4 cases, a whole article of valuable information! Make your edge computing not "edge"

Edge computing means different things to different users. One thing that remains constant, though, is that location matters.

With edge computing, autonomous mining equipment can respond to unexpected situations a mile underground even if it loses connection to the network. When a hotel guest places an order on their phone and wants it delivered poolside, edge computing can direct it to the server closest to the guest.

Currently, the number of sensors, smart devices, and mobile users is proliferating across industries. In response to the increasing amount of decentralized data that needs to be processed locally, companies are beginning to invest in edge computing. Low latency is critical for many applications, and edge computing eliminates the time it takes to move data to a data center or public cloud for processing.

Another factor is the cost of backhauling data to data centers. Performing processing and analysis locally reduces the need for expensive bandwidth.

[[380764]]

In the field of immersive interaction, edge computing can bring the generation of real-time data closer to humans and machines. According to Gartner data, by 2022, more than 50% of the data generated by enterprises will be created and processed outside of data centers or clouds.

Edge computing deployments are not isolated. Edge infrastructure is usually centrally managed through enterprise data centers or the cloud. Edge services can be delivered through content delivery networks, hosting providers, and cloud service providers. These service providers are currently expanding their businesses through computing power on regional data centers, local micro data centers, and telecommunications towers. The main task of enterprise IT is to integrate all components such as sensors, mobile devices, access points, gateways, edge servers, storage and network devices, and manage operations that extend from the edge to local and cloud data centers.

Gartner believes that in terms of the use of edge computing, enterprises currently tend to start with a single simple use case led by a seller or system integrator. After a few years, enterprises will begin to try various edge computing use cases. Gartner predicts that by the end of 2023, more than 50% of large enterprises will deploy at least 6 edge computing use cases for the Internet of Things or immersive experiences, a figure that is in stark contrast to less than 1% in 2019.

Here are four companies in different industries talking about their forays into edge computing and their experiences with employee safety, productivity, customer service, and revenue.

Edge computing accelerates

Artificial intelligence maintains the speed of freight trains

Idle time is the enemy of freight train operations. Scott Carns, COO of Duos Technologies, noted that railroads measure freight train performance by average speed, which is typically between 35 and 37 miles per hour. "Every mile higher in average speed equates to $10 million more in profitability. As a result, people are investing heavily in technologies that can improve efficiency."

One factor that slows down freight trains is inspection. In North America, when freight trains cross rail lines, operators must take time to perform mandatory inspections on vehicles. Historically, workers have been required to perform visual and physical inspections of mechanical parts on trains parked in rail yards. To streamline the process, Duos began using sensors, imaging equipment, and analytical tools deployed in the yard.

[[380765]]

Duos is building multiple rail inspection gates, which are like large carports. Trains passing through Duos inspection gates are automatically inspected without having to slow down or enter the inspection yard for physical inspection as in the past. The inspection gates are equipped with high-powered LED lights that illuminate the trains and cameras that can capture images at high speed.

“We build these large inspection gates that trains can go through at 40, 50, 60 miles per hour,” Carns said. “As the train goes through, we image the entire train from every angle and then run an AI algorithm through it to look for leaks or mechanical defects. The system can detect oil leaks, broken parts, and missing doors and indicate what needs to be fixed.”

The system can observe each train car from nine different camera perspectives. Each camera generates about 1GB of data, so a train car generates about 9GB of data. An average train has about 120 cars, so the amount of data generated by each train is about 1TB. Because many tracking sites across the United States are located in remote areas, the available WAN bandwidth varies, and the huge amount of data makes it impractical to send the data to a central data center for processing.

Adjacent to the Duos detection door is the IT system room. Each room houses two 45U racks with Dell XR2 servers and PowerVault storage. "The idea is to build a small data center next to the track to process all the data at the edge," Carns said.

To comply with rail safety regulations, inspections must be carried out quickly so that repairs can be made when the train arrives at a nearby station. "We have to do our inspections in near real time because the trains have to be inspected before they reach their destination."

Additionally, depending on the location of the gates, there may be 30 to 40 trains passing through each day. The processing must be completed before the next train passes. These data-intensive operations are completed quickly at the trackside.

Duos migrated from Dell's PowerEdge VRTX blade servers to the XR2 about 18 months ago because the XR2 platform makes the system more modular.

It is also important to give the servers additional reinforcement considering the environmental conditions they must withstand. For example, in Winnipeg, Canada, one of the locations where Duos inspection doors are deployed, the temperature can rise to 90 degrees in the summer and drop to minus 40 degrees in the winter. One of the biggest challenges in the railway industry is maintenance and maintainability. In addition to military and marine applications, it is railways. Duos also plans to continue to expand the construction of its inspection doors. In the next few years, the company expects to deploy 60 or 70 sets.

On the development side, the company will also continue to expand the scope of its AI applications. Recently, Duos began using thermal imaging technology to capture and analyze models of the operation of electric traction motors installed under train cars at different temperatures. The move is intended to improve predictive maintenance for these expensive parts that are prone to damage and burnout.

Location-aware apps make hotel rooms smarter

Performing a lot of operations at the edge can sometimes reduce the amount of hardware. At Nobu Hotels, upgraded wireless infrastructure is supporting more IoT and AI-driven applications with less proprietary hardware.

Nobu is currently standardizing its hotel network infrastructure to enable Wi-Fi access to be integrated with guest and employee applications, including content delivery, lighting control, door locks and security alarms. This will allow Nobu to simplify its on-site hardware requirements, saving time and money, said Rodney Linville, global IT director for Nobu Hospitality.

“To make it easier to open the hotel and integrate high technology into the hotel, I needed to make sure the technology I chose could do that,” Linville said. “We wanted to do that now, not in the future.”

[[380766]]

To this end, Nobu began deploying Aruba equipment in three new hotels in Chicago, Warsaw and London. The equipment uses Aruba's edge and core switches, ClearPass Policy Manager for role-based and device-based secure network access control, and location-ready access points that support IoT devices running WLAN, Zigbee, Bluetooth and third-party protocols.

The network infrastructure supports management applications such as mobile POS, property management systems, and visitor applications for the hotel. In addition, they can provide technology such as door locks, temperature control, and hotel services for hotel rooms. Guests can use their mobile devices to check in, check out, order room services, request bedding changes, or raise curtains. Employees can monitor room access, reprogram door locks, or handle guests' requests for help in emergencies.

“Being able to connect anywhere in the room and be able to control the services in the room is what we’re looking for in the guest room,” Linville said. Nobu’s goal is to allow guests to use their own phones or wireless devices to manage the hotel’s services and control different systems in the room (such as door locks, TV or air conditioning).

To achieve this, IT teams are pairing Bluetooth beacons and Zigbee gateways in Aruba access points with IoT and analytics applications that can generate, analyze and act on data in real time.

In addition, Nobu has begun integrating Aruba's infrastructure with access control technology from ASSA ABLOY Global Solutions. ASSA ABLOY technology allows guests to check in online and unlock their room's door using their smartphones. In the past, Nobu had to deploy a dedicated indoor gateway for the ASSA ABLOY contactless entry system. With the integration, Nobu can use Aruba access points to secure communications between indoor smart door locks and centralized door lock management software, rather than relying on a separate Zigbee gateway.

Linville asked ASSA ABLOY to replace the dedicated gateways with Aruba hardware. “These dedicated gateways almost set up another network within the hotel to control the guest room door locks,” Linville said. “We didn’t want to pay for this technology when we already had it in place at the hotel.”

Another key project is to use Aruba devices with third-party employee safety applications. React Mobile provides portable panic buttons that can alert security personnel at any time when hotel staff encounter dangerous or threatening situations such as guests being threatened or medical emergencies. The alarm system integrates location-based services, which in the past required IT departments to deploy and maintain as a separate network and dedicated Bluetooth beacons installed in the rooms.

Through system integration, Nobu can use Aruba IoT-enabled access points to provide the necessary connectivity. If an employee presses the “help” button on the React Mobile smartphone app, a Bluetooth beacon in the Aruba access point will use the app to determine the user’s location and send a request for help over Wi-Fi.

At Nobu’s request, multiple vendors worked together to integrate their systems in time to open the Chicago hotel. These deployments require trial and error, but the effort pays off. With the experience in Chicago, the deployment of the Warsaw hotel facilities became very simple.

Aruba also has a willingness to invest in R&D and explore AI. “One of the reasons we chose Aruba is because they were as eager as I was to do these types of integrations,” Linville said. “They saw the value in it.”

Another consideration for Nobu to use the Aruba platform is security. Aruba devices integrate role-based firewall and intrusion prevention capabilities. These features help Nobu make its mobile POS application compliant with payment card information (PCI) data security requirements. Other enterprise vendors can also provide this security, but they need to add new equipment to the network.

Hyperconverged infrastructure at the edge simplifies IT work

A demonstration of a hyperconverged infrastructure (HCI) platform many years ago caught Jeff Miller's attention, but the scale of the system was a big problem. Miller said: "I personally think that the demonstration was perfect, but the scale I need is not that large, 10 times smaller." For this reason, he has been in contact with the supplier Scale Computing and has been closely following the development of HCI until the scale of use can meet his needs.

Miller is the IT director for Jerry's Foods, which operates more than 50 retail stores in Minnesota and Florida, specializing in groceries, liquor, and hardware. In the past, each store had a virtual server and SAN, but these legacy systems were difficult to manage and expensive to maintain. Today, Jerry's Foods is nearing completion of its deployment of Scale Computing's HE150 edge device project. This move has enabled the company to restructure its entire IT strategy.

[[380767]]

Built on the Intel NUC mini PC, the HE150 is an all-flash computing device using NVMe storage that is about the size of three smartphones stacked together and features include disaster recovery, high-availability clustering, rolling upgrades, and integrated data protection.

On-site processing is critical for some core store applications. For example, product information needs to be managed on-site. These stores have 65,000 to 70,000 merchandise items. These items need to be tracked in a database that is constantly updated as suppliers make changes to their products. POS transactions rely on access to data that includes UPC codes, supplier numbers, product categories, and prices. All of this must reside locally and be viewable locally.

Another important field application is the Electronic Benefit Transfer (EBT) system. To verify the availability of funds, the system requires a cross-link between the local POS application and a system run by the U.S. government. Because it is clustered, the HE150 infrastructure is highly available. After the EBT application is migrated to the HE150 infrastructure, the infrastructure pings the U.S. government's system to confirm the user's account status. If shoppers cannot use EBT funds, they will choose to shop elsewhere where EBT funds can be used. This can have a significant impact on Jerry's Foods' sales performance. "For example, if the Lake Street store in Minneapolis could not use EBT, we could lose $40,000 in less than an hour," Miller said.

The current workloads in the store are not heavy in terms of computing and intelligence, mainly due to social unrest in the United States. During the riots in Minneapolis this summer, a branch was severely damaged. Jerry's Foods is currently rebuilding it, including strengthening the building security system, which will put more burden on local IT resources. Elsewhere, Jerry's Foods is considering installing a solar roof to reduce energy costs, and if built, the air conditioning system in the store will be intelligent.

With edge devices, Jerry's Foods can customize the applications that run at each location. In stores where edge processing requirements are higher, Miller plans to deploy large HCI appliances from Scale Computing. Currently, the decision on how large to use depends on how much video needs to be stored locally. "Whether it's big or small, it's still a win for us," Miller said.

Overall, the Scale Computing appliances reduce the amount of time IT spends managing in-store infrastructure because fewer resources are used. Jerry's Foods expects the deployment to cut costs per instance in half over five years compared to traditional virtualization systems. "The great thing about Scale is that all the old infrastructure is gone," Miller said. "We can get rid of VMware completely." At the same time, many dedicated hardware appliances will also be eliminated. In the past, all branches had to maintain their own dedicated hardware for certain applications.

Some of Jerry's Foods' stores will also be switching to the new edge infrastructure model, which will be completed next spring. Miller's next step is to consider connecting cloud storage with applications running at the edge.

IT/OT Convergence Takes Shape in Mining

Workers at Swedish mining and smelting producer Boliden are extracting copper, zinc, lead, gold and silver, but they are doing it from the comfort of a surface control room rather than underground in tunnels.

At the Boliden site in Gapenberg, Sweden, autonomous mining equipment is operating in areas that are not suitable for sending workers down into the mine. These areas are filled with dust and vibrations, and are prone to water infiltration and toxic gases. In this environment, communication equipment also cannot work properly.

Automation has been changing the mining industry for years. Sandvik Mining and Rock Technology built the first autonomous mining equipment as early as 2004. Today, equipment that operates completely autonomously in underground mines is the most advanced equipment available. "The onboard computer can independently control the machine without human intervention," says Petri Mannonen, Product Line Manager for Mining Information Management at Sandvik, which provides equipment and services for operations such as rock drilling, crushing, screening, loading, stowing, hauling, excavation and demolition.

A major driver of modernization in the mining industry is employee safety. Because underground mining is a dangerous environment, keeping workers out of harm's way improves safety.

[[380768]]

Another key goal is to improve operational efficiency.

As mining technology advances, the need for network connectivity is increasing. It’s not just laptops running underground, machines with network connectivity are critical to operations. As mining becomes smarter, connectivity is expanding from small dedicated networks to mine-wide networks. With greater connectivity coverage, mining companies can detect and keep people, equipment and other resources at a safe distance. In emergencies, high-reliability connections can help with communication coordination and response.

Sandvik provides mining equipment and operating software for Boliden's Garpenberg mine. These software can be connected to Cisco's Industrial Internet of Things IoT network via wired or wireless. Sandvik's AutoMine software enables autonomous and remote control of underground mining equipment, and engineers in the control room can track resources and make adjustments as needed. In addition, Boliden also uses Sandvik's OptiMine software to collect and analyze data from mining equipment to optimize production.

To ensure the safety of the environment and equipment, a strong and reliable communication network is required from equipment to control room to security solutions. To this end, Boliden uses a low-latency IoT network built with Cisco's industrial switches and access points.

It’s a requirement that the network stays connected underground. Dave Wilson, managing director of global sales for Cisco IoT, says this connectivity has helped transform the mining industry, and Sandvik’s business model in particular.

“We can now provide connectivity in harsh environments, so that devices can connect to the network. This allows Sandvik to rethink its business model and grow it further,” Wilson said. “Connectivity also makes vehicles smarter. With the addition of tools, sensors and software, the vehicle becomes a large sensor that can analyze what is happening. Users can optimize entire processes based on the analyzed data.”

Although the Garpenberg mine is located a mile underground, engineers can track the status of personnel and equipment in real time. Onboard sensors can help with navigation and track the health of equipment to ensure it is operating optimally. The data collected can also help make business decisions.

These capabilities are revolutionizing mining operations. Self-driving trucks can drive autonomously without the “driver” getting tired. These vehicles can navigate mines with centimeter-level accuracy and go to places that have never been mined before.

Many of the calculations are done independently at the farthest edges of the mine. Because the equipment can work in areas with poor network coverage, much of the analysis is done onboard vehicles.

More advanced analysis is done on the ground. Typically, IoT data collected by equipment is either sent to a local central repository where the data is stored, or to the cloud for analysis. With the development of edge computing, the trend of integrating IT systems with operational technology has been highlighted in the mining industry.