Long Range Low Power networking

Started by rtomkins 6 years ago5 replieslatest reply 6 years ago283 views

I started to read about The Things Network, which led me to LoRaWAN and then Link Labs and their Symphony product cropped up on the RADAR.

Golly, there seems to be a an ever increasing number of players in the field of Long Range Low Power networking, just a few minutes ago, along came a reference to Neul.

I read a paper produced by Link Labs that Symphony Link is a better choice than LoRaWAN, and the paper is very compelling, for a large number of reasons.

Would anyone using these forums have any comments/thoughts about LoRawWAN, possibly some folks are using something and can jot down the pitfalls, or gotchas.

I was impressed with the The Things Network, and their ability to set up a WAN across Amsterdam with ten gateways. As I read that, I started to think of how one could monetize the use to create a sustainable infrastructure.

Amongst everything I have read so far, I have come to understand (AM I CORRECT?) that the various solutions being offered are strictly a firmware implementation on piece of hardware that they want to sell you, as to the protocol stack, that's proprietary and not part part of the deal.

Thanks in advance for insights you care to share.

Richard Tomkins

[ - ]
Reply by iansMarch 25, 2018

Hi Richard,

We've not done much with LoRa but it seems to be very much a proprietary solution in that everyone has to use the same basic chipset (at least I've not heard about it being cross-licensed). The advantage obviously is that you can own both the base stations and end nodes for a very low price. The disadvantage is that you then have to manage the basestations and keep adding them to your network as your solution grows in scale.

I like the idea of the Things network, but it does leave you at the mercy of a variety of different providers if you piggy back on their basestations - no big deal for a little proof of concept deployment but it could be a major sticking point for a proper commercial rollout.

Other alternatives such as SIGFOX are a bit more open in some ways - you can use your own chipset (could just be an 8051 with an RF front end) and port the stack yourself, but then you do need to get it certified (the minor detail that a lot of people forget about when dealing with comms solutions!) which will be a few thousand. You are also tied to the national provider but then at least someone else is dealing with the headache of managing the field population of basestations.

The other emerging technologies such as NB-IoT , LTE-M etc. will probably be along the same model as SIGFOX.

I think it is a case of "horses for courses" - if you want a solution you can roll out across the country or across a continent then with LoRaWAN, Symphony etc. you're going to need really deep pockets. Alternatively if it is a relatively local system then these would be a better choice than SIGFOX et al.

BTW - if you're think of any sort of mobile asset tracking using any of these LPWAN solutions, I'd recommend a lot of up-front testing to make sure they will do what you need as there are a lot of pitfalls for the unwary...



[ - ]
Reply by rtomkinsMarch 25, 2018

Lot's of reading and many late nights later, the following PDF from Microchip, presented in 2015 gives a good overview of LoRaWAN.

Getting Up and Running with LoRaWAN™ Long - Range Networking

My reading so far says Semtech own the IP for the Radio which is what forms the MAC layer. They created an open source network stack.

Lot's of different companies have developed raido modules using the Semtech chip and some have created LoRaWAN network stacks, and these are available as libraries, such as that offered bt Multitech through embed. All very nice, but, being stuck using embed and the embed OS, and having to use the embed online tools to compile your own code, may be a serious limitation for some.

The LoRa Alliance appears to be overseeing LoRa, specifically a certification body that has their hands out for a membership fee. Frankly, I don't see a need to join, they are not really promoting the technology on their website, they are promoting certification.

Semtech appears to have some support in place, which appears to be geared to selling chips.

There is a lot of disjointed information all over the internet, and I am beginning to see how The Things Network lucked into something by assembling the pieces, putting it together, packaging it and selling it.

As an example of the disjointed pieces, Stackforce claim that they now officially maintain the open SOurce code from Semtech. Yeah, right, no links to this official repository anywhere on their website. I stumbled on this, Documentation of the API, but have yet to find any code, but it looks promising and appears to be well documented.

I see a lot more searching, reading, indexing, and many more late nights ahead.

[ - ]
Reply by gustavovelascohMarch 25, 2018

I have been working with LoRa and LoRaWAN for some months. It is important to clarify that both, LoRa and LoRaWAN are different: LoRa is the modulation (PHY layer) and Semtech owns the IP. LoRaWAN is a MAC layer specification over LoRa.

LoRaWAN Network consists of three elements: Devices, Gateways and Network Servers. Devices are registered to applications and gateways are just that, gateways. Devices should have a key for the network and a key for the application in order to exchange data. The gateways are registered in the network and they will receive packets for any device from any application.

I have used devices like the ones from Microchip (RNxxx series) and from IMST (im880x), but there are more options out there. They are basically a module which contains a processor unit + Semtech radio interface (Sx1272/6). Some modules are certified as LoRaWAN class A or C (I haven't seen any device certified as Class B).

The gateway or concentrator is the one who forward packets from devices to network server and viceversa. It commonly uses Semtech Sx1301 which support thousand of devices (?), but there are some gateway implementations based on a simple Sx1272/6, though. Some commercial gateways are from Multitech and Kerlink. There are also raspberry-Pi-based gateways using iC880 concentrator from IMST.

The network server is in charge of handling incoming data from devices (through one or more gateways) and dispatching it to the appropiate applications. This is what The Things Network and Loriot offers.

[ - ]
Reply by rtomkinsMarch 25, 2018

Many thanks for the information.

I continue reading and exploring technologies and hope to update my original note here sometime this week.

[ - ]
Reply by rtomkinsMarch 25, 2018

We're using two  STMicroelectronics STM32L0 Discovery kit LoRa operating in OTAA using the LoRaWAN software expansion for STM32Cube (UM2073) code.

We're using a Sandbox Electronics LoRaGo PORT mounted to a Raspberry Pi 3 running the Semtech Lora-net code.

We're using the LoRa Server, open-source LoRaWAn network-server.

It took eight weeks to get all the bugs out, not having up front knowledge of this stuff.

There were many undocumented stumbling blocks encountered and jumped over. 915 MHz operation does not appear to be a consideration in most documentation, the majority of which is geared towards Europe. The ISM band is different between Europe and Canada and that means a lot of documentation assuming Europe as the operating region is not helpful, simply misleading. Semtech's json file global_conf.json.US902.basic readily shows the same European assumptions, there are no tx_lut entries to manage the transmit power levels. The supplier, Sandbox Electronics should have run some tests and characterized the radios, but again, they assume Europe.

So, we have been able to receive and decode payload data from both End-nodes and have also been able to transmit to both End-nodes and have them change controller states.

When we get a chance, we may write something up on all this, right now we are prepping for a presentation.