Choosing IoT connectivity technology needs careful consideration of multiple technical and commercial factors linked closely to individual use cases.
Different applications favour different technologies so what the product is designed to do is a critical factor in the decision. Features and benefits often conflict and need to weighed carefully in the decision so these articles distil the key characteristics that define an IoT connectivity technology to help you identify those important to your particular use cases.
Here are the parameters that we think define the strength of an IoT connectivity technology:
- Quality of Service
- Battery life
- Proprietary vs Standard
Last time out we considered reliability. Now, over the half way point in the series, we’ve reached one of the critical parameters for IoT use cases where the end point is battery powered – energy consumption.
In previous articles I have grouped technologies into three categories – LAN/PAN at one end, 3GPP at the other and LPWAN technologies operating in unlicensed spectrum sitting between the two and borrowing some characteristics from each.
No single technology is right for every product or use case and these articles are not intended to advocate a particular commercial proposition. The Internet of Things market covers a diverse range of use cases and each is associated with different priorities. It is segmented and some technologies suit some segments better than others. In this article, looking at energy consumption, if your end device is powered from a mains supply you can for all practical purposes ignore energy consumption altogether – it is simply irrelevant to your decision. If this is you, you can stop reading now.
You’re still here so we can assume that connecting to a mains supply is not technically possible or commercially feasible for your use cases. For these applications energy consumption quickly rises toward the top of our list of criteria for judging connectivity options. Most use cases simply will not support a maintenance schedule that calls for frequent battery changes – an end point’s in-service life needs to be measured in years.
The formula for calculating battery life is actually quite complex. It is fundamentally impacted by both the power consumed during downtime (or “idle”) and the power consumed when transmitting data but we need to look at a lot more than simple headline numbers like power consumption during transmit and idle. Let’s now explore this in a little more detail.
The length of time a device spends idle is often a compromise between longer battery life and the ability to contact or page a device when required. The optimal balance will depend on the application – for some a daily window for device communication will be sufficient, for others there may be a need to contact a device in just a few seconds at a time that cannot be predicted or scheduled in advance. The best systems do not impose any particular idle time but allow the devices to make the decision themselves based on pre-programmed or downloaded information.
Once out of idle mode and checking paging channels it is important to minimise receive time. Some solutions require devices to listen to an entire system information transmission and paging channel to check whether anything has changed. Others are more intelligent, providing flags at the start of the reception process which help the device determine whether it is worth listening to the remainder of the frame. Having multiple paging sub-channels also means the device only has to listen to one, reducing reception time.
Each bit transmitted brings the battery one bit closer to its death
Transmitting is hugely more energy intensive than receiving and significant reductions in energy consumption can be made with careful design of the Tx regime. There are obvious steps such as selecting modulation formats that do not require linear power amplifiers and hence enable operation in efficient parts of the power curve. But far more important is to reduce the number of bits that need to be transmitted. Essentially, all systems face the same amount of path loss and overcome this with various mechanisms that lengthen the transmit time per bit such as the low data rates in ultra-narrowband or the multiple spreading code bits sent per bit in spread spectrum solutions.
All IoT technologies use the same amount of energy per bit
Where they differ widely is in the number of bits actually transmitted. For example, some solutions automatically transmit the same message several times (e.g. three) to increase the probability of successful reception. Obviously this regime triples the power drain for transmission, or described another way, will typically shorten the battery life by a factor of three. A system that, all other variables held equal, had a battery life of, say, a decade will need new batteries after just three years. And that’s significant enough to make a proposition unfeasible.
Other solutions require longer packets with long addresses or other overheads. This can easily incur a 10X increase in the number of bits transmitted compared to a highly optimised solution.
It is relatively easy for any LPWAN solution to claim 10+ year battery life – and if a device has prolonged idle periods with very few transmissions this can probably be achieved. But we all have the personal experience of how manufacturers’ claims for battery life can often be hugely overstated in the real world and how very low power devices can be annoying slow to respond. It is difficult at this stage to verify whether a 10-year battery life under “typical conditions” can actually be met – until networks are loaded to near capacity and devices have been in the field for some time there will be limited evidence to go by. But finding out halfway through the anticipated device lifetime that 50 million units all need an unexpected battery replacement will be the kind of costly and disruptive surprise that most users will not appreciate.
When picking an LPWAN technology it is worth asking whether the designer really understands how to deliver extended real-world battery life. Did they minimise the number of bits transmitted? Have they allowed the device to optimise its balance between battery life and the ability to be contacted? Have they selected radio parameters predominantly on the basis of their ability to be energy efficient using non-linear power amplifiers? Have they avoided blindly repeating messages or long address fields? The Weightless-P designers certainly took all these factors into account.
Many IoT applications terminals will be powered by batteries leading to the need for low energy consumption. All LPWAN technologies operating in licence exempt spectrum are low power but Weightless-P uses a number of techniques to offer best in class power consumption performance and consequently very long battery life. Overall energy consumption is a factor of both the power used during transmission and the amount of the time that the transmitter is active – the duty cycle. Weightless-P uses GMSK and offset-QPSK modulation schemes which deliver optimum power amplifier efficiency. Offset-QPSK modulation is also inherently interference immune and using Spread Spectrum for improved link quality in busy radio environments minimises the required transmit power. A limit of 17dBm in ISM spectrum means that terminals can operate from coin cell batteries. Adaptive data rate also permits minimal transmit power for nodes with a cleaner signal path to the base station so maximising battery life. And since a terminal device will nearly always spend a very large proportion of the time in an idle state, the power consumption in this mode becomes critical. Weightless-P consumes less than 100µW when inactive.
What Weightless offers:
- GMSK and offset-QPSK modulation for optimal power amplifier efficiency
- Interference-immune offset-QPSK modulation using Spread Spectrum for improved link quality in busy radio environments
- Transmit power up to 17dBm to allow operation from coin cell batteries
- Adaptive transmit power and data rate to maximise battery-life
- Power consumption in idle state when stationary below 100µW