Missing traffic

Hi thingpeople,

I’m new to LORAWAN and TTN.
I got a new RAK7258 gateway and went through the basic instructions to set it up. I live in Perth Western Australia a few KM north of the CBD on a hill and the GW sits in a window with city views. I don’t have any sensors (yet)

After the basic TTN config I started to get a LOT of traffic. (300 message in an hour or so!! Whoohoo it works) Most of them were joins… then I realised the channels were a little strange and it looks like they were a default config not the AU915. They looked like the image below…

image

I then set “import channels -> AU915” and it looks like this now.
image

image

Most of the packets then stopped.
Now; with the new channel config the RAK main screen shows I get around 1-2 packets of traffic per minute.
BUT… only 1-2 packets per HOUR actually turn up in the packet logger window. (and get sent to TTN)

Q1: On my initial setup I’m pretty sure I was listening on the wrong channels. does that mean there is / was someone transmitting a huge amount of data on an incorrect / illegal channel? Or was it just someone is transmitting on slightly offset channels from standard?

Q2: I assume my channel config is now correct for my region. Why are so many packets arriving but then not showing up in the packet logger? Are these non LORA packets its seeing? Or are these somehow invalid packets and getting discarded? (CRC on or off on the logger shows nothing different).
Most “missing” packets are turning up on DR5 with a SNR of -5 and RSSI of -100

I guess my main aim here is to confirm I have my gateway setup correctly for my region and its not eating packets.

Thanks for the help…

Brom

First, without running nodes of your own or awareness of local efforts, there’s no reliable basis to expect to see anything. And if you do see something, it might be some proprietary LoRa scheme from a smart meter, not LoRaWAN and especially not TTN LoRaWAN. There is also the potential issues of packets with low level or “hardware” CRC errors. Typically these are not logged by packet forwarder software, but merely counted. While it’s possible to pass them on to the network server, that’s rarely done as they’re not useful.

In terms of frequency settings, you can compare the enabled channels to those listed at https://www.thethingsnetwork.org/docs/lorawan/frequency-plans.html

Since there are two distinct bandplans applicable in your area, it’s possible that people locally are using one (for TTN or other purposes) and not the other.

It’s not entirely clear to me that the RAK packet logger is supposed to do anything when used in basic gateway mode pointed at TTN (vs using the internal stand alone network server), but my RAK gateways have essentially never run factory firmware, so I could be completely wrong on that.

Would would be informative would be if you could share details of the sort of of packets that are showing up in one place and not the other. And while screen shots are great, actual cut-and-pasteable text is better, especially for anything that has a based64 encoded payload. It would be especially interesting if there seems to be a repeating packet sort of thing showing up in one place and not the other.

In Au there are 64 channels allocated to this radio band and for LoraWan these can be considered as 8 blocks of 8 channels. Your initial configuration was the first block of 8 whereas in Au TTN sits in the second block of 8 channels.

You are correct your current configuration is correct for TTN.

In addition to TTN there are other LoraWan networks, it makes sense to locate these in another block of 8 channels and avoid traffic congestion with TTN. The traffic in the first 8 channels could therefore be another LoraWan network, a series of Lora point to point radios or some other radio technology quite different to Lora. All of these are quite legal as Lora and LoraWan are located in a part of the radio spectrum which has shared use, you don’t need a licence but also don’t have exclusive use.

From the description of your gateway’s location and elevation, it sounds like your are in the ideal location for maximum coverage and so its not unexpected that you are receiving so much data.

So the next step is to get a node. You may also want to reach out to the nearest TTN community, they may be able to test your gateway for you.

Thank you very much @cslorabox, @TonySmith

The channel details make total sense now. I also just noticed there is another dropdown for the multiple channel sub-bands!

I have a few sensors and a improved antenna on order which will arrive in a few weeks. Ill then be able to do more playing and actually get an understanding on the sort of range my gateway has. Ill also start getting involved with the local community.

Just for interest sake… here is my RAK overview pages on the TTN frequencies.
image
I didn’t manage to capture any packets via the packet capture window over the last 30 or so min.

So then changed over to the 0-7 channels… and saw lots of join request packets like this.

{
    "freq": 915200000,
    "chan": 0,
    "tmst": 22569260,
    "utmms": 1607235234438,
    "rfch": 0,
    "stat": 1,
    "rssi": -109,
    "size": 23,
    "modu": "LORA",
    "datr": "SF12BW125",
    "codr": "4/5",
    "lsnr": -6.5,
    "data": "AEFYSU9NLUxHWBUVdAQAAABLs2VB3Wg="
}{
    "MHDR": {
        "MType": "Join Request",
        "RFU": 0,
        "Major": 0
    },
    "JoinRequest": {
        "AppEUI": "47 4C 2D 4D 4F 49 58 41 ",
        "DevEUI": "00 00 00 04 74 15 15 58 ",
        "DevNonce": "B34B"
    },
    "MIC": "68DD4165"
}

{
    "freq": 915200000,
    "chan": 0,
    "tmst": 17255076,
    "utmms": 1607235229121,
    "rfch": 0,
    "stat": 1,
    "rssi": -119,
    "size": 23,
    "modu": "LORA",
    "datr": "SF12BW125",
    "codr": "4/5",
    "lsnr": -16.8,
    "data": "AEFYSU9NLUxHXRYVdAQAAADMPFzQzck="
}{
    "MHDR": {
        "MType": "Join Request",
        "RFU": 0,
        "Major": 0
    },
    "JoinRequest": {
        "AppEUI": "47 4C 2D 4D 4F 49 58 41 ",
        "DevEUI": "00 00 00 04 74 15 16 5D ",
        "DevNonce": "3CCC"
    },
    "MIC": "C9CDD05C"
}

On the 0-7 frequencies It seems around 3/4 of the packets that arrive in on the overview page DO show up in the packet logger; But only 10% of the packets in the 8-15 band are shown in the logger.

If I receive a non LORA packet would this show up in the receive list but then not in the packet logger? ie. I have a number of sensors around the house which use the 900ish Mhz band which are not LORA based. But… saying that I would expect they would show up with much stronger signal as some of them are within a few meters of the gateway. (They are Aircon temp sensors)

Thanks for the help anyhow.
My main concern is that I am loosing valid packets somehow.

Brom

If by improved you mean higher dBi spec be aware that depending on local regulatory environment you may have to dial back GW Tx power to saty withing local EIRP limits - its a zero sum game! A small amount of gain can be useful to offset local cablerun/connector losses but this should only be <1db to <3-5db for a good installation with Ant close to GW system anyhow. Very high dBi figures usually also associate with directionality (e.g increased down angle of main beam - good if location high masts/locations otherwise not so good… also typically they are associated with non linearnon-isotropic radition patterns so great for reaching toward local horizon but can give supprising null or reduced sensitivity zones closer in making them counter productive for uniform coverage in a area - there is a reason the emergency services often stick to simple dipoles :wink:

Not wanting to burden Tony but suggest you seek his advise/guidance on plans if he is willing as he is something of a local ‘expert’ given his years of handling these kinds of system :wink:

1 Like

Thanks @Jeff-UK

Currently I have the RAK sitting in the window and not only does it get full Australian sun its also hard to place in a good location just because of its size and shape.

I have a few antennas on order. The intention isnt to get something with a higher dBi rating. Just something which is easier to mount in better locations and is a bit more aesthetically pleasing. eg

Also… I cant find the exact dBi rating on the included antenna. I know the new window antenna above says its 3dBi but with the 3 metes of cable and the detailed product documentation which wont be included I have low expectations is going to be any better than what is included.

I will do some testing and look at reducing the transmit power if it looks to be better than the included antenna. (I assume you CAN reduce the transmit power on these devices)

Brom

@Jeff-UK gave you already the most important informations concerning antennas. The antenna you selected has 3 m RG174 cable, 1dB loss per meter. The radiator shall have a gain of 3dBi - at what frequency?
A 1/2 lambda dipol has a gain of 2.1dBi or 0dBd and a length of about 0.35m (depends on resonance frequency). I would recommend an antenna with a length of 1/2 lambda - gain is not everything!
Much more important is where you mount your antenna. Inside a building is always a compromise, an antenna should be as high and as “free” as possible if you want to have a good coverage. Additionally you have to consider the radiation-pattern especially when using antennas with high gain.
Don’t trust the gain-values given by the sellers if they are not verified by a competent RF-laboratory. There are no “miracle-antennas” with “8dB gain” and 0.15m length for our frequencies.

PS: Does your gateway have a RP-SMA connector?

Thanks @wolfp
Re connector. I think I have the correct one.
I got a another similar antennas coming too (A 4db dipole with 2m cable); its all going to be experimentation. Nice to know that a <3-5db is a good sweet spot. The length longish of cable on some of the antennas is bit annoying (It was hard to find something shorter with the correct connector from Aliexpress). I have a friend with antenna crimp equipment so might cut it down sometime in the future or might be able to open the antenna and cut the cable from that end.
The units I am getting are all around $10 each so I have high expectations they are going to be cheap with questionable ratings. Its a home / fun install so happy to tinker a little. I would be taking a very different approach if this was a corporate orientated system.
Ill see if I can track down a low cost 2dBi 1/2 lambda dipol too. (I’m all ears if anyone has any good recommendations.)

Re placement - I have been thinking about putting it on the top of the roof… but I have a 3 floor house and steep incline. So my choices are, attempt to kill myself and climb up on my own; pay for professional riggers; or experiment with low cost antenna in the window.
Ill start with the easy option first… If I can get a low cost antenna with similar performance which allows me to take the RAK off the windowsill then ill call it a success.

I also need to wait for my sensors to arrive too. Right now I have no idea how far away the transmitters are. All I can tell now is I am seeing around 30 unique devices over an hour with RSSI’s between 90-120.

Brom

1 Like

Just a quick update.

I received one of my cheap antennas today and tested it out.

Its supposed to be a 12dBi antenna… in reality… its not; At least not for the LORA AU frequencies and when you include the cable loss. (No surprise there)
Using it on my RAK7258 I get exactly the same performance as the default antenna (both uplink and downlink)
I have 2 static LORA sensors and I am getting identical RSSI and SNR. I am also seeing identical statistics from other peoples random chatter. Which for me is a 100% success.
My aim was to just be able to move the gateway out of the window. The gateway is now hidden behind furniture and the smallish antenna is now in the window.
I am graphing the RSSI and SNR on my devices and cant see where I switched antennas.

I have a couple of other (potentially more suitable) antennas coming in the next few weeks which I will also test.

I hope other people find this useful. Especially if you have a similar aim where you just want to be able to split the antenna away from the gateway while keeping the costs low.

ps I also got an antenna cable coupler to convert the connector. (RPSMA SMA JK)


Antenna cable is around 1m (not mentioned on the listing)

Brom

2 Likes

This kind of antennas have a magnetic feet. This feet is used to make a capacitive coupling to ground (metallic plate, roof of a car …). This connection to ground is part of the antenna system. If your windows are made of wood or plastic, the antenna works better if you put a piece of metall sheet underneath.

“Its supposed to be a 12dBi antenna”. This are Chinese dBi. They seem to use an other system to determine the gain of an antenna.
Imho this kind of antenna has abt. 5 dBi at 900MHz.

yeh, Its on an aluminium window at the moment.
Originally I was going to put a ferrite plate of some sort under it… but seems to be working pretty well so have not tried it yet. Any idea how aluminium performs vs something ferrite?

One strange thing is although all of my graphs are showing the same antenna performance I am now getting a lot more “active nodes” showing up. Previously I was seeing 5 or so active nodes. Now I am seeing 40.
Any idea how the RAK calculates the active nodes? I don’t seem to be getting any more packets than previously. I might be getting slightly higher SNR numbers on average but it would probably only 1 or so dB.

Re antenna 5dBi sounds logical too. loose 1dBi for the cable. I understand the stock is 4dBi.

Brom

Aluminium is even better than steel (for RF). Imho it is a good solution to put the antenna on the frame of the window. Ferrite would be worse because it absorbs the RF. We use ferrite plates e.g. in Fully Anechoic Chambers (FAC) to simulate free space conditions.
The reason why you might see more nodes can be the radiation-pattern of your antenna-system consisting of the antenna and the alu-frame of your window. But I don’t know what RAK means by “active node”.
BTW: The attenuation of a glass-windows at 900 MHz is abt. 2dB, if it is made of coated glass the attenuation can be several 10dB.

1 Like