- AI, IOT and their diffusion
- Edge Computing and its future
- Ecosystem, not together
- Startups and new solutions
AI, IOT and their diffusion
Predictions regarding the spread of the Internet of Things have proven to be underestimates – in 2015, Gartner predicted that there would be 26 billion connected objects by the end of 2020 – this figure was reached and surpassed already by the end of 2019.
We have seen a significant acceleration in the adoption of these new tools, but what are the current estimates for this sector?
Waiting for the data from the Observatories of the Politecnico di Milano, the from Cisco Visual Networking Index tells us that in 2021 there will be more than 85 million connected objects in Italy – more than smartphones (about 63 million) – and that 51% of all connections will be machine-to-machine – M2M.
But are M2M and the Industrial Internet of Things the same thing?
No – they are two different paradigms of communication.
Both solutions have automation in common, i.e. they don’t require human interaction, the only difference is in the way the machines relate to each other.
M2M uses point-to-point communication while an IoT system typically places its devices in a global network, allowing large-scale integration.
From a networking point of view, both the Internet of Things and its industrial variant – IIoT – can be said to be evolutions of M2M, but of a qualitative, not merely quantitative kind – forming large cloud networks of heterogeneous devices, communicating with users and with each other, they make it possible to deliver increasingly complex and sophisticated services.
Another important aspect of the IoT/IIoT paradigm is scalability – integrating new devices into existing systems is quick and easy, unlike in M2M, where adding a new machine or changing the system requires more time and effort, as new point-to-point connections have to be created for each device.
Until a few years ago, IoT was relegated to certain sectors, such as health, fitness, fuel control or in-car infotainment devices, but we have seen a very rapid expansion in its use – in short, we have realised that you cannot talk about ‘Digital Transformation’ – business or private – without IIoT/IoT.
IoT is becoming more and more ubiquitous but, to exploit its full potential, it too must evolve.
Today, the stumbling block is that most of these devices, while capable of sensing data, have no capacity to process it – talkative but still dumb technology as it were.
Processing power becomes more and more crucial, because data collection without processing is almost useless, and real-time data analysis becomes increasingly important.
Smart objects therefore cease to be gadgets and become embedded systems with AI and, in some cases, machine learning capabilities.
It is expected that by 2025 25% of IoTs will be equipped with this kind of capability, growing from the current 0.5%. Their relative rarity at the moment is also due to the fact that there is still a lack of dedicated AI/ML IoT hardware, and a lack of developers to boot – out of around 22 million software developers worldwide, only 1.2 million focus on embedded systems and, of these, only around 0.2% have minimal AI/ML skills.
There is therefore a clear need for IoT hardware, simple and familiar, but able to make AI tangible. In short, we need development platforms for the new generation of value-added services.
In the industrial sector, the data confirm that those who have introduced these technologies achieve tangible and measurable results in a number of strategic areas: from predictive maintenance and asset management, to supply chain traceability, logistics, improved customer care and much more.
AI delivered by the new IoT solutions will no longer be confined to the world of specialists, but will have a radical impact on businesses, PA and consumers – a winning combination that will change the way we see and experience the world.
We are, in fact, in the midst of a real revolution, not only economic but also cultural – the McKinsey Global Institute’s study Modeling the Impact of AI on the World Economy forecasts an increase in global economic activity of around 13 trillion dollars by 2030, with a growth rate of around 1.2% of GDP per year, yet all this is in no way comparable to the cultural upheavals that have happened and will happen due to AI- we are witnessing the birth of a new world, where things that were unimaginable until recently will be possible.
There is, therefore, an urgent need to prepare the new generations and to ‘update’ the older ones to the conscious use of these powerful tools.
Edge Computing and its future
Data is the engine of digital transformation – information must therefore be easily accessible and usable, creating a path to sustainable value creation. It is important to collect data, but even more important to process it. This, until now, was mainly done in the Cloud – a necessary step as AI algorithms require a non-indifferent amount of computational resources.
Edge computing provides a solution to the typical limitations of Cloud infrastructures – it avoids latency problems, overcomes the coverage problems of fixed and mobile networks and, above all, allows decisions to be made in real time.
Edge computing is not new, but the innovation is applying it to the IoT.
IIoT research conducted in 2020 by IOT Analytics summarises the benefits of IOT edge computing very well:
- Open architectures: freedom from proprietary protocols and closed architectures, avoidance of vendor lock-in, increased interoperability.
- Pre-processing and filtering of data: “cleaner” data to send to the Cloud, thus reducing bandwidth costs.
- Edge Computing: high computational capacity to analyse data, low latency and high data throughput.
- Distributed applications/multi-functionality: Applications are decoupled from the hardware hosting them, thus giving rise to flexible architectures where, depending on the need, they can move vertically to the Cloud or horizontally from one intelligent computing resource to another.
- Consolidated workloads: presence on the same hardware of several operating systems (Linux, Windows, RTOS) and containerised applications (Docker 1, Win Container) leading to workload consolidation.
- Scalable deployment/management: Intelligent Edge computing resources are securely connected to local or wide area networks (LAN, WAN) and can therefore be easily deployed and managed from a central location.
The IIoT Edge computing market is estimated to reach $30.8 billion by 2025, up from $11.6 billion in 2020.
The interest in Edge computing is justified by the fact that replacing ‘dumb’ IIoTs with IIoTs that make use of it has several important implications for companies in all sectors: increased system flexibility, increased functionality, dramatic cost reduction.
With the advent of new IoT devices, equipped with computing capabilities that were unthinkable just a few years ago, the ability to collect and process data where it is produced has opened up, resulting in lower latency and more efficient communications.
In addition, Edge processing makes data quality management, such as filtering and prioritisation, more efficient. Performing these tasks on-site means cleaner data sets that can be processed in the cloud for further analysis more quickly and efficiently.
Ecosystem, not sum
The IoT increasingly brings the physical and digital worlds closer together.
In addition, we are seeing a progressive reduction in costs and energy consumption.
Concepts such as sensor and data fusion are starting to become popular among IoT applications – by pooling data from multiple sensors and applying AI algorithms, new information is extracted, and by comparing multiple sources, the data itself becomes more reliable.
Around their home speakers, OTTs have started to build important ecosystems with a pervasiveness well beyond the boundaries of the Smart Home, with new applications ranging from the Smart Car to the Smart City and more.
The spread of IoT/IIoT Edge Computing therefore allows the creation of an ecosystem that is no longer a simple collection of objects. The connection between intelligent objects is no longer enough, an onboard intelligence enabling new scenarios and services that make the most of the data collected is necessary.
Start-ups and new solutions
According to data from POLIMI Observatories in 2020, the contribution of startups in creating new solutions continues to grow.
These solutions can be divided into three macro-categories: Software Solutions (85%), Enabling Technologies (8%) and Physical Solutions (7%).
In Software Solution we find startups that develop algorithms for AI, in Enabling Technology those that develop structural and preparatory components for the realisation of an AI project, and finally in Physical Solution those that develop Autonomous Robots, Vehicles and Intelligent Objects.
It is very much the latter that are most involved in the realisation of the above-mentioned intelligent systems.
When it comes to start-ups, it is important to monitor not only the funding received but also their actual ability to conquer/maintain the market. 2020 has been a challenging year for many innovative startups, to say the least. The world of R&D has been frozen by the pandemic, and many projects have suffered, if not a setback, then at least a slowdown.
For start-ups developing AI hardware in Italy, it is not easy to find investors, because by its very nature hardware is demanding, expensive, complex to manage, and therefore very risky.
Moreover, there is still a slight scepticism on the part of the major players with regard to the world of start-ups, which are often considered as ‘hobbyists’ and rarely worthy of attention for Open Innovation projects, despite the fact that there is a lot of talk about them.
This is a delicate moment in history, but also a crucial one – with its power, IoT on the edge can not only sustain, but also speed up the digitalization of our country – we just need to bring it on the field.