Edge Computing and IoT, Why This Coming Together Matters Now!

In the world of AI, data is like oxygen.  If you do not possess and harness data, you will be left out in the wild.

I have been advocating edge computing as an essential component of digital transformation journey in all my previous articles.  It is gaining prominence daily.  Cloud behemoths Microsoft, Amazon, Google and their lesser known competitors have increased their attention to grab the market share in this new frontier.  Microsoft through Azure IoT Edge seeking a larger foothold in Industrial IoT segment, and AWS with its Greengrass.  Intelligent edge (computing) was a critical component in most of the testbeds showcased in the just concluded Hannover Messie 2018 (Industrie 4.0) conference as well.  Nearly every industrial company on earth were exhibiting in this year’s conference promoting, digitization, analytics, digital twins, industrial software, and applications, promising to fundamentally transform manufacturing forever.

From the early days of computing, two forces are in play, centralization and decentralization.  Computing power, flexibility and user experience were the deciding factors.   Intelligence and power got centralized in mainframe, then got distributed in client-server for better user experience and again consolidated as cloud for scalability.  Cloud definitely proved to be a game changer compared to all of its predecessors.  Cloud along with GPUs put enormous computing power instantaneously for further innovations, and has given a flip to other technologies that were languishing for decades, like AI.  It has provided avenue to store and process vast amount of data that in turn kindled the need for collecting more actionable data from the field, and Internet of Things has become the foundation for digital transformation. 

While cloud helps companies monitor and maintain equipment, provides infinite storage for enterprise and consumers and merges data from multiple points in a network, it fared not so well in mission critical environments where decision making in real time is key for successful outcome.  The learning from the cloud architecture, and technological advancements like run-time environments, server-less computing, containerization (Docker, Kubernetes etc.) high powered GPUs, edge analytics, 5G,  etc. are making the case for edge computing even more powerfully.

What is edge computing?   It refers to the processing of data at the edge of a computer network, closer to the source of data.  It is far more efficient and practical to process the data where it is collected than at the cloud in certain use cases.

Why now?  Gartner predicts, by 2021, there will be $2.5 million per minute in IoT spending and 1 million new IoT devices sold every hour.  Robots, wearables, healthcare equipment, autonomous cars, drones, consumer appliances, smart factories etc. generates vast amount of data.

Industrial Internet of things (IIoT) - The foundation for the fourth industrial revolution (Industry 4.0) is the Internet of Things.  IoT enables continuous data exchange between all participating units – from the production robot to inventory management to the microchip. This connects all production and logistics processes together, making each industry more intelligent, efficient and sustainable. The promise of IIoT is that better collection of streams of data from equipment will improve information and analysis, leading to more efficient production.  This results in terabytes of data crisscrossing internal and external networks.  Sending all the data to cloud as-and-when it is collected, and taking decision on them based on the analytic input may not be always efficient, and could negatively impact personal and equipment safety in certain scenarios.

Smart manufacturing requires flexible production systems and cross-platform, cross-industry applications, however this demands large-scale real-time industrial communications networks which present significant challenges.

Edge Computing comes in handy for the IoT implementations by coexisting and complementing well with cloud.  Having seen the need and value, major cloud and platform service providers and digital trendsetters are ramping up their solution offerings around edge computing. 

Data localization & actions (example - autonomous car, factory)

With edge computing capability, data generated by the sensors can be analyzed locally and actuators can take real time actions.  This is very critical to prevent accidents in the case of autonomous car where split second decision is vital or in a smart factory to shut down a malfunctioning equipment.  Edge computing also helps to sustain local operations, whenever there is loss of connectivity to central datacenter or cloud. 

Resource optimization: 

Research by industry analysts shows 40 to 50% of the data collected never gets analyzed.  Back-hauling all the data to the central datacenter or cloud only increases the resource utilization, without any tangible benefit.  It puts more demand on the transport (bandwidth) and storage.   Securing the data will be another challenge.  Edge computing enables local data processing for time sensitive decision making, and only the sifted (for example performance data) can be sent to cloud for deeper analysis.

Security & Privacy:

When it comes to IoT adoptions, data security and privacy remains as major concerns impacting the decision making. By processing the data locally, before the data crosses the firewall on its way to the cloud, measures can be put to protect that data from attackers. Also edge analytics can be configured to sanitize machine generated data before it is sent to the cloud. If the data is processed locally, it gives the power to control what goes in and out of the assets. 

Compliance:

I discussed in detail in my previous article on GDPR, that will be in effect from May 25, 2018.  Similar to that industry specific compliance requirements, like GxP, HIPPA etc. puts data segregation, need-to-know, storage & processing of personal data to the forefront.  Complexity of complying with these regulations increased exponentially with Internet of Things, as every device - embedded, standalone and connected, gathers data through its life cycle.  Processing data locally increases the chances of being on the positive side of the compliance framework.  It gives opportunity for companies to prevent data leakage at the source itself, and reduces the resources needed to safeguard the archived data.  It also helps to comply with industry specific, country and region specific regulations on data protection.

In addition to being a solution to some of the known problems discussed above, edge computing along with IoT creates value and new business opportunities for companies.

 

Digital twin capability:

Edge computing along with the recent technological breakthroughs in connectivity - 5G, TSN (Time Sensitive Networking) etc., companies can build digital replica of the entire factory to better understand how each element works together.  This enables the real time loop between design and execution. Digital twins can now expand from individual components or machines to visualize entire factory, location or even organization.  Companies can create value by optimizing processes, removing bottlenecks and by improving human and machine interactions.

Future proof investment – ready for emerging technologies like blockchain:

Blockchain is gaining wide acceptance across the industries, banking, healthcare, logistics etc. as a decentralized data management framework.  The potential of IoT and blockchain working together to solve identity and integrity of transactions is undisputed. The immutability and proof of work (PoW) is key for autonomous machine-to-machine transactions in the future. Solving the proof-of-work consumes substantial resources in terms of CPU time and energy.  Limited computing resource and energy supply of IoT devices become major barriers when blockchain is applied to IoT systems specifically because of the mining process. Edge computing enabled IoT architecture will provide enough resources to support blockchain applications.

Machine Learning and Artificial Intelligence:

Edge computing extends the machine learning and AI capability to the edge.  With the necessary horsepower available at the edge, now companies can train and infer machine learning models at the edge, closer to the source of the data.

Augmented Reality (AR) applications:

Augmented Reality has huge potential in industrial environments, like workers’ training, equipment maintenance, troubleshooting etc.  The biggest obstacle for deployment of AR solution  is that current devices’ lack of computational and graphical performance. While 5G has the potential to solve the connectivity puzzle for AR, offloading high computation tasks to edge computing solves another piece of the puzzle.

Regain control and monetize data:

By processing data at the edge, companies can regain control of the data they generate.  They can decide what to share to the outside world. Platform providers require data improve the products and services, and companies now will have the ability to negotiate with service providers for a win-win outcome.

Edge computing enabled Internet of things architecture takes the distributed computing to the next level.  It provides much needed power and control over the data.  It extends the capabilities of cloud to the edge, and poised to create new opportunities for business and consumer, patients and provider, and so on.