Industry 4.0: Rise of the Edge

Cloud and Edge computing are distinct but complementary.

The term Edge can be used for a variety of use cases – Consumer-centric, Commercial, or industrial settings. This article is the third of a 4-article series, and primarily addresses manufacturing and industrial operations, including smart factories, connected workers, intelligent supply chain, video surveillance – all in an industrial setting. These principles also extend to smart utilities, smart grid, industrial aspects of Retailers (e.g., refrigeration), etc. However, commercial IoT use Cases, such as smart buildings, retail stores, autonomous cars, or set-top boxes are out of scope for this article.

What is Edge?

The industrials are a massive generator of data – equipment data (time series), process data, and people activities and interactions as transactions. However, most data go to waste — as data is either not captured or captured but not analyzed. The cloud is no longer sufficient to instantaneously process and analyze the troves of data generated — or soon to be generated — by IoT sensors and devices, smart infrastructure, connected cars, and other digital platforms. Sometimes faster data processing is nice-to-have and, at times, it is mission-critical. With so many more devices connected to the internet and generating data, cloud computing will not be able to handle all these end-points. Edge computing offers an alternative to cloud computing with applications at ‘the edge' and may extend to edge analytics and edge AI.

The Edge and the Cloud need to work together for speedy connectivity for the specific use case(s), quality of service (QoS), and service levels (SLAs) as outlined in Exhibit 2 below. Although a relatively new(er) space, edge computing offers some obvious benefits, including:

1. Real-time or faster data processing and analysis: Data is processed closer to the source, not in an external data center or cloud, which reduces latency, lag time. Edge servers and gateways will fill the critical role of intermediary processing node and integrator between multiple “cores” (hyperscale cloud or enterprise-owned data centers) and a diverse set of people, devices, and workloads at the edge.
2. Lower costs: Enterprises spend less on data management solutions for local devices than for cloud and data center networks.
3. Less network traffic: With an increasing number of IoT devices, data generation continues to rise at record rates. As a result, network bandwidth becomes more limited, overwhelming the cloud and leading to a greater bottleneck of data. 5G is also poised to support IoT at scale.
4. Increased application efficiency: With lower latency levels, applications can operate more efficiently and at faster speeds. The technical evolution of edge will follow the evolution of application development, delivery, and operations, relying on repeatable and reusable microservices at scale.

Industrial IoT

Within the industrial IoT (IIoT) context, also known as Industry 4.0, edge processing involves process integration with multiple technologies – Robotics, automated guided vehicles (AGV), unmanned aerial vehicles (UAV/Drones), IoT, Machine Learning (ML), computer vision, image processing, speech recognition, NLP, content caching & distribution, augmented/virtual reality (AR/VR), data hyper-automation, location-based services (mapping & Geospatial sciences), personalization & targeting, etc.

Edge computing occurs close to data sources, e.g., gateway or a device within points in a network mesh. The primary benefit of edge computing is reducing the risk of network outages or cloud delays — that are attributed to sub-optimal, high latency — when highly interactive and timely experiences are critical. Edge solutions decrease Cloud usage and costs, and it enables these experiences by embedding intelligence and automation into/near the physical world. Edge computing data feeds into local AI models, termed ‘edge AI,’ to make low-latency local decisions. The inferencing at the Edge starts with bringing together data for experimentation and AI model training, and that takes a lot of decentralized, local computing. For a holistic view across the industrial enterprise, the Cloud remains the best solution to combine edge, enterprise applications (e.g., ERP, CRM, SCM, etc.) and third-party data for discovery and AI model creation.

IIoT integrates IoT and related technologies into the industrials, which will revolutionize manufacturing and supply chain processes as depicted in Exhibit 1 in smart factories. As IIoT rapidly evolves, there is a dramatically increasing number of devices that impose high demands on the existing cellular network, with the expectation to support the ubiquitous connections from both critical and non-critical IoT devices.

To reach this expectation, massive machine-type Communications service has been standardized in the 5G Radios. Beyond 5G, ultra-reliable low-latency communications are envisioned to move to 6G to support massive mission-critical IoT devices with 1-10ms latency target. The?speed and precision that 6G allows could enable more advanced technologies and robotic procedures?in industrial use cases like specialized manufacturing, robotic surgery, and robo-military actions. 6G will create a negligible amount of lag between queries and data processing needs, which is especially helpful for digital and holographic imagery. Optimizing the operation of the cellular network to meet the diverse quality-of-service (QoS) requirements is challenging in a large-scale IIoT environment, where a collection of heterogeneous devices is geographically distributed.

Massive non-critical IoT Devices

• Field Sensors: To monitor and support manufacturing, an increasing number of sensors and actuators will be deployed in the smart factories to guarantee safety. Although the data rate of a single field sensor is quite low (e.g., 10kbps), the collision from the simultaneous massive access may lead to severe access delay.
• Asset Tracking: Accurate locations and inventory status of the assets are critical for operational efficiency and safety. According to requirements in the 5G evolution, we need to provide centimeter-level positioning for a large number of devices in IIoT.
• Video Surveillance: Video surveillance is widely adopted in the factory to guarantee warehouse security, and automation process safety, which generates Gbps-level uplink traffic from hundreds of connected cameras. Thus, the uplink enhancements and techniques are critical for this IIoT use case.

Massive Critical IoT Devices

In Industry 4.0 settings, supporting ultra-reliable and low-latency communications is a key prerequisite for use cases, such as:

• Vehicles: As the typical mission-critical applications, Vehicle-to-everything (V2X) and AGV communications improve the asset transportation efficiency in outdoor and indoor environments, respectively. Unmanned aerial vehicles (UAVs) are important in applications, such as aerial monitoring, package delivery, etc.
• Integrating Augmented Reality (AR) and IoT is to provide real-time control of real-world IoT devices and virtual-world objects in a virtual environment. Apart from the latency and reliability requirements, AR applications need high computational capability and sufficient bandwidth to provide an immersed experience to users.
• Robot Motion Control: Due to the lack of flexibility in robots, human knowledge is indispensable in complex manufacturing processes, where robots respond to human instructions. Due to the fast-changing environment, stringent latency and reliability requirements should be satisfied to guarantee the effectiveness of real-time decision-making.

Typical Quality of Service (QoS) requirements in terms of latency, reliability, data rate, and connection density, depends on the Use Cases (discussed above) are outlined in Exhibit 2 below.

Connected Ecosystem: Enterprise Architecture

At the enterprise level, the connected ecosystem is comprised of numerous participants. While there is a lack of standard architecture, a 4-tier ‘simplified’ architecture model – sensing devices, Edge, Cloud, and the IOT platform(s) – is emerging as depicted in Exhibit 3 below.

Organizations may need configuration management and orchestration systems to manage and move across these tiers dynamically.

1. The Intelligent Edge: Industrials are increasingly integrating artificial intelligence (AI) and machine learning (ML) into edge use cases, creating new data orchestration requirements. A well-architected 4-tier computing architecture model can unlock the potential for enhanced data security, privacy and trust, improved latency, and expanded bandwidth.
2. Innovation: Communications (CSP) and tech organizations continue to introduce new cloud, edge, and networking solutions while businesses explore edge-native applications. This innovation further blurs the boundaries across the 4-tier architecture, and harnesses the innovation potential with business cases that require proximity, bandwidth, and latency around a mobile/platform sensing ecosystem.
3. Implementation challenges and solutions: Smart factories, utilities management, and connected vehicles demonstrate the potential and complexity of the 4-tier architecture model. For ease of implementation, organizations can partition data and workloads across architecture tiers.
4. Cross-tier optimization: A strong Edge business case considers ecosystem computing needs to determine what belongs on premise, on mobile, on the edge, in the cloud, and across sensing platforms. As such, organizations will pay a premium for proximity, latency, bandwidth, performance, scalability, and other technical benefits of edge computing. However, one must factor-in human costs to manage the solution while calculating the break-even point.

Hyperscalers extend to the Edge

The 3 dominant Cloud players – Amazon, Google, Microsoft – are also emerging as edge computing leaders. Amazon with AWS Greengrass (2018) service extends AWS to devices so they can “act locally on the data they generate, while still using the cloud for management, analytics, and durable storage.” Microsoft’s Azure IoT Edge solution extends cloud analytics to edge devices and can be utilized offline. Azure is looking at AI applications at the edge (Edge AI). Similarly, to extend GCP to the edge, Google launched two new products to improve the development of connected devices at the edge: hardware chip Edge TPU and Cloud IoT Edge, a software stack.

About Cybersecurity

Cybersecurity must be both preventives to avoid threats and responsive to threats.

Enterprise adopters can get enormous value from reducing transaction costs by connecting assets. The trade-off is that connecting assets also provides a potential virtual (or physical) breakpoint, i.e., a larger attack surface and more risk when assets are connected. IoT solution providers and enterprises need to work together to develop security that strengthens and protects breakpoints, as well as enables rapid detection and mitigation of security breaches. Experts agree that 100?percent defense of the perimeter is impossible, but it matters as much or more whether a security breach will result in bringing down one machine for an hour or the entire electrical infrastructure of the Western United States for days. While there is a lot to be done to protect the extended attack surface, fortunately, technology is advancing to provide greater levels of security.

One of the interesting developments is the ability for homomorphic processing to perform real-time algorithms on encrypted data, alleviating the need to de-encrypt and re-encrypt data, making cybersecurity purely an encryption problem. According to Global Industry Analysts, the homomorphic encryption market is forecast to grow quickly, given the escalating need for tightened security measures at each tier of data transmission across public networks, e.g., Internet and cloud-based services of smart computing and connected devices.

Conclusion

In conclusion, infrastructure and operations leaders must focus on Edge + Cloud as an integrated network, compute, storage, and AI & analytics system. Some of the best practice recommendations:

• Raise the profile of edge across applications, technologies, and business units by establishing the position of an Edge Architect or an edge innovation team, with representation from the relevant business units. At times, the edge architect could be an extended role of an Enterprise Cloud Architect.
• Develop a business case framework for mapping the benefits of edge to the business outlining goals and priorities, edge use cases, functional capabilities, risks, and associated risk-mitigation tactics.
• Prioritize projects by performing 2-3 bounded proofs of concept, using repeatable patterns or an open framework to avoid wasting effort on projects with little chance of success. [Note: The business case must be a cost/benefit analysis and the return on investment for a scale rollout.]
• Conduct an application and infrastructure inventory focusing on the 4 imperatives of edge computing outlined above – faster processing, lower costs, less network traffic, and increased application efficiency.
• Choose between building internal edge competency or external edge as a service (EaaS) implementation focused on business outcomes and SLAs by consulting with system integrators or ISVs with expertise in edge in their vertical industry to determine which is more appropriate for the organization.

The Edge is still a relatively new field when compared to a decade-old Cloud. Therefore, the PoCs don’t quite look like ‘scale implementations’ with holistic Business Case. Also, the architectural standards and implementation tools need to advance. In parallel, for scale implementations, there is a need to rev up the talent supply.

Over the next 3-4 years, we’ll see a lot of innovation in the Edge across computing, network, storage, and AI and Analytics dimensions, as well as distributed device and edge orchestration and management.

What do you think??

Note: For the 2 other articles of this 4-article series on Transforming Industrials unleashing the Power of Data, please click the corresponding links below:

Article #2:?Data Strategy in the Cloud Era

Article #4:?Opportunities and Barriers in realizing Industry 4.0