One of the defining characteristics of the Internet of Things is the requirement for connectivity. IoT applications are inherently about remote monitoring and management of devices or processes. The remote aspect necessitates some form of connectivity for delivering data from point A to point B. There are many different technology options for providing that connectivity, as illustrated in the chart below.
Technology selection is all about trade-offs. Ideally every IoT device would be connected using a single ubiquitous, ultra-secure, high bandwidth, low power, low latency, low-cost network. Unfortunately, no such network exists. So, any selection needs to balance the requirements of the application. Every deployment has to work within some constraints. Fortunately, IoT has evolved dramatically in the last few years to provide an increasingly rich array of connectivity technologies (as well as operating systems, networking protocols, middleware platforms, and more) to better address the needs of IoT. We term these the ‘Thin IoT’ stack. Despite this progress, any decision over which technology to select to connect an IoT device needs to weigh up a range of criteria, including location, mobility, data speeds, bandwidth, power, latency, security, and cost, all explored below. These must be balanced against the requirements of the application.
The location of a device will be perhaps the most important determining factor of what types of network technologies might be useable. Those that are located within the walls of a home or a factory can be connected using private networks, typically in the form of WiFi (and with a growing interest in the enterprise space in Mobile Private Networks using cellular technologies). Those that are located outdoors will typically need to rely on cellular networks (e.g. 4G or 5G), satellite, or may have no suitable network coverage at all.
We can break down the technologies into four main categories. Personal Area Networks (PANs) such as Bluetooth or NFC, have a range of a metre or two and are used for connecting personal devices such as headphones. Local Area Networks typically have a range of a few tens of metres. The most widely deployed of these short-range technologies is Wi-Fi, almost universally deployed in both residential and enterprise settings for supporting general data communications needs. Other examples, much more relevant to IoT include Bluetooth Low Energy (BLE), Thread and Zigbee. In most cases these networks are deployed by the users themselves, and operated as private networks.
The reverse is true of Wide Area Networks (WANs), which provide national and international coverage, albeit limited generally to areas of higher population density. Communications service providers (CSP) run these networks for multiple customers, who pay them an access fee for doing so, these are referred to as ‘public’ networks. For our purposes, there is one category of technologies that is particularly relevant here: cellular networks, be they 2G, 3G, 4G or 5G. There is also a growing trend for public networks using unlicensed spectrum, including LoRaWAN and Sigfox (discussed in our Low Power Wide Area page).
The final group is Global Area Networks, which relates to satellite connectivity for the globe, including those areas not otherwise connected by WANs. This is a rather niche area today, but there are dozens of companies looking to launch Low Earth Orbit (LEO) satellite constellations to provide internet connectivity and to connect IoT devices, often using adapted terrestrial technologies such as LoRa and NB-IoT.
These different network technologies have varying capabilities in terms of data speeds. Meanwhile certain applications, particularly those involving the transmission of video, require higher data rates, possibly continuously, whereas for simple sensors it is enough that a few small messages of a few bytes are delivered irregularly.
There are technologies available to suit whatever the deployment requires. 5G, for instance, can deliver speeds of around 100Mbit/s. 4G LTE has varying capability depending on the category, with Cat-1 delivering 10Mbit/s, Cat-4 150Mbit/s and Cat-M offering just 1Mbit/s as the Low Power Wide Area (LPWA) variant. Other LPWA technologies are event more limited. LoRaWAN provides only around 50Kbit/s, and Sigfox has a daily limit of just 140 messages of 12 bytes each, giving not much more than 1KB per day.
Within buildings, applications that demand high bandwidth will tend to rely on Wi-Fi, whereas there are more options for low bandwidth, such as Bluetooth Low Energy (BLE), Thread and Zigbee.
The concept of latency is increasingly discussed in the context of IoT. It refers to the delays in getting data packets from point A to point B. This is typically a function of the data speed, i.e. higher bandwidth networks have lower latency, with most technologies providing an appropriate level of latency for their applications. But for some use cases, having a reliable real-time connection is increasingly important, for instance in gaming, augmented reality/virtual reality (AR/VR), or some other sophisticated use cases, such as remote surgery or managing autonomous vehicles. Some technologies have latency levels in the tens of milliseconds (below about 30 milliseconds is considered ‘real-time’), while for LPWA technology Sigfox it’s in the tens of seconds (which is fine for the kind of non-real-time use cases that it’s designed for).
Some IoT devices have access to mains electricity or another constant source of power. Most devices within homes, factories or workplaces, for instance, have ready access to a plug socket, while any connected car application tends to have access to direct power from the vehicle battery.
There are many potential IoT use cases, however, which do not have such access. Environmental monitoring or track & trace use cases, for instance, will need to rely on battery power. This has the obvious draw-back that these devices will, from time-to-time require their batteries to be replaced. This can severely increase the cost of supporting such an application. For this reason, the arrival of technologies specifically aimed at keeping power consumption low opens up a lot of additional opportunities in IoT. The most prominent example is in the arrival of the Low Power Wide Area (LPWA) technologies, but also in more efficient short-range technologies such as Bluetooth Low Energy (BLE) and Thread. The LPWA technologies promise to be able to support devices for up to 10 years between charges, whereas for 5G (for instance) the equivalent would be only a matter of days or weeks.
Energy harvesting is an interesting emerging trend, whereby the device can gather enough energy from ambient sources or through kinetic energy (i.e. the movement of the device) to send and receive (a very small amount of) data.
Security is typically quoted as the number one consideration and concern for any organisation deploying IoT. Different technologies will provide different levels of security. A private network for instance is the most secure. Communication over public networks can be enhanced through the addition of the likes of private APNs, IP VPNs, Transport Layer Security (TLS), hardware security modules, and IoT SAFE SIMs.
The choice of connectivity technology will have implications for security. The more constrained a technology is in terms of data speed, generally the lower its grade of security. Higher bandwidth devices can run more sophisticated security protocols, along with more sophisticated device management to ensure it is up to date.
It should be noted that any security breaches are far less likely to be the result of any inherent limitation in the connectivity technology, and far more likely to be due to human error or some kind of oversight in developing the application or the hardware. Networks are generally very secure, but there are still grades of security.
Cloud integration also poses issues as it requires certain levels of security that some connectivity technologies simply cannot provide. Hence the need for what are termed ‘cloud connectors’.
The ‘right’ level of security will depend on a lot of factors, most critically the sensitivity of the data.
The choice of which technology to select will also have to include considerations of cost, both of hardware and of connectivity.
All of the above-mentioned features have variable costs associated with them. Higher data speeds, lower latency and enhanced security all result in higher costs. A 5G cellular module, for instance, will have a price tag of hundreds of dollars, and even for LTE the price varies between around $10 and $40 depending on the capabilities. In comparison, and in the LPWA technology group, cellular devices can cost around $5, a LoRaWAN module costs perhaps $2 and for WiFi it is under $1.
There is also the question of the cost of the connectivity, i.e. the payment made to a network operator. In the case of private networks this is zero (although there may be a third-party company managing the network), whereas for public networks it could be anything from $1 per year to $50 per month per connection depending upon the location, bandwidth and volume.
The price of both elements has been coming down significantly in recent years, opening up the opportunity for new use cases.
As we wrote at the top of this page, the need for connectivity is one of the defining characteristics of the IoT and, as with almost everything to do with the IoT, there is no simple answer as to which technology is best to use. Rather, some technologies are better suited to certain applications than others. Generally, there will always be compromises including potentially in software application and device functionality, both of which can be optimised to work better with different connectivity technologies even for specific applications. As such, the discussion of connectivity technologies contained in this introductory page can only ever scratch the surface of what is an extremely wide-ranging and complex topic.