Build out a LoRaWAN with RPi edge nodes and Gateways?

I’m new to LoRaWAN - reading everything I can… :slight_smile: The duty cycle is a problem for me in the sub GHz range, so I need to build out a 2.4 GHz deployment. I need to connect an SX1280 to an RPi to build an edge node, but I don’t find RPi compatible HATs (or USB connected) for the SX1280 that have antenna connections. Suggestions, please? I’d ideally like to keep the SX1280 connected to the RPi because of all the other things the RPi is handling for me - I’d rather not have to connect a Raspberry Pico or Zero to the SX1280 and then figure out how to have the RPi 4B talk to that to send/receive.

I was going to use the Semtech PCie 3 channel GW module (Semtech 3 channel GW) for the gateways.

Do these choices make sense and work together?

Thanks,

Bret

How much data are you sending from each node, if that is where the duty cycle issue is ?

What sort of maximum distances and environment do you have between the nodes and Gateway ?

Thanks for the response! Yeah, sorry, I didn’t provide any of that info. This is basically a vehicle fleet management problem. I need to send telematics, errors, warnings, etc. from each car every 5-10 seconds. The expected size of a telematics message is around 200-400 bytes compressed. In addition, when the car is idle it’ll receive messages telling it where to go and where to stop - expected message payload is about 100 bytes compressed. I’m using MQTT within the system, so there’s a little overhead with that, too. I’ve seen some whitepapers on 2.4Ghz range and I plan to have gateways every 1,000 meters or so (testing will determine actual placement and number, based on obstructions, etc.). I need real-time, always-on connectivity. Early guesstimates of total bandwidth is about 270Mb per car per month.

I have a working prototype using 5G, but to get the cellular modems certified to work on major worldwide carriers it’ll cost upwards of $100,000 each. The monthly cost of the cellular connectivity is prohibitive, too. LoRa in 2.4GHz seems to be the ideal solution - initial hardware outlay and an enterprise TTN plan for hundreds of nodes is reasonable and sustainable.

Bret

Wrong technology… even on 2.4Ghz?

Probably right technology…

There is a reason for that…

And LoRaWAN/device full cert likely has a cost too…

And whose 2.4Ghz GW’s are you going to run over? TTN (actually TTI for your use case) is back end only - you still will need GW infrastructure, and 99.5% of TTN/TTI is <GHz not 2.4G!

1 Like

Sending lots of information to and from the nodes, does not sound like a typical ‘LoRaWAN’ application.

Hi Jeff,

Why is sending statuses every few seconds via 2.4GHz LoRa the wrong technology? I can certainly see it’s wrong for sub GHz LoRa bands due to the duty cycle limitations.

That’s just for ensuring my device isn’t going to disrupt the entire Cell network they’re responsible. That’s not for monthly cell data.

Does it? Haven’t seen that in any context yet. Do you have a link or anything? Certification by whom? I thought the ISM bands were open and didn’t require certification. I’ll be using SX1280 modules, which I assume (wrongly?) are certified to conform to LoRa requirements already.

I’ll have local ISP backhauls from my own 2.4GHz gateways. Unless I’m mistaken, a gateway built around Semtech’s Reference Design should work, shouldn’t it? Sure, that’s a lot of infrastructure - 8-10(?) gateways around and a lot of edge nodes, but the lower long-term costs over cellular might be better in the long-run.

Bret

Agreed, especially from a standard sub-GHz band standpoint where you get 30 seconds a day of uplink time. Is there another WAN technology you’d apply here? The vehicles I’m tracking have their own power source, so low power isn’t a consideration. I need wide area, always on, near real-time connectivity (i.e. < 1-2 second latency) that’s not going to force me through months and $100,000’s for certification (and then pay thousands a month for bandwidth in addition for all the SIMs I’ll need).

Bret

5G seems a reasonable suggestion.

Just because you want a much much cheaper option, does not mean there is one available that is legal.

I have done a fair bit of 2.4Ghz LoRa testing both in urban and line of sight situations, and my initial thoughts are that 2.4Ghz LoRa would not have the range capability you want at the data rates you want to use.

But before you go any further, why not do some simple range checks, in a typical situation you want to use. The 2.4Ghz LoRa modules are low cost after all.

Yeah, I’ll definitely have to look into the legality of what I’m proposing, but I haven’t seen limitations yet. If, like duty cycle limitations, those legal issues exist, it’ll certainly reshape this in my mind. I’ve seen posts here about people using the free version of TTN like I’m suggesting, but I’m ABSOLUTELY willing to get a full Enterprise account (or more, if required). I understand there are servers to maintain and I’m not trying to freeload. :slight_smile:

That’s kind of why I posed the initial question - yes, the bare modules are inexpensive, but I can’t find boards with the SX1280 embedded that are compatible with an RPi that have an antenna connection except something like this.. Ideally it’d have GPS capability, too, like all the Simcom modem boards I’ve been using to test this in 5G. I’d love suggestions about SX1280 RPi compatible boards. I’m searching everywhere and am shocked I can’t find many.

And simulating an urban environment will be challenging… I guess I could build a battery powered node and a battery powered gateway (minus network connection to TTN since I’ll be mobile) and get some help to carry them around the city to test various scenarios.

Bret

LoRa devices are LoRa devices there are no reception differences between nodes and Gateways.

Put a simple Arduino and SX1280 module as a transmitter sending a packet every second or so and place it, as the transmitter, where the Gateway would normally be.

Put another simple Arduino and SX1280 module as a receiver in the vehicle, and have it flash\beep when it receives a packet. Drive around the city and see where you receive packets and where you do not.

Range testing the SX1280 setup is really very easy.

Before embarking on this project, you really should have some idea as to whether 2.4Ghz LoRa is suitable, within legal power limits.

Thanks - great plan! :slight_smile:

Yes, I’m trying to talk this over with TTN support now to see if their Cloud Plus plan will allow this, particularly from a bandwidth perspective.

Do you mean that I could inadvertently transmit at a certain power level that might be illegal in certain localities and I should do range testing at legal power levels to determine how many gateways I’ll need or whether it will work at all in my urban environment? That seems like a reasonable thing to do. Thanks!

You need to check the power level that is legal for your part of the World.

Why? But also more importantly, as we still lack some overview of what you are trying to achieve, what & where are these cars and how the heck are they so likely to change that updates need to come in every 5-10 seconds.

That’s a whole heap of data - every 7.5 seconds x 300 bytes expanding to say 500 = ~4K, that’s huge even for a car, not so much for a SpaceX rocket.

You can’t push MQTT over something that doesn’t provide a TCP transport layer.

Frankly this is total :ox: :poop: - there are many many many vendors of fully certified global cellular modems - some of which I use to talk to satellites with - so off planet as well. I know they are globally certified because I’ve got them (UK), my client has them (Australia) and the satellite provider supports them (USA).

Cellular base stations handle hundreds of streams of prioritised voice data and thousands of streams of just data. How so you think your special modems may disrupt a cell network? How many cars are you having in an area. And why can’t you run your own 5G base stations?

I think Jeff was just guessing. No, wait, maybe he’s done this enough that he knows stuff.

If you think you need to get a cellular modem certified, why should you not get your LoRa devices certified?

There are several sorts of certification - FCC certified for radio emissions, LoRaWAN certified as conforming to the specification, general electrical safety issues. The SX1280 modules aren’t certified to confirm to LoRa requirements because, and this is a bit meta, they are designed by the people that invented LoRa, so they don’t need to certify the modules, they can but only exist to be a LoRa radio. More Schrödinger than Heisenberg, but definitely they exist therefore they are LoRa.

You don’t certify a band - but if you want to go splat over the top of it with lots of traffic in an area you will annoy other users who will invite the authorities to visit you, not that they can do much about it. But what the other users can do is get a few random rogue devices and broadcast Karaoke so much your system grinds a halt. ISM bands are shared. Everyone has to play nicely or it falls apart.

Is this a rhetorical question? As above, Semtech who are the sole source of all things LoRa design a 2.4GHz gateway around the chips it designed and …

And sure you can make the reference design. But then you’ll need the firmware to run on the chips, and that isn’t on GitHub.

8-10 gateways are a small test setup.

So, given it’s 2.4GHz with much less range, I’d guess maybe this is a command & control system for a small lot that parks the cars for you?

Plus, LoRa is NOT LoRaWAN. LoRaWAN has a whole backend infrastructure with a round trip time that runs to seconds - the protocol is just not suited to pumping data at a server with a fast response time - the system is asynchronous.


At present it looks like you’ve read the flyer and then the marketing brochure and you’ve found your solution and you really want it to be THE ONE to solve your technicals. It may well appear to be cost saving, but it may not be up to the job but we won’t know without knowing more. You are likely to get 10’s of $1,000’s of free advice here if you can tell us what you actual application is - the missing part is what’s up with these cars that they need so much data to be transmitted - and over what range - and then we can advise.

Thanks for the reply, Nick! Wow, lots to get to… :wink:

The “cars” are actually autonomous electric “pods” on a track. Check this link, please: http://theptsolution.com/. We hope to have thousands of these on tracks (“ribbonways”) running around places like NYC, Dubai, Sao Paulo, etc. We want to get GPS info, charge level, drive status, etc. from each pod so we can take them out of service when the charge level gets low, a drive fails, people report an issue (cleanliness, malfunction, etc.), or anything else.

Yeah, I do use MQTT in other aspects of the system, but I shouldn’t have included that comment here, it’s irrelevant for LoRa. My monthly bandwidth estimate should be closer to 207MB.

I had a call with Verizon’s IOT sales team and that’s what they told me directly. Even if I use a SIM vendor like RedPocket or others, they said I still need to get my entire device certified or they would start kicking them off their network. My POC uses a Simcom 7600A-H, which isn’t certified with Verizon, but I could easily switch to another cell modem module that is (i.e. Simcom 7676G). They said even if I used a certified cell module I’d have to get my ENTIRE build certified so they can protect their network. I’m open to ANY solution that satisfies my messaging needs and doesn’t require a costly, slow certification.

Huh? This is an option?? Seems like that would DEFINITELY need certification so it doesn’t clash with cell networks.

I guess I naively assumed that, since there’s no real authority over LoRa, anything that conformed to LoRa protocols would be allowed. Further, I assumed (wrongly?) that I wouldn’t have to get an RPi along with an SX1280 certified by anyone.

The HAL and packet forwarder are available here: Github link.

Understood. I said “real-time”, which might’ve overstated my requirements a hair. When a rider summons a pod via our app, waiting until the next duty cycle would be prohibitive. A 5 second delay is acceptable - no one would use Uber if it took 15 minutes for them to receive the ride request. Likewise, if they press the “Emergency” button in the pod, a few seconds won’t be a big deal - 15 minutes would. I should’ve said “near real-time”.

Please let me know if you need further clarification on my use-case. :slight_smile:

Thanks for sharing your experience and time, Nick! :slight_smile:

Sure, because sales staff are up on technical regulations.

How would they’d know what you are using on their network, a network that allows me to fly in to the US, go to Walmart, buy a Verizon PAYG SIM, put it in a global enabled phone and crack on.

Cellular modules are certified by the manufacturer, mostly at a regional level rather than a provider level.

You keep being surprised about tech that’s to be found by a little bit of reading around. If you buy a 5G femtocell base station you go up to your router, plug it in and then commission it for the backend provider - like Verizon - don’t know if they offer than, but plenty of other providers do. This acts EXACTLY like a cell tower but smaller with lower capacity - it allows you to in-fill coverage.

I’m not sure you’ve quite got the idea. Semtech is the real authority over LoRa because it’s their IP and only they supply the silicon. It’s therefore rather technically difficult to breach the LoRa ‘standard’. If you buy a module you can do whatever you want with it as long as you don’t breach local law.

If you want to use LoRaWAN then there is the LoRa Alliance that maintains the specification which is now an ETSI standard so has independent ratification. Some LoRaWAN Network providers require that the devices are certified - both at an RF as well as a protocol level.

Hmmm, wonder what I was thinking of - I know firmware for something 2.4GHz is by application only - I’ll have to watch some of the TTN Conf. 2022 videos to remember.

However your computer system will know where available pods are - so you don’t have to communicate with a pod to provide user feedback. And a few more seconds whilst the message gets through is neither here nor there as the pod has to start moving due to mechanical inertia and not moving at blinding speeds.

However LoRaWAN isn’t suitable for this as it’s asymmetric - lots of status message up - not so many down. A gateway can’t hear uplinks when it’s transmitting. However having a hybrid system of the status data going up via LoRaWAN makes some sense once you’ve figured out how often you need to send messages about how clean the pod is, hint, if it’s clean and not used, it’s still clean. And if it’s not moving, you don’t need to send location. And if it’s in Baltimore then you can abbreviated co-ordinates unless it goes off the rails.

All that said, I gather new government policy is to discourage electric vehicles unless it’s a Tesla. So all this may be moot.

Looking further at the website it appears to be early days for the concept so, from 30+ years of experience of being given an aspirational brief, there is a certain amount of guessing, which is fine, when creating something new I do a lot of guessing.

As such, the absolute simplest thing to do is to go and buy a pile of stuff, find someone that can wire it up and code it, see what you end up with, do more research, refine & repeat.

When I say someone, initially it should be one person so there are no stupid meetings impeding prototyping whilst people argue about which byte goes where plus the colour of the enclosures.

Along the way you hopefully find out that a location message with speed means the central computer cluster can figure out the location for the duration of that section of track. Which is pretty much how train networks work …

The more this application is described, the less it seems possible.

Lots of data to move around very quickly, and it appears it needs to done in a reliable way, so send and ack on each transmission. Very unlike LoRaWAN.

2.4Ghz LoRa is not ‘long range’ in urban areas (and yes I have tried it) yet with all the uplinks and down links going on and only 3 channels, collisions seem inevitable, even if packets have a short air time, which unfortunately means short range. An avalanche of collisions might ensue.

It seems a significant error to plan the later stages of the project until you have at least some idea if the basic level comms will cover the distances required. If the basic range of the LoRa is not adequate, then the only way you could improve the situation is by installing more and more Gateways.

Yep, that’s where we are. Hi, I’m Bret - the guy who’s researching (i.e. this thread), buying, coding, assembling! :slight_smile: We can’t buy “a pile of stuff”, so I’m trying to make informed decisions before purchasing.

Like I said, I’ve got a POC working with 5G cell connectivity, but certain aspects are cost prohibitive, so I was looking for other options for connectivity.

Thanks for the sanity check! :slight_smile:

Earlier you wrote:

I did some quick checking and found that femtocells appear to only be meant for indoor range extenders (i.e. in a business or a house). Are you aware of outdoor models/brands one could employ easily? Any idea what getting those approved with respect to existing cell networks might entail?

Thanks,

Bret

Obsessed much? Just buy one that’s approved.

But if “a pile of stuff” is out of price range, then a femtocell will be.

No, not obsessed, but those certifications keep popping up as hidden fees, and they’re ungodly expensive! For example - I’m using a Verizon approved cell module now and they still insist on a $100,000 certification! I’m just trying to ask all the questions so I’m not caught off-guard.

Yeah, I was seeing things costing $40,000. That’s out of our range for an experiment, sadly.