Innovation on the edge: how compute is shifting away from the core and demanding new ideas

In an era defined by real-time decision-making and connectivity, ‘edge’ computing is becoming indispensable. As organisations increasingly shift compute resources away from centralised IT and towards the edge – the locations where data is generated and used – new challenges and opportunities are emerging. The push towards the edge is fuelled by the demand for ultra-low latency, the exponential growth of smart devices, and the rise of the Internet of Things (IoT). But the transition to edge computing isn’t just about geography, as it also requires an entirely new way of thinking about infrastructure, security, and connectivity. Moving compute to the edge calls for innovative approaches to harness its potential across sectors, from healthcare to logistics and renewable energy.

The ongoing explosion of big data is playing its part, impacting the way we need to sift and analyse information. Traditional centralised approaches just don’t let us exploit information as effectively as the right kind of edge solution, powered by AI and brought to life by superfast connectivity. Shifting compute to the edge enables critical business processes to function in new and better ways, and speeds up intelligence gathering and decision making to make us more responsive and competitive.

In this context it is easy to see why analyst firm IDC is expecting worldwide spending on edge computing to hit a staggering $232 billion this year, up over 15% in 2023 and on track to reach $350 billion by 2027. But many of us still struggle to define exactly what constitutes edge computing. That’s probably because it is not easily boiled down to a narrow set of elements. Analyst firm Forrester defines it as “a family of technologies that distribute application data and services where they can best optimise outcomes in a growing set of locations and connected assets”. The crucial point here is that it’s a set of technologies, some established and some still emerging, generally software-defined but often rooted in the physical. So let’s consider some of the most important constituent technologies, and the innovative ways we can expect them to be deployed over the coming years.

Creating a greener, smarter edge

As compute power devolves from the centre, we are seeing the emergence of a whole new class of architecture: self-reliant smart buildings, effectively able to manage themselves through a tightly integrated set of technologies. 

“Smart buildings are being developed in a number of verticals – hospitality, healthcare, education, airports,” observes Xavier Mongin, global director for Government, Defence, and Smart Cities at Alcatel-Lucent Enterprise. “Common to all is the need for a hyper aware infrastructure, made up of sensors and other devices and supported by the ongoing convergence of operational technology (OT) and IT. Transportation, heating, ventilation, air conditioning, lighting, and security are moving to internet-native technologies, such as IoT and IIoT [Industrial Internet of Things]. We’re also seeing greater and greater use of generative AI and ML too.” 

What’s emerging, he says, is a greener as well as smarter type of construction: “We are going beyond the net zero ambitions of today towards buildings that are much more energy efficient, sustainable and intelligent, plus safer for humans,” he notes. “We’re starting to see biomimicry, where we imitate how insects such as bees and ants live. We can use AI to create swarm intelligence, and a big part of that will be distributed management and intelligence in the form of edge compute. That can’t be done easily out of the cloud. It’s about on-prem compute, close to the device we are managing, and it needs middleware to allow all the different building blocks to work together and deliver fuller automation.”

Staying secure outside the perimeter

Perhaps the most obvious impact of edge compute comes from migrating the enterprise application nearer to the end user – giving them access of perhaps sub-100ms compared with more like 300ms for an application sitting in a public cloud. That’s LAN-like performance. But where does that leave data security? Conventional wisdom holds that a remote and fluid perimeter leaves data more vulnerable than it would be inside a corporate setting. Not necessarily so, argues Prakash Mana, CEO of Cloudbrink, a provider of secure connectivity for hybrid workers at the network edge.

“The nature of edge compute is that it’s so transient,” he observes. “But actually that’s a strength. If you give an attacker an immovable perimeter to chip away at, and unlimited time to do their work, sooner or later they will find a way to get past it. Offer them a continually changing edge, and they no longer have the leisure to crack it. A moving target will always be harder to hit than a stationary one.”

Cloudbrink is one of a number of proponents of Automated Moving Target Defence (AMTD), as defined by analyst firm Gartner and based on orchestrating movement in various IT environment components and layers across the attack surface, creating uncertainty and complexity in the eyes of an attacker.

Defining the future of the edge

Quantum computing, though still a work in progress, looks like a technology that has the potential to make edge data perform in important new ways. But at present its role is to complement existing classical computational resources, not to replace them, points out Krisztian Benyo, technical business developer with quantum technology player Pasqal. “It is useful in tasks like resource management, inventory management and itinerary planning,” he says. “Quantum is great for edge tasks that are hard to solve with CPUs alone. Its qualities make it useful, for example, in factory floor automatisation where there are a lot of resources that you need to track and allocate and constraints that you need to deal with. It would be applicable too in any kind of supply chain management, helping to optimise processes. I see a future for it in transportation and logistics, as well as energy. These all have a lot of complex problems and a variety of resources to manage.”

Benyo also points to ongoing development work looking at how quantum can synergise with AI, especially generative AI: “By integrating quantum with machine learning you can greatly reduce the time needed to train algorithms and reduce computational costs and energy consumption,” he claims.

An example of how artificial intelligence is already transforming the edge is Netherlands-based Axelera AI, which also has a UK office based in Bristol’s Quantum Technologies Innovation Centre (QTIC). It is reinventing security surveillance with a new toolkit that empowers developers to quickly build edge AI applications, such as multi-channel video analytics, quality inspection, and people monitoring. The platform allows developers to innovate in ways that aren’t possible with CPU and GPU-based solutions by offering accurate AI processing.

Connecting the edge with blockchain

Massive growth in the importance of AI and IoT at the network edge is demanding new approaches to handling and networking data. One innovative approach has been developed by blockchain platform Quranium in the form of distributed ledger technology based on post-quantum cryptography (PQC). 

“Lots of the blockchain platforms we have today are crackable by powerful computers,” claims Mumbai-based Kapil Dhiman, the company’s co-founder and CEO. “We have used PQC to create our architecture so as to make it practically uncrackable. It’s also one of the fastest in the industry.”

Dhiman expects many of the future applications of this technology to be IoT-related, not least in the development of driverless cars: “Autonomous vehicles rely on edge-based technology that has to interact with a centralised server,” he explains. “Just imagine if that driverless car gets hacked. That could result in havoc. What you need is infrastructure that has no point of failure and is, in practical terms, unhackable and uncrackable.”

The same technology, he says, can apply to everything from smart watches to robotics: “Edge compute is going to become incredibly important over the next decade, in applications ranging from satellites in space to gaming, and you can’t afford to have slow compute or devices at the edge become unusable. I would expect a lot of innovation here, to make edge processing much faster, more efficient and more secure.”

Security, claims Dhiman, is at least as important as latency, and he is confident that post-quantum cryptography is the answer: “Lots of major organisations, including Apple and Google, are moving their data communications and data storage to post-quantum cryptography.”

When edge is off the planet

The science of quantum photonics, though still at R&D stage, is promising a new era of edge compute, taking the processing of data beyond the stratosphere and into the cosmos. “By reducing size and weight and cutting power consumption, you can take compute and put it in a satellite,” says Roberto Siagri, CEO of Padua-based Rotonium. “The photonic qubit that we have designed will, once it is put into manufacturing, work in a classical compute regime or a quantum one. The magic is that, through entanglement, you can conduct compute in space as though it were all in a single computer on earth. By entangling quantum processors, they act like one object, meaning you are not forced to have all your compute in the same place. If we can move more and more compute tasks into space, we can dramatically reduce the energy needed for them. We can use the sun as the power source. And we can increase resiliency too. I envision it as a whole new way to manage data.” 

But with commercial deployment of quantum photonics still a few years off, edge compute is going to remain Earthbound for a while. “We’re not suddenly going to get rid of all our centralised clouds and data centres,” warns Azfar Aslam, vice president and CTO at Nokia Europe. “In any case, it should be the applications that are dictating which architecture option we use. It’s about striking the right balance between economics and performance. Our job at Nokia is to support our customers no matter which infrastructure route they go down, whether that’s on-prem, close to prem or in a centralised data centre.”