My name is Fadi and I recently graduated from Technical University of Denmark.
My thesis was about building a fully functioning Off-the-Grid LoRaWAN Gateway with mobile backhaul, that can last 365days in Denmark without any downtime. I believe I have managed to do so. As it looks right now, I am starting a company to build and install my gateway for municipalities in Denmark. They will not be able to buy the product but rather rent the network that I am going to build, specifically in rural areas. What I am looking for is people with experience using and testing gateways to help validate the usability of my MVP and to give some feedback about how to improve the product. The only criteria that I have is that you actually want to use it for something that solves a problem for you that otherwise can not be solved with traditional wired gateways.
Our MVP Gateway includes:
A 50W solar panel
A 100Ah Battery,
An antenna
An electronics box.
Inside the electronics box there is:
A MPPTcharge controller
A Raspberry Pi Zero W
Waveshare sx1303 LoRa concentrator
SIM7080G Modem
TTGO T-Beam for power measurement and to transmit the data back via LoRaWAN
And a bunch of tested code used to reduce the power consumption without compromise functionality.
Currently, I do not believe that there is anything in the market that is equal or lower in power consumption compared to my gateway (maybe I am wrong?). The average power consumption is 1.2W, with peak power of 1.9W. I do believe with the next planned version we can go down to an average power consumption of 0.8W, but I am not 100% sure yet.
Below is a picture of one installed in a municipality in Denmark:
How is this related to the sponsor of this forum: The Things Industries? Will you be using TTN/TTI? If not this message will quickly disappear as of topic…
In collaboration with the municipalities the goal is also to use the multiplexer from Chirpstack to support TTN. This is why we are renting the gateway as a service such that we can use it for creating other networks and supporting TTN.
This suggests attempting to use the GW on atleast 2 LNS concurrently? If so this is frowned upon and should not be done on TTN (Forum Search will give reasons).
All GW’s basically use the SMTC ref designs for Concentrator card implementations (the SX130x bit), with minor tweaks. The Power consumption of each typically therefore very similar. The Control card - the MCU bit (and firmware/s/w load implementing e.g. Packet Forwarder, Network setup/GUI Management etc.) - added to create the complete GW is typically where many devices will therefore vary wrt overall power consumption. And whilst Pi0W is a good choice in that respect afraid to say your implementation therefore very unlikely to be unique or lowest power in that context as SX130x + Pi0W combinations have been around for many years (delivering low power implementations ) - since atleast 2018 from my own personal knowledge and experience with many versions used/posted to the Forum etc. by a number of users. Other solutions have been implemented using alternate control plane/management processing using alternates to Pi0W’s as well…
…what makes your so ‘low power’ or why do you believe it to be a case of
Is it the GW (LoRaWAN RF through to network interface) you consider low power or overall system? Or subsections?
Happy to test and compare, when I can spare some time, if you want to contact me by PM…
this is a very promising research initiative! We have always the issue with the power supply of LoRa Gateways in high alpine areas and every Watt (hour) counts to stay alive during times of bad weather or snow on the solar panel.
Yeah I do some research about using GW for 2 LNS concurrently, it does not look that promising with multiplexer. It seems that another way to go about it is using Packet Broker, or ideally what I am thinking is to have 2 concentrators connected independently to each of their own LNS, this does require a redesign of the hardware and a lot of testing. It will also increase power consumption.
When I am comparing mine, I do so towards other commercially available gateways out there, it seems that there are nothing that I can find that breaks the 2W barrier and most of them still use the older SX1301 and SX1308 chips which are 4-5 times more power hungry in RX mode compared to the SX1302 and SX1303. For homebuild DIY gateway probably there are other people who have done it here on this forum, which breaks the 2W barrier.
Some of the improvement that we have done to make it ‘low power’ is:
Obviously, use a processor that is tailored for our need, no need for quad core cpu’s etc.
use LTE CAT M1, no need of 4G and 5G most gateway can work fine with much lower data speeds
We want to maintain full control over LTE Modem, we use Cmux for UART multiplexing to avoid using the power hungry USB port on the Raspberry pi Zero W that is a power save of around 0.2W to 0.3W, which does not sound a lot, but it really is.
We have tested 5 different MPPT controllers to find the best one that fits our needs, it has all the functionalities that we need, a very low idle current and being ‘cheap’.
Again the 1.2W I am talking about is for the WHOLE gateway setup, not just the Raspberry pi Zero W and concentrator but everything; Modem using LTE CAT M1, MPPT controller with is idle current, CPU and concentrator. Basically if you measure the power from the battery it will read 1.2W. If you choose to use WiFi the AVG power consumption is around 0.8-1W. For the whole thing.
The next version we are going to look into we will probably switch the Raspberry Pi Zero W with either MangoPi MQ Pro or Lichee Pi Zero for even lower power consumption. Hopefully we can push it down to below 1W using LTE CAT M1, but I can’t promise anything.
Is it the GW (LoRaWAN RF through to network interface) you consider low power or overall system? Or subsections?
I am not sure what you mean about this, but I am specifically talking about the hardware, some of the homemade scripts and code for the Cmux.
I got a bunch of people writing to me to help testing it, we are going to build another 10 of them through summer and probably send a few out for testing and feedback.
You are allowed replicate the gateway I made, but only for private use and to support TTN, anything beyond that please contact me. The version that we are working on right now is much better and stable than the one you read about in the thesis, especially for the electronics part.
Your unsubstantiated assertions about it being the lowest power gateway could be forgiven for enthusiasm. But where did you get the information to build it from that you then want to restrict?
Perhaps you should look at the licensing on the sources you’ve used and then apply an appropriate one to your evolution of the body of work that the community has & continues to work on.
The ESP32 module represents an unnecessary power drain, but if you are going to use LMIC-node, created by a forum member, you should at least credit them in your thesis.
There is no information about how to build it, the thesis itself is trying to solve this issue. I basically did a lot of testing with different modules to figure out the lowest power consumption, here is a link for some of the tests I have done: General power consumption - Google Sheets.
The ESP32 is used as a backup module for monitoring the gateway, this only works when there is another gateway close to it to relay the data back or when there is no problem with the gateway itself. The ESP32 runs on its own internal 18650 battery using deep sleep that allows it to last more than 2-3 month without the use of the big AGM battery. The AXP192 of the TTGO unit is programed to only charge the 18650 battery when it registers that the voltage of the AGM battery is 13V or above meaning fully charged (or close to). That way the ESP32 only drain power from the main battery when it is full.
About licensing, anyone can replicate the work I made, I just want people to contact me when they choose to build 100 and sell them, mostly for collaboration.
And for crediting, yeah you are right about that one, the list is long. The master was state funded.
our team had also deployed similar low power gateway last year in Hong Kong, it’s has running very good. My consumption is around 12V 150mA to 240mA depending on Cat1 4G TX. we used a 40W PV, 40AH battery and a very expensive Mppt chip.
Also, we created a battery management circuit to monitor the PV charging current and the battery voltage to decide if the PV is covered with bird poo or need cleaning.
To further reduce power, we decided to filter unnecessary packet at gateway side so only meaning ful package will be sent via 4G, this will also drastically reducing power consumption.
The whole kit is very tiny , really a good idea for fast deployment.
In the context of TTN this is not a thing! All packets need to be treated equally in line with the free access concepts and the TTN Manifesto. You as a GW owner have no concept of what is a ‘meaningful’ packet. White listing as it is usually refered to might block legitimate packets from other TTN users and more importantly other networks altogether where there is a peering arrangement in place, most usually via PacketBroker, and where it is agreed exchange and routing of packets to the benefit of all. If your GW is connected to the TTS(CE) (TTN) please turn off filtering as soon as…
Otherwise well done on the implementation…inclement weather in UK means 60W practical min limit here with much larger back-up battery so most often go with min 100W often 120W panel paired with >90Ah battery…depends on if sun path is obscured at all…if goal is 24/7/365 operation. Even 60 can leave some offline periods especially in winter months. (Also tend to use 3G historically which typically has lower peak power envelope but as £G networks ramp down having to look at 4G/5G migrations…)
) - since atleast 2018 from my own personal knowledge and experience with many versions used/posted to the Forum etc. by a number of users. Other solutions have been implemented using alternate control plane/management processing using alternates to Pi0W’s as well…