Nobody doubts the success that a technology like the IoT (Internet of Things) has achieved and still has to achieve. However, as always happens in the world in which we live, the inexorable passage of time reveals new trends and technologies that challenge the position of existing ones, improving them and proposing new paradigms. And this is exactly what happens with IoT technology.
From IoT to Edge Computing
The basic architecture of an IoT system consists of 3 main technologies: embedded systems, an intermediate layer or middleware, and cloud services. In this architecture, embedded devices, which typically consist of sensors and actuators, use different routing nodes such as modems, cellular base stations, etc. to communicate with the cloud, which provides the comprehensive mechanisms for storing, processing and managing all the information generated.
However, over the past decade, there have been various standpoints that have tried to extend centralized cloud computing (CIoT) to a more geo-distributed way in which computing, networking and storage resources are much closer from the data sources and from the end user, what was called Fog Computing at the beginning of the last decade. Its purpose is to solve some of the challenges faced by the IoT such as bandwidth, latency, communication interruptions, Qos degradation and security. Some of these aspects will be discussed in more detail below.
However, even with this new type of architecture, not all these resources are computationally ready to run large-scale IoT applications simultaneously. That is why we could think of extending calculations from the cloud (CIoT) and intermediate network nodes (Fog Computing) to directly the end devices of the network, capable of doing all kinds of processing and decision-making, driven mainly by the Machine Learning and deep learning capabilities embedded in them. This technique minimizes communication with the cloud and is known as Edge Computing.
This type of data processing, that is directly embedded in the end devices of the network, helps the applications of healthcare, automotive and manufacturing, computer vision and image recognition in a faster, efficient and secure treatment of information .
Edge Computing: The new great disruption
Edge Computing technology is here to stay. What a few years ago would have been unthinkable for technological reasons is today a reality. Undoubtedly the take-off that 5G technology is expected to bring in this decade, together with the democratization of Edge AI technology, will act as catalysts for the growth of this potential market.
According to a new report by Grand View Research , the size of the global Edge Computing market is expected to reach 43.4 thousand million USD by 2027, showing a compound annual growth rate (CAGR) of 37.4% during the period from 2020 to 2027. Specifically, a value of 1.94 thousand million USD is expected in the European Union by 2023, expanding with a CAGR of 29.3%.
These figures highlight the importance that this technology will have in the present and near future.
IoT Edge Computing Hardware
Undoubtedly, one of the key factors for the success of Edge Computing technology has been the advances in dedicated hardware that allow the computational calculations demanded by these types of applications to be optimally and quickly carried out. Thus, many companies have chosen to develop ASIC solutions for this type of problems.
Although these two examples can be interpreted as advanced solutions for intensive calculation operations, there are also others focused on energy efficiency.
NXP, the manufacturer of microprocessors (CPUs) and microcontrollers (MCUs), has recently released a new family of MCUs called Crossover due to its combination of power and graphic acceleration typical of application processors, with the simplicity and efficiency of MCUs. This is intended to combine unprecedented performance with high levels of integration and security in order to drive industrial, IoT and automotive applications.
In this sense, the processor design company ARM has released its latest Cortex-M55 family, which is its Cortex-M processor with the highest AI capacity and the first with the new Helium technology specially created for Machine Learning. Likewise, it has also designed the first Neural Processing Unit (NPU) called Ethos-U55 intended for embedded environments. This announcement will allow to take the processing speed and efficiency to another level, given the relevance of the company in the IoT world. Surely we will see the first MCUs that implement these solutions in the coming months.
Edge Computing philosophy offers a wide variety of possibilities. Even it is possible to avoid processing the information within the embedded systems themselves, microcontrollers or microprocessors, and to carry it out in the information sources: the sensors. Thus, the manufacturer ST has IMUs with a Machine Learning core and decision-making trees in its catalogue. In the same way, Sony has recently announced the first image processing sensor with integrated AI processing.
Secure IoT Edge (Trusted Edge Computing)
One of the great challenges of IoT technology is in the security field. By placing devices away from the full control generated by the cloud or by a secure data centre, the chances of attack are increased when equipment with few protection resources are deployed in insecure and sometimes unknown locations.
In an effort to address this challenge and maintain the required security, the different actors involved are turning to different solutions.
The most widespread solution nowadays in this field is based in hardware security modules (HSMs), and the most common type is Trusted Platform Module (TPM). This device consists in a set of security protocols created and maintained by the Trusted Computing Group (TCG) for the creation and secure storage of keys. Although there are also software versions called virtual TPM (VTPM), they are much less secure than their hardware versions .
Other HSM solutions are based on allowing small pieces of code to run in a safe and secure space called the Trusted Execution Environment (TEE). Any code that is executed in a TEE is not visible or accessible from outside, which reduces the chances of attack. Two of the main hardware components that support these TEEs are Intel SGX (for x86 and x64 processors) and ARM TrustZone (for ARM processors).
ARM is just one of the companies that is betting the most on bringing these secure boot solutions and the creation of a chain of trust, from the world of CPUs and high-performance devices to the most resource-limited such as MCUs. The result of this is its latest Cortex-M33 processor designed for embedded and IoT applications that require a special emphasis on security, being NXP the first manufacturer to design a family of MCUs (LPC5500 series) that integrates it.
The dizzying era in which we live today, technologically speaking, forces us more than ever to have to continuously recycle and learn almost constantly. The evolution of IoT technology towards Edge Computing is one example of this. It is a technology that takes advantage of the progress of 5G technologies, Machine Learning and Artificial Intelligence and Security among others to take the definitive leap towards a much more connected society that will undoubtedly give plenty to talk about in this decade.
At TEKHNĒ we always bet on the latest innovative technologies as a key differentiating aspect. Our mission is to be the perfect technology partner in which firms from different industries rely on to transform ideas into the next great market disruption.
Do you want to jump into the technology of the future? Contact us and we will be happy to assist you.
 Beginning Azure IoT Edge Computing – Extending the Cloud to the Intelligent Edge. David Jensen, Apress, 2019.