An edition of Ericsson’s Mobility Report published in November 2019 projected at least five billion active cellular internet of things (IoT) connections by 2025. Numerically speaking, there will be a fairly even split between LTE-M and NB-IoT. In recent months, high-profile large-scale LTE-M/NB-IoT roll-outs on both sides of the Atlantic have backed this up.
IoT connections growth
(Image: Ericsson Mobility Report)
These low-power wide-area networking (LPWAN) protocols each have their own respective merits in terms of performance parameters and dominance in certain geographic regions.
Though 2G and 3G can provide the low bandwidth, power-frugal operation that’s still seemingly a great fit for contemporary IoT node deployments, migration to 4G networking (along with eventual 5G adoption) means cellular operators must make spectrum available to accommodate such next generation technologies. Depending on which part of the world we’re talking about, operators are decommissioning either their 2G or 3G infrastructure, but still keeping one of these in place to offer some level of legacy network support.
Cellular IoT connections by segment and technology – billions (Image: Ericsson Mobilty Report)
Across North America, Oceania, and much of Asia, 2G has already been canned – with just 3G remaining of the older cellular generations. In contrast, from a European perspective, it’s likely that 2G will have greater longevity and 3G will be phased out. The reasoning is that there are enough installed IoT nodes in Europe which are dependent on 2G networks to justify such retention.
Given that long-term consistent 2G/3G coverage cannot be assured on a worldwide basis, the time has come for engineers to start considering LTE-M and NB-IoT as the foundation of their future IoT developments. NB-IoT and LTE-M were both introduced as part of 3GPP Release 13, back in early 2016, with the intention of establishing communications requiring only minimal power budget – thereby enabling IoT nodes’ battery lifespans to reach decade-long periods (and minimizing battery replacement activity).
Though NB-IoT and LTE-M both utilize a stripped back form of LTE, there are distinct differences between them. LTE-M can deliver data rates of around 150-200kbps (through its 1.4MHz bandwidth), with the capacity to support some higher bandwidth workloads, such as voice services. NB-IoT has a mere 200kHz bandwidth to play with and can only really handle 50-60kbps data rates. Utilizing batch communication (to curb power drain) is suitable for very low bandwidth use cases (where intermittent transfer of small amounts of data suffice) but can attend to a huge number of nodes. These will generally be fixed. Conversely, LTE-M shows much better mobility credentials. Another favorable attribute, in certain circumstances, is its 50ms/100ms latency response, whereas NB-IoT has way over an order of magnitude longer latency periods associated with it.
Once again, there are regional variations to be aware of. LTE-M has gained significant initial traction in the United States, Japan, Australia and New Zealand, while NB-IoT is so far proving more popular in China, South Korea, plus European and CIS states. Therefore, availability should be taken into account before embarking on an implementation.
It’s straightforward enough to replace 2G/3G with either LTE-M or NB-IoT. Furthermore, these protocols present a simple upgrade path to 5G LPWANs in the years ahead. Commercial versions of LTE-M and NB-IoT currently available are forward-compatible with 5G, meaning there’s provision for them to work within these networks as they are brought online. In most scenarios, it will be possible to upgrade NB-IoT and LTE-M modules to 5G LPWAN via over-the-air software updates (so the expense and inconvenience of sending engineers out into the field can be avoided). Often it will probably be necessary for 4G and 5G IoT solutions with global scope (using NB-IoT or LTE-M hardware) to have the option of reverting to 2G, if no alternative is available.
Alongside choosing the best suited cellular connectivity type for IoT nodes, there are other factors that require contemplation. Among these is specifying microcontrollers that are fully optimized for cost-sensitive and resource-constrained use cases, with a well-defined technology roadmap. Among the parameters that will need to be mulled over are:
- How frequently will data be acquired?
- What frequency will the microcontroller core need to run at to deal with the expected processing workload?
- What will the associated power consumption profile be?
- How much data storage capacity should be provisioned for?
- Which are the most prominent security threats that will need tackling?
- What is the budget that must be adhered to?
If some of the actions are time-critical then measures must be taken to ensure the microcontroller does not add to system latency. Also, if human interaction with the IoT nodes is likely, then voice recognition or biometric mechanisms may need to be included. Thought should be given to the operating systems and the supporting development tools too.
The restricted nature of edge-based IoT will mean that trade-offs are always going to have to be made, but having the right set microcontroller options to choose from will certainly help. Continued architectural advances are leading to much higher degrees of microcontroller integration. This, in turn, is equating to stronger security, a wider breadth of functionality and smaller form factors, while keeping down the unit pricing and power requirements involved. Multi-core devices are being introduced that have completely independent power domains. This means that cores can run different applications in parallel with one another, or be turned off if not needed.
>> This article was originally published on our sister site, EE Times Europe.
|Tom Pannell is senior director of marketing for edge processing at NXP Semiconductors, where he is responsible for narrow band connectivity products and product marketing strategies for the company’s 15.4 and BLE solutions. He also oversees its wireless partner strategies that combine the NXP’s MCUs and MPUs with wireless solutions from third parties. Tom holds a Bachelor of Science degree in electrical engineering from Wright State University in Dayton, Ohio, and a Bachelor degree in philosophy and the humanities from John Carroll University in Cleveland, Ohio.|