Posted on Friday, March 8, 2019 by Gerben Kuijpers
The good, the bad and the unknown: 6 key learnings from working with NB-IoT, part I
18 months into our work on developing an NB-IoT solution for smart electricity metering, we have made a lot of headway. So, what have we learned so far? The short answer is: A lot! And frankly, that includes some things that came as quite the surprise. In a series of blog posts, we’ll now dive into topics like connectivity, coverage, network capacity and much more and share our learnings from numerous trials and field tests, so read on to find out more about the reality behind the specs.
As we’ve previously stated, NB-IoT is definitely a contender for the communications technology of the future in the realm of smart electricity metering. That said, we have seen an unjust inflation of expectations for its merits in recent years. There is simply a mismatch between the specs on slideware and the performance that can be obtained for real-life use cases with many devices. Utilities can easily get the impression that NB-IoT is the answer, no matter what the question is, but reality is – as reality tends to be – a lot more nuanced.
At Kamstrup, we started working on our NB-IoT solution for smart electricity metering in the summer of 2017. Since then, a series of lab tests and trials based on typical electricity meter scenarios have given us the experience and knowledge we have today and continue to build on.
What we did
One deployment involves 50 meters in cooperation with a major international mobile network operator. Meters were installed in and around a large European city in carefully selected locations, including the homes of some of the operator’s employees. This makes up a very realistic environment for metering use cases and as a result, the data we get is highly relevant. The meters are connected to our head-end system, which allows us to collect meter data as well as signal data to learn more about reading performance and the connection quality.
Collecting and interpreting data on how the solution performs in different locations and set-ups enables us to determine the use cases where NB-IoT is ideal and, just as importantly, identify its limitations and the cases where it makes more sense to use one of the other communication technologies available.
Among the next steps is a pilot project with meters used in live operation in Q1 and Q2 of 2019 together with Finnish electricity company Herrfors, and we will of course keep you up to date on the development and new learnings right here on the blog.
What we learned
So, let’s get started with learnings 1 and 2:
1) Basic connectivity is solved
Today, we have reached a level of plug & play, where all that is required is to insert an NB-IoT enabled SIM card and power up the meter. During the last 9 months, no expert involvement has been needed to get the meters to connect to new networks in new countries, so we consider this basic challenge solved.
2) Coverage exceeds expectations
As we’ve previously stated, NB-IoT is definitely a contender for the communications technology of the future in the realm of smart electricity metering. That said, we have seen an unjust inflation of expectations for its merits in recent years. There is simply a mismatch between the specs on slideware and the performance that can be obtained for real-life use cases with many devices. Utilities can easily get the impression that NB-IoT is the answer, no matter what the question is, but reality is – as reality tends to be – a lot more nuanced.
At Kamstrup, we started working on our NB-IoT solution for smart electricity metering in the summer of 2017. Since then, a series of lab tests and trials based on typical electricity meter scenarios have given us the experience and knowledge we have today and continue to build on.
What we did
We developed an NB-IoT plug-in module for our OMNIPOWER electricity meter in the autumn of 2017 and today, there are more than 100 NB-IoT enabled meters deployed as trials in live networks in seven countries across Europe.
One deployment involves 50 meters in cooperation with a major international mobile network operator. Meters were installed in and around a large European city in carefully selected locations, including the homes of some of the operator’s employees. This makes up a very realistic environment for metering use cases and as a result, the data we get is highly relevant. The meters are connected to our head-end system, which allows us to collect meter data as well as signal data to learn more about reading performance and the connection quality.
Collecting and interpreting data on how the solution performs in different locations and set-ups enables us to determine the use cases where NB-IoT is ideal and, just as importantly, identify its limitations and the cases where it makes more sense to use one of the other communication technologies available.
Among the next steps is a pilot project with meters used in live operation in Q1 and Q2 of 2019 together with Finnish electricity company Herrfors, and we will of course keep you up to date on the development and new learnings right here on the blog.
What we learned
So, a year and a half down the road and with a lot of NB-IoT enabled meters up and running in live networks, we have some important learnings to share. This blog series will break down what our experience has taught us in six different areas, and today we kick it off with the first two. But before we begin, a small service announcement: In the bottom of this blog post you’ll find a small NB-IoT dictionary with some of the technical terms explained. If you don’t find the word you’re looking for, let me know in the comments section and I’ll help you out.
So, let’s get started with learnings 1 and 2:
1) Basic connectivity is solved
Stable communication is a prerequisite for fulfilling the daily needs of any utility, but in the beginning, just getting an NB-IoT connection established was a challenge. We experienced all kinds of weird behaviour, including chipset crashes, unexplainable loss of data and the need to configure the chipset in different ways for Radio Access Networks (RANs) from different vendors. This resulted in a lot of communication with modem vendors, chipset vendors, mobile network operators and the telecom equipment manufacturers. Over time and after numerous chipset firmware updates and newer versions of the RAN software, the situation slowly improved.
Today, we have reached a level of plug & play, where all that is required is to insert an NB-IoT enabled SIM card and power up the meter. During the last 9 months, no expert involvement has been needed to get the meters to connect to new networks in new countries, so we consider this basic challenge solved.
2) Coverage exceeds expectations
Without coverage, performance is irrelevant, so we have carried out various tests to assess both coverage and penetration; areas where NB-IoT was expected to perform very well. Actually one of the key selling points of NB-IoT is a link budget that is improved by up to 20 dB compared to conventional cellular technologies.
One such test was performed in our own basement here at Kamstrup, where a new building with thick concrete walls and triple-pane windows provided the ideal conditions for testing deep indoor coverage. As seen on the image, NB-IoT enabled meters still have a good and stable connection in a basement area where 2G, 3G and regular 4G (Cat 1 and above) connectivity is unavailable. We have obtained similar good results for deep indoor coverage in field tests.
The improved link budget for NB-IoT can also be used to increase the communication range, when compared to 2G, 3G and regular 4G.
One such test was performed in our own basement here at Kamstrup, where a new building with thick concrete walls and triple-pane windows provided the ideal conditions for testing deep indoor coverage. As seen on the image, NB-IoT enabled meters still have a good and stable connection in a basement area where 2G, 3G and regular 4G (Cat 1 and above) connectivity is unavailable. We have obtained similar good results for deep indoor coverage in field tests.
The improved link budget for NB-IoT can also be used to increase the communication range, when compared to 2G, 3G and regular 4G.
To test this, we have done mobile tests, measuring the signal levels further and further away from base stations and collected performance data from field test meters far away from base stations. Our findings from the drive tests and other field tests show that for realistic scenarios, a range of 10-12 km between meter and base station is feasible.
Drive test with field test meters: The vertical lines indicate the signal levels along the approximately 11 km long route from the base station to the coverage edge.
In the next chapter of this series, we’ll dive head first into one of the major challenges, namely that of assessing the actual network capacity. We’ll also touch upon roaming issues. Until then, please get in touch via the comments below.
Dictionary
Chipset:
An NB-IoT chipset is the hardware and firmware implementation of the NB-IoT radio protocols. Our plug-in module for the electricity meter uses an NB-IoT modem from a 3rd party. The chipset is a component of the modem that the modem vendor typically sources from yet another vendor.
Link budget:
A way of representing how much a signal can be weakened from sender to receiver before the connection is lost. The higher the value, the longer the range and the better the penetration through buildings/walls, etc. The scale is logarithmic and defined in dB.
Radio Access Networks (RANs):
A Radio Access Network is the part of the mobile network that implements the radio access technology. This includes the network side of the lower layer protocols of NB-IoT.
Base stations:
Base stations are part of the Radio Access Network and are the fixed stations that are used for radio communication with the mobile devices. Base stations for 4G are also referred to as evolved NodeB (eNodeB/eNB).