by Terry Edwards, Engalco Research
Already the 21st century has been characterized by dramatic advances in technology—much of which impacts the day-to-day lives of ordinary folk as well as professional people. The Internet of Things (IoT) represents a powerful and pervasive example of such advances. It is well known that the IoT is extensively used for many practical purposes, ranging from automatically (remotely) controlling domestic equipment (cooking, heating, lighting, etc.) to examining the security status of properties as an important example. Now that cellphones are so powerful, the extensive 4G networks are regularly used by their owners to connect through to the IoT.
Taken globally, in 2019 the total number of IoT connections is expected to amount to well over 1 billion. And by year 2026, this figure is expected to have way-more than tripled—reaching 4.8 billion. The detailed data is shown, year-by-year, in Figure 1.1
The IoT as such is of huge importance, but it has a highly significant subset known as the Industrial Internet of Things (IIoT). The IIoT is associated with ever-expanding bandwidth demands and is inextricably linked with “Industry 4.0,” which is often described as the 4th Industrial Revolution. As the bandwidths required for effective IIoT communications increase so does the demand for cellular networks with greater bandwidth capabilities than 4G. This effectively means 5G, indeed the IIoT is already a massive driver for the relatively new 5G networks.1 In monetary terms the markets for IIoT connections are expected to grow rapidly from US$7.1 billion in 2019 to US$40.7 billion by year 2026 (CAGR ~ 28%).
It is also important to appreciate that both the IoT and the IIoT necessarily embrace multitudes of relatively narrow bandwidth internal connections, implemented by links such as Bluetooth® or LoRa. Unit prices are in the “several US $” range for the required transceivers. In fact the great majority of IoT networks are classed as narrowband: NB-IoT.
There already exist many applications for the IIoT, for example:
- Factory premises
- Petrochemical plants
- “Smart” electricity distribution grids
As well as providing much faster network connections than earlier standards, 5G is also characterized by having major software improvements—particularly an approach known as “network slicing.” The concept and action known as “network slicing” is a completely vital aspect of 5G. Basically “network slicing” comprises the adaptive, automatic capability of a network to uniquely connect using appropriate systems and applications. Network slicing ensures that specific connections dynamically use appropriate links that suit the specific application. In this way, users are presented with a link that is effectively customized for their purposes. For example, not everyone needs a broadband link—only those who specifically need such a link over some identified period of time.
The massiveIoT (mIoT) and the mission critical IoT (mcIoT) refer to machine-to-machine (M2M) and high-reliability/low-latency connections, respectively. When these classes of connections are demanded, the 5G networks automatically dedicate an appropriate network “slice” (low-latency) to handle the critical requirement. Latency refers to the total throughput time—end-to-end through the network for a signal. “Low-latency” usually means around or less than 1 ms.
In this context, automotive and medical applications are particularly relevant. “Automotive” includes intelligent communications with cars and other road vehicles, i.e. “smart” cars and also autonomous vehicles (AVs). Increasingly, highly intelligent and agile communications will be required as millions of vehicles—mainly cars—enter and exit towns and cities on a daily basis worldwide. At first (and for several years to come) 5G mcIoT will just about cope with this requirement, but it is almost certain that yet another new technology (i.e. 6G) will be required. 6G will almost certainly demand THz transmission.
Medical applications (using mcIoT) are of course absolutely critical because life and death situations are ever-present and near immediate reactions are necessary. Only low latency connections are applicable for these applications.
Targeted on 1 ms latency, these network slices must utilize what is known as edge computing (the “edge cloud”) because this means the fastest possible route.
5G and IoT Security
It is the “new normal” that cyber attacks are increasing in all countries —both in sheer numbers and also in levels of sophistication. This feature as well as the advent of much faster and more complex networks such as 5G and the IoT means that security is ever more important. 5G networks are designed on a cloud-domain basis and are focused on third-party services—which increases their exposure to cyber attack. The situation is now so critical and sensitive it is already impossible for effective security to be maintained manually. Instead totally automated (therefore very fast) technology must be applied and this is where artificial intelligence (AI) comes in.2 With the IoT (also the IIoT) there is the additional cyber-sensitivity danger associated with retaining simple passwords for individual IoT devices. The application of AI overcomes such dangers. The automation of security is mandatory and dynamic security is highly desirable so as to adapt to changing circumstances, i.e. continuous learning and automatic application.
At their multitudinous nodes, all IoT (and IIoT) networks require sensors. Every IoT sensor must also be a “smart” sensor, i.e. must be compatible with all the digital network characteristics. Additionally MEMS (micro-electro-mechanical systems) technology is utilized in almost all sensors.
The great majority—by far—of electronic systems (including 4G, 5G and the IoT) implement mechanically static semiconductor chips (ICs) that function internally in a totally electronic manner. Basically, massive arrays of transistors are switched in nanoseconds to provide computer-like functionality. MEMS are totally different in that they implement arrays of tiny mechanical switches. Stepan Lucyszyn’s book is a good reference concerning MEMS.3 In practical terms, a MEMS chip may (superficially) look a lot like any other semiconductor product and would typically be housed in a QFN package. But this is where any similarity stops.
MEMS manufacturers and sales outlets exist in many countries worldwide and several online references provide useful information, notably Reference 4.
But in the present context, the focus is on the availability of MEMS-based IoT sensors, which reduces the selection down to those cited in Table 1.
The number of players headquartered in each country are summarized in Figure 2.
The total 19 OEMs cited in Table 1 (and Figure 2) exclude the other important players known as fabless providers. Such providers are responsible for designing, testing and marketing the final MEMS products but necessarily they sub-contract to specific fabrication facilities (“foundries”) to manufacture the products.
Somewhat surprisingly, the above information does not indicate any MEMS-based IoT sensor OEM headquartered in China.
The IoT (also the IIoT) working together with the new 5G networks represent vital and strongly growing aspects of 21st century life in both personal and business terms. The IIoT is associated with ever-expanding bandwidth demands and is inextricably linked with “Industry 4.0,” which is often described as the 4th Industrial Revolution. Software-defined approaches—notably network slicing —enable low-latency communications that are needed for applications including automotive and (especially) medical. AI is a must-have technology for on-going network security. MEMS devices are almost always the mainstay in IoT sensors.
1. The 5G Report. Released May 2019 by Engalco-Research: www.engalco-research.com
2. Jim Hodges (Heavy Reading) and Ravi Raj Bhat (A10 Networks), “AI-based Threat Protection for 5G & IoT Networks,” broadcast August 15, 2019
3. Stepan Lucyszyn (Ed.), “Advanced RF MEMS,” Cambridge University Press, 2010. ISBN 978-0-521-89771-6