The Edge Computing Paradigm in Manufacturing
It is estimated by Gartner, the global technology Research and Advisory leader, that by 2025, 75% of digital data will be produced and processed outside the traditional data center or cloud.
What is Industry 4.0 or the Fourth Industrial Revolution or 4IR?
Let us look at the evolution of the industrial world. The First Industrial Revolution witnessed a fundamental change in how industrial production happened – steam powered machines emerged, which raised industrial production levels. This happened between approximately 1760 and 1840.
During the Second Industrial Revolution, another productivity dimension was added to industrial manufacturing – railroads made personal and industrial transportation faster while electricity began replacing steam, which pushed production still higher, and telegraph networks revolutionized communications. This was circa 1870’s to around 1915.
The Third Industrial Revolution, also called the Digital Revolution, started after the second world war. Industrial automation and digital computing emerged which meant faster business and industrial production. The Fourth Industrial Revolution which is currently underway, introduced new ways of computing and manufacturing – AI/ML, automation, robotics, IoT, Edge Computing became mainstream and technologies such as 5G made them more practical and applicable to real-life situations.
What is happening in Industry 4.0?
The world of industry could be broadly divided into the physical world – mechanical, digital, and electronic machines, the biological world made up of human beings, and the digital world that stores and processes the data produced by the other two worlds. The technologies mentioned under the 4IR – AI, ML, Robotics, IoT, 5G, machine-to-machine communication, etc. are enabling the blurring of lines between these disparate worlds.
Need for Edge Computing
According to industry analysts, it is expected that the number of IoT devices will exceed 55.7 billion by 2025 leading to the production of 70 zettabytes of data generated by these connected IoT devices. This will see an explosion in computing resources – devices and power and storage – needed to manage this data explosion.
Data, that enables business decision-making and supports real-time business operations, drives today’s business world. It is expected that huge amounts of data will be collected from sensors and IoT devices that operate in real-time from the remotest parts of the world. From personal devices to connected automobiles to factory sensors, the quantum of data is enormous, yet most of this data is not utilized at all. For example, a McKinsey study found that an offshore oil rig generates data from over 30,000 sensors of which less than one percent is currently used for decision-making.
This has exposed the current methods of computing – centralized data centers or cloud platforms – as being unsuited to processing these floods of real-time data. Bandwidth limitations, latency and network unpredictability are the limiting factors. Speed of computation is hit, and results are not meaningfully utilizable for real-time processes.
The solution – Edge Computing
Enterprises are leveraging edge computing architectures as a response to these data explosion challenges.
With edge computing, data is partly stored, processed and analyzed closer to the point of creation – a more efficient solution to issues of latency and bandwidth. Since data does not have to travel over the network to a cloud or data center for further processing, latency is drastically reduced. Add 5G networks to the mix and you get faster processing and more comprehensive utilization of data, thus enabling for deeper insights, faster response and enhanced user experiences.
In summary, cloud computing leverages big data while edge computing operates on “instant data” – real-time data generated by sensors or users.
What is Edge Computing
Edge computing is a distributed computing paradigm that brings enterprise applications and computation and data storage closer to data sources such as IoT devices or local edge servers. It is an architecture rather than a new technology.
Edge Computing isn’t exactly new. Content distribution networks of the late 1990s that served web and video content utilized edge servers deployed closer to end-users, in multiple locations, over multiple Internet backbones.
In principle, edge computing shifts part of storage and computing resources out of the central facility closer to the source of the data. Thus, what gets transmitted to the central location are the results of processing and analysis – real-time business insights or equipment maintenance insights – which now happen at the “edge”, performed where the data has been generated – a retail store, a factory production line, or a smart city.
Thus, new, real-time experiences are created for the users, while confidential data is retained closer to the source and cost of data transmission is brought down. Processing now happens at higher speeds and increased volumes.
Adding an edge to Edge Computing through AI
Edge AI embeds AI functionality into Internet of Things (IoT) endpoints, gateways, and other devices data at the point of use that are producing the data. When AI acts on data at the Edge, it reduces the need for centralized computing power. AI in Edge Computing enables augmented and virtual reality, autonomous operations, digital twins, smart factories, etc.
For the enterprise, this provides clear opportunities for differentiation, creates innovation capabilities, operational excellence, and new methods of customer engagement, as compared to the standard edge computing architecture.
In 2017, Ford and Domino’s worked on a project for a delivery system involving pizza delivery using autonomous vehicles. Vehicles are GPS-enabled, which lets them get tracked in real-time to inform the customer about order progress and send a retrieval code. All data points, location, anticipated timeframe updates are edge AI enabled. Thus, key insights into customer experiences and preferences were made available. For Ford, this meant promotion of the concept of self-driving cars and hence sales, while for Domino’s, this translated into increased operating efficiency and better customer experience.
Edge Computing and 5G
5G has matured and is entering the real world. It promises to make the Edge more real and reliable. It is creating new and huge opportunities in every industry vertical. It helps bring computation closer to where the data is generated, thus also allowing for the data to be stored near source. This enables better data control and reduced processing costs, faster analysis and insights and hence actions.
It is estimated that by 2025, 50% of enterprise data will be processed at the Edge, and this would be made possible in large measure by technologies such as 5G.
Edge Computing – application areas and use cases
Bringing to life applications that require low latency is one of edge computing’s main value propositions. For example, processing at the edge reduces lag in virtual reality systems used in field situations – disaster sites or battlefields. Similarly, in environments with weak or unreliable connectivity, such as an offshore wind farm, or an oil refinery with geographically distributed assets with smart sensors / devices, edge solutions could be of help. Edge enables faster data processing, enabling robots and sensors to make real time decisions and complete tasks smarter and safer in mission-critical and remote applications.
Other use cases and examples:
- Use of AI and ML for automated tracking and management of remote facilities, to aggregate sensor data and collate with other sources of information, to provide deeper insights into changes in the work environment.
- The United States’ Federal Emergency Management Agency (FEMA) has used edge computing for disaster response. It can set up a portable satellite-based network and use it to collect visual data via drones before sending in human teams.
- Edge solutions used to fight forest fires more safely and effectively by tracking equipment and teams in remote locations, enabling integration with existing firefighting resources, and offering insights and recommendations.
- Autonomous vehicles – a convoy of trucks could travel behind one another thus saving fuel costs. A driver would be needed only in the first vehicle, while all others would be able to communicate with each other using ultra-low latency.
- Hospital patient monitoring – various health parameter tracking devices such as blood pressure meters, glucose monitors, etc. are generally not interconnected, or if they are, enormous quantities of unprocessed data from these devices would be stored at a cloud site or data center. An edge server at the hospital location could help retain and process data locally, making for better data privacy and faster response times via real-time notifications to clinicians.
Advantages of Edge Computing
With Edge Computing, reduction in latency and breaking the limitations of poor bandwidth are the most visible benefits. This leads to clear business benefits: faster processing, improved response times, tighter and real-time control over physical assets, richer user experiences.
The objective is to strike a balance between speed and volume of data processing and location of the data. Data that calls for instant processing is stored and processed in devices at the very edge. Where deeper processing (such as analytics) is required to take advantage of cost models and high-level cognitive applications, data is moved and processed in the public cloud. Sensitive data can be retained closer to its source, helping meet confidentiality norms.
However, the biggest opportunity edge computing creates is to bring analytics, AI and ML to the network edge. It helps unlock the potential of the large volumes of data that is created by connected devices. New business opportunities can be discovered, operational efficiency can be raised, and users can be given enhanced experiences.
The emergence of Fog Computing
Fog computing is yet another variant of the decentralized computing paradigm. It creates a new layer between the Cloud and the Edge, bringing the Cloud closer to the Edge. In application areas where the number of IoT devices and the Edge computing devices is huge, and instant action or response is key to success, Fog breaks down potential latency between the Edge and Cloud. Fog might also be introduced for security and compliance reasons.
The terms Fog Computing and Edge Computing are at times used interchangeably because both involve bringing intelligence and processing closer to where the data is created; the line of distinction is hence blurred. Connected devices such as drones, smartphones, smartwatches, fitness bands, security monitors, home automation, industrial automation, weather sensors, etc. could be considered as either Fog or Edge.
As an example, consider a smart city where data is used to track, analyze, and optimize public services – transit systems, utilities, etc. Limited Edge deployment locations would just would not suffice to handle the resulting numerous, unending, and large streams of data. Hence Fog nodes could be deployed to collate, process, and analyze data that needs instant action, while meta-level data could be moved to the Cloud for deeper, long-term processing and use.
Looking forward
Even though cloud-based computing is ideal for many application scenarios, Such as data warehouses,
there are some inherent limitations with the architecture – it depends on networks to transport the data. Consider a computer that is making an important and instantaneous decision – guiding a self-driving car; if the network fails or is slow, that would be an acceptable or even dangerous situation. Similarly for medical monitoring, or natural disaster response. Moreover, moving large quantities of data is highly energy-consuming and hence costly. Hence Edge Computing will be around and have a key role to play going forward. It has a valuable role to play in health management, fighting climate change, managing cities and streamlining industrial production. Though it will certainly evolve as technology evolves. It will strengthen as advanced technologies such as 5G, satellite mesh, AI, ML, and Blockchain improve. It will become more real-time as processors become more powerful and storage becomes cheaper.
For example, as medical devices add more capabilities, become more powerful in terms of complexity and speed of processing, they will be able to sense more health parameters and respond suitably in real-time. Hence one could expect:
- Emergency calls and response even before a heart attack has occurred
- Real-time monitoring and response of other body metrics
- Personalization health guidance
- Smart agriculture
- Smart Cities
- Smart traffic management, smart parking