How to work AI/ML and Edge Computing in IoT

The “edge” is magical. For example, environmental science studies habitat borders where certain plant varieties grow strongly at the edge and not further. Similar phenomena have been observed in astronomy at the edges of the universe.

Human societies are no exception. A new revolution is underway with high computing powers that move to the edge, a phenomenon becoming increasingly known as Edge Computing.

Industry 4.0, IoT

IoT is about collecting and analyzing data, insights, automation, and automating processes that involve machines, people, things, places, and other objects. IoT is therefore a mixture of sensors, actuators, and connectivity. It also includes storage and cloud computing.

Industry 4.0, also called the fourth industrial revolution, is heavily dependent on IoT tech. It will reshape automotive, transport, and healthcare as well as commerce. Their merging is a natural phenomenon, as AI and ML are the driving forces behind all 5G/IoT innovations.

5G supports strong IoT

One of the most important innovations in 5G is its strong support for IoT. This includes support for low-cost and long-battery-life sensors.

Data storage and computing will become a ‘fabric continuum’. That fabric touches all industries and a slew of AI, ML and Edge Compute solutions will eventually discover their optimum zones and roles depending on the use case.

Hyperscalers

Google, Amazon, Microsoft and other cloud service providers have introduced a new type of IaaS (Infrastructure as a Service) with manageability and transparency and competitive pricing.

Connectivity

Offering advanced IoT solutions has never been easier thanks to the availability of a variety of options, including LoRAWAN, NB IoT and the rapidly expanding 5G.

Edge Computing

The massive advances in computing silicon have enabled edge devices to be powerful. They allow local data assessment and real-time decision-making, completely avoiding the trip to the server.

Some other transformations can be seen more clearly in terms of player engagements. Telecom operators using Hyperscalers (mentioned previously) and the transition of Operations Technology (OT, such as machine control systems moving to the IT domain).

In the hopes that they can be used and paid for by consumers and enterprises, IoT applications are becoming more popular. It is evident that it will be most effective to deliver it from the closest location to the customer for a variety of reasons. This zone is also a hub of activity.

All these players seek to control the space between the enterprise and the consumers.

This space will basically mean that workloads are shifting from the central data centers to on-premises or near-premises dedicated edge computing cloud. Sometimes, the central data center shift can mean moving an on-premises data center to an on-premises dedicated cloud.

This implies a shift from the Capex to an Opex model which will have an impact on enterprise financial models.

Edge Cloud in IoT World

This can be seen as a positive change in the value chain. In what is now known as the Edge Cloud industry, the cloud is moving towards the edge.

An easy explanation of Edge Cloud

The Edge Cloud is simply a combination of smart edge devices, including sensors, nodes, and gateways, with software (algorithms and security stacks. Connectivity modules. Sensing and actuating parts. and A processor in a full stack) to manage hundreds of sensors per gateway.

These nodes can be connected automatically by advanced clustering, nodal reconfiguration, and nodal re-routing.

Where does the tech for the cloud come from?

These technologies are available and are being used in scalable ways. The protocols are standardized to allow data to be offloaded to the cloud for non-critical data processing but to retain real-time processing at Edge Cloud.

Can AI and ML be stopped?

We expect millions of IoT devices and services to use Edge Clouds to help human life as Silicon prices fall and increase in volume. AI/ML is moving toward the edge is an unstoppable process, and in fact, it’s a requirement.

These technologies have clear benefits for the following industries:

• RT processing ensures low latency responses (improves safety, reduces defect rate)
• Data security at the local level
• Data sorting, filtering and pre-processing (eases cloud load).
• Combining licensed and unlicensed spectrum results in more efficient transport systems.

AI image processing, object detection and audio/video recognition capabilities also have improved dramatically. These capabilities are now available with some Silicon vendors as an additional capability. As deployments increase, we expect them to be more common and better priced.

Is this Edge Cloud real?

Are there overhypes about Edge Computing? There are many assumptions that underlie this incredible boom in Edge cloud theory.

1. It is not clear if Edge Cloud will grow as big as it seems. It is a highly-growth industry, there is some evidence. The ownership of this service is still being determined. Telco versus Hyperscaler, specialist enterprise system integrator, or Cloud Edge specialist.
2. Edge Cloud is not a one-size-fits-all solution. It is not clear what applications will require Edge Cloud capabilities, such as those in manufacturing RPA or healthcare. Companies are still evaluating business cases and pricing models.
3. Edge Computing is a solution searching for an application. Are consumers less willing to spend because of the short round trip delay? Is it possible to confuse edge storage with edge computing, and treat all applications the same?
4. What about the sunk costs of central data centers? They could be underutilized, or even end up storing non-critical data. If Edge Computing is all about Real-Time, then a 5G slice could be an important part of the solution. What business model will be sustainable between hyperscalers and telcos?

Conclusion

These new revelations will be revealed soon. As the space develops, it will be more obvious how AI/ML technology will align in the new world order. This could lead to new positions for service providers and their alliances with the solution providers.

AI/ML will eventually combine to create a unique landscape that will attract users and consumers over the next decade.