Heltec ESP32 Lora GW

Hi all,
I wanted either to deploy a lora-wan GW software or creating something proprietary on ESP32. ESP32-GW with ESP32-Sensor1, ESP32-Sensor2, … ESP32 sensors N
The latter (home made loraWan) is something I wanted to design starting to automate and manage the comunication between my sensors: I think it will not take so much time to cope with a limited scope.

A very simple protocol to discover the GW, addressing the enpoint and handle a sort of transport layer protocol.

Isn’t there anything I could reuse?

Maurizio

Yes, LoRaWAN.

Can I install it on ESP32/any low cost board?
If not, it’s not a solution for the use case I have in mind.

So are you after developing some form of simple or lightweight LoRa server, not based on TTN and LoRaWAN ?

There is a Single Channel Gateway software implementation for ESP8266/32 that works, although as any single channel, not fully compliant with LoRaWAN specifications (because one channel is not sufficient). However, to experiment is not bad (many of us went through this), although you will soon cross its limits, in particular if your nodes send often, since they share one channel and one SF - loss of packets etc.
A real gateway cost starts from the equivalent of 3-4 ESP32-LoRa boards (the TTIG), so in a real use case, gateway price is not a real constraint.

Yes, something I can use at least to handle LORA sensors in my private network.
I have been defining the finite state machine for the sensors/gateay to implement it .
I do think it can work and run effectively on ESP2

So, likely a moderator will stop the thread, not being TTN :slight_smile:
Anyway, I would still consider the same: single channel gateway on a ESP32, and if you want your own network, just install the TTN open source stack on your server. Reaching the same level of reliability and also security is very, very hard. Software is already made, no need to reinvent the wheel.

1 Like

Is this the library you are rrefering to:

Said that, the packet loss you mentioned in your post can be easiliy addressed with a simple retrasmission mechanism and up to 10 sensors in a private network a lightweight apporach can perferctly fit.

Single channel gateways have been done, but they dont provide a full service to others in the community yet they are using the TTN infrastructure for free.

There may well be a demand for an application that implements a private server and its own internet communications setup, but this is a TTN forum, not a general purpose LoRa one.

And if your careful the re-transmissions dont cause you to exceed the legal duty cycle limits.

Spectrum is a precious resource and designing a system/implementation with deliberate intent/need for re-transmission if frankly anti-social IMHO! :wink: Better to use a more robust mechanism in the first place - as highlighted the cost of a full LORaWAN GW isnt much more and if using it for e.g. TTN then has social benefit in being part of an open deployment…remember ALL your TX & Re-TX will be ‘seen’ by other local LoRa users and will act as potential increase in traffic and raise noise floor…you may be in a remote location not impacting others then get away with it but it doesn’t scale (what if a friend living in an urban environment with other LORa/LORaWWAN systems around ‘likes’ your implementation and wants to replicate?) and posted as a caution to others thinking the same…just my 2 penneth worth! :_)

Single channel GW’s also ‘look’ like std GW’s and may confuse users or potential deployers whilst only serving to cause lost packets and confusion - not a great help when debugging a deployment…a wider question to TTN Forumites in general: dare I suggest a personal view that TTN has matured enough that single channel GW’s should be blocked as causing more trouble than they are worth (note deliberately provoking debate on this vs playing a Troll! - may be worth having a discussion thread to canvas opinions as beyond just this device topic?). When GW’s cost >£1k there was excuse for development and experimentation but with wide availability of <£300 & even several now <£100/150 GW’s do we need to move on and reset limits for use?

1 Like

what’s the cheapest solution to deploy a loraWAN GW (Standard) on Raspberry?

The cheapest is not on raspberry but The Things Indoor Gateway that altogether costs, now, less than a concentrator hat for RPi ( 80€ vs around 110€). However, the latter solution could be more flexible.

Besides the TTIG called out by @UdLoRa (I’m evaluating a handful on long term test right now) if you can look away from RPi systems an alternate might be the RAK 7258 Micro GW RAK7258 - micro indoor gateway which seems cheap enough and if similar to what I have seen with other RAK platforms shoudl be worth considering - its probably next on my scaled eval list (Q3 tasks) after I finish with the TTIGs and some of the other RAK units I have deployed or am evaluating inc the RAK Outdoor/Enterprise GW eval that is way behind schedule!. Having 100M wired Enet as well as Wifi plus RS232 Console access makes the 7258 potentially more accessible and flexible than the TTIG for not a huge amount more money :slight_smile:

1 Like

I agree with Jeff, many features more with not much money (by the way, SX1301 vs 1308, slightly more sensitive). However, aiming at super-cheap, and replacing an ESP32 single channel, TTIG is more than sufficient.
Maurizio: did you just check for availability of public gateways around you? Some regions are well covered.

Thanks all for the feedback.
I have already several sensors in my home running on Arduino, ESP8266, ESP32 sending via MQTT messages to Mosquito running on Raspberry still in my private network.
Sooner or later I will purchase and deploy a standard TTN LoraGW (the RAK7258 seems interesting), in the meantime to cope with a specific use case, that is, control a sensor close to my house where I do not have wifi coverage, LORA point to point (with some customizations :)) is more than enough.

Sorry for the stupid question…is it possible ask to a standard lora GW to connect with a specific MQTT server and forward specific MQTT publish/subscriber received from a client? I have checked the documention: it is supported.

Maurizio

That depends on the software running on the gateway, but in most cases that will not work. For LoRaWAN you need back-end software in addition to the gateway. There are gateways that run some back-end software as well but those are the more expensive ones most of the time.

I would like to have a feature like this:

  • broadcast to verify the gw with related capabilities (MQTT)
  • if they support MQTT (natively), send the name of the MQTT Server and handle transparently the connection and the publish/subscribe handling