Why We Operate At India’s Frequencies

Why do we, a national wireless network operator in New Zealand, operate using LoRaWAN frequencies meant for India? Why not just be like everyone else and make the “obvious” choice of operating at the frequencies designated by the LoRa Alliance for New Zealand (915 – 928 MHz)?

The short answer is noise, customer convenience, and strategy.

Noise

LoRaWAN networks use Semtech’s superb, patented radio modulation technology LoRa  at the physical (RF medium) layer. The result is an ability to reliably operate wireless networks in the public, unlicensed radio spectrum. By eliminating huge licence fees for exclusive use of radio spectrum, LoRaWAN networks globally are offering low cost wireless connectivity and enabling the promise of billions of connected devices.

However, LoRaWAN networks share the same public radio spectrum with many other wireless users. For a LoRaWAN network, this is “noise” and a source of interference. The more popular and usage of a frequency band, the more the noise and interference.

Under New Zealand regulations, the 3 possible public spectrum choices under 1 GHz are 433.05 to 434.79 MHz, 864 to 870 MHz, and 915 to 928 MHz. The first one is not practical which leaves the other two. Have a look at a typical snapshot of what the two bands look like:

868 MHz Radio Spectrum

920 MHz Radio Spectrum

Clearly there’s a lot more noise in the 915-928 MHz spectrum.

Now, LoRaWAN networks are specifically designed to operate in noisy RF environments. But, project the current radio spectrum forward a couple of years and factor in millions of new devices using the “obvious” frequency band. The congestion will be a lot worse.

That’s not to say a LoRaWAN network can’t function in that future situation. It is specifically designed to do so. However, if a better alternative is available, which it is in the form of the 864 to 870 MHz band, that makes better sense.

Given both bands are public radio spectrum, there is no guarantee of the future situation. Still, going with a currently uncongested band seems the right choice to one that is already noisy and likely to get worse.

Customer Convenience

We originally launched with the 865 – 867 MHz band. An earlier blog post provides more details of the reasoning.

While there was a clear preference to adopt the LoRaWAN EU 868 band, this proved impossible as its three default frequencies are higher than 868.0 MHz. In New Zealand, it is practically impossible to operate above 868.0 MHz even though the public band itself extends to 870 MHz due to duty cycle and other limitations.

This decision, however, meant we were the only ones in the world operating at that particular frequency band. The consequence was customers, and we ourselves as users of our own network, had to buy EU 868 modules and configure them to work with our specific frequencies. That had many advantages, such as availability and scale pricing, but also some disadvantages.

In a small number of cases, such as the MultiTech’s mDot with locked binaries, it simply wasn’t possible. In other cases, there was the one-time hassle of channel configuration. Take a look at the configuration instructions for the popular Microchip RN2483 to see what’s involved. It’s not hard but is difficult to scale. Further, since these modules are designed for the EU default frequencies in mind, some of them had limitations due to their firmware which didn’t allow OTAA join.

Things were fine but there were a few niggles.

Then, a couple of months ago, India’s LoRaWAN frequencies choice was known and, by sheer coincidence, it was broadly the same 865 – 867 MHz.

The benefits of tweaking our band plan to match that of India exactly are huge. Our customers could simply buy LoRa modules and devices meant for the Indian market off-the-shelf. No more configuration hassles. As the Indian IoT market is growing rapidly and will be globally significant in size, most manufacturers will offer India’s band plan as an option. This takes care of scale pricing.

Further, as the Indian band plan is an option in the standard LoRaWAN software code, customers building their own devices have to simply pick it rather than modify the code in any way.

A bonus is that the Indian LoRaWAN band plan has no duty cycles, just like ours. This makes our band plan exactly the same as India’s.

India’s band plan has three default frequencies- 865.0625, 865.4025, and 865.9850 MHz. So we had to slightly change five of our eight channels to allow for their three default ones plus two to even out the channel spread.

No wireless network operator likes to change its operational frequencies but the benefits were so clear and so big that we had no hesitation to do so as quickly as was feasible. On 1 June 2017, we completed the switch over to our new (Indian) band plan.

Strategy

Both originally and as borne out by subsequent developments, there has been a strategic underpinning that reinforced the noise and customer convenience angles.

Looking forward a couple of years, every analyst predicts the cellular companies will be taking a big or major share of the Low Power Wide Area Network (LPWAN) connectivity market with NB-IoT and LTE Cat M1 standards. LoRaWAN will compete directly with NB-IoT.

There are many dimensions to this coming competition. No doubt I will write more about this as NB-IoT is commercially rolled out next year and becomes clearer but for now will only focus on one aspect.

One of New Zealand’s major cellular carriers is globally positioning NB-IoT as a “premium service” compared to LoRaWAN (and Sigfox). That’s likely to be used to justify premium prices. For example a recent local quote is, “There are many IoT networks available now but we think NB-IoT is a premium technology choice that is worth waiting for.”

Why will NB-IoT be a “premium technology choice” compared to LoRaWAN and worth paying more for? Certainly there is nothing in their respective technical specifications that justifies this characterisation.

We believe that one of the justifications that will be advanced is LoRaWAN’s use of public unlicensed radio spectrum compared to NB-IoT’s use of licensed spectrum. This will be used to promote NB-IoT as more reliable, hence “premium” and worthy of premium pricing compared to LoRaWAN.

While this overlooks the fact that LoRaWAN is specifically designed to operate in noisy radio environments and has in-built reliability enhancements like confirmation of transmissions, we expect this argument to get some traction globally.

This may be an issue for our LoRaWAN competitors in New Zealand using the “obvious” choice of 915 – 928 MHz.

However, this is not true for us as our choice of frequencies mean we also operate in an almost noise-free radio environment. Except that we don’t pay licence fees for that privilege of exclusivity. Our operational performance and reliability matches those of the cellular companies. The “premium” NB-IoT positioning and pricing then becomes highly suspect in relation to our LoRaWAN network.

It’s quite possible that this will be drowned out by the marketing muscle of the big cellular companies. That doesn’t make it less true.

Summary

We believe we are in a sweet spot of using unlicensed public spectrum that is noise free while still being able to use LoRaWAN standards and global scale pricing of modules and devices. This provides us critical competitive advantage in providing low cost and reliable wireless IoT network connectivity to our customers nationally in New Zealand.