of data that is currently produced is already vast, and a recent report from IDC indicates that by 2025, every digitally connected person in the world will have some type of digital data engagement over 4,900 times per day, and that IoT devices will produce over 90 zetabytes) of data each year.

Additional benefits of the intelligent edge, as AI-enabled edge computing is referred to, include offline applications, since there is no internet requirement for edge-to-fog node communications.

IoT and Edge Computing , Not Just for B2C Brands

Edge computing also has applications in the security, manufacturing, communication, healthcare, and transportation industries. Security services use intelligent edge-powered facial recognition applications in airports, subways, and train stations to scan for known terrorists and individuals that are wanted by law enforcement.

Factories are also using edge devices to continually monitor products for defects and analyze real-time data to create predictive trends that preclude from happening in the future. It’s also being used to connect real-time data from the factory floor to its enterprise resource planning (ERP) system. “I see edge as a layer above IoT that enables IoT devices to integrate with the core data center and cloud,” Collier said.

Wrap Up

Edge computing is relatively new, but given the potential applications and opportunities for retailers, marketers and manufacturers it looks bound to grow. For brick and mortar stores, it can enable a consistently exceptional customer experience, and provides personalization, marketing, and logistical solutions for many other industries.