Packet loss issue with increased range


(Shankardevy) #1

we have tried a Lorawan Network with Catenna 44550 node and Multitech Gateway.
we can able to achieve data transmission for 15 minutes, when the gateway is kept in 500mts to the 1000mts range.
specifications are below
1.Band = IN865.
2.Frequencies Used by Node and Gateway = 865.0625, 865.4025, 865.985.
3.Coding Rate = 4/5.
4.Modulation = LORA.
5.Data Rate = SF12BW125.
6.payload = 27(include lorawan header)

Using the above specifications. we tried transmissions for 3Km, but the output we got is uneven data transmission with non-regular interval.
we have also checked TTN-fair policy, using spreading factor, bandwidth, payload length.
we have calculated duty cycle and time on air using the above data.

Suspecting that we have exceeded the fair usage limit, we changed the frequency to every one hour.
https://docs.google.com/spreadsheets/d/1uzm7PBfrGCzVAF2DkoGn1eBQKrM-UguQLVT-VCk6CwE/edit?usp=sharing
Using the above calculator we found that we could send 15 packets/day at data_rate of SF12BW125. we are sending data on three channels
865.0625, 865.4025, 865.985. so we have a total budget of 3*15 packets per day and at a rate of sending data every one hour we will be much below the fair usage limit.
However even with these reduced frequency we are loosing several packets every few hours as you can see below

Date        time      Battery Volt  Water Level
|13/05/2019, 15:53:24|4.2412109375|26.7617352089704|
|13/05/2019, 14:53:18|4.24755859375|26.7649417329256|
|13/05/2019, 12:53:04|4.2412109375|26.7697515188583|
|13/05/2019, 11:52:58|4.228271484375|26.7694308664628|
|13/05/2019, 10:52:51|4.221923828125|26.7796917431193|
|13/05/2019, 09:52:44|4.208984375|26.763979775739|
|13/05/2019, 05:52:18|4.202392578125|26.7614145565749|
|13/05/2019, 04:52:12|4.208984375|26.7598112945973|
|13/05/2019, 02:52:00|4.208984375|26.7614145565749|
|13/05/2019, 01:51:53|4.21533203125|26.7614145565749|
|13/05/2019, 00:51:46|4.221923828125|26.7630178185525|

Questions;
How do we narrow down to the real cause of lost packets? is it due to exceeding limits or due to range issue or is it because the node is sending on wrong frequency for those lost packets.
right now without clear way to debug, we are unable to solve this riddle.


#2

what did you try ?
and please what sensor do you use exactly (link?)


(Shankardevy) #3

Water level sensor https://dalianruike.en.ecplaza.net/products/high-stability-submersible-water-level-sensor_4062084.

we have tried transmitting data from Node but the interval is uneven. As mentioned in the above post.


#4

reading the docs I understand that you refer only to a pressure sensor, So I think you must have a (LoRawan) node that it is connected to.
What is that node (DIY or commercial) and did you wrote the code for it ?


(Shankardevy) #5

we bought catena 4450 from MCCI


(Shankardevy) #6

catenna 4450 code https://github.com/mcci-catena/Catena-Sketches/blob/master/catena4450m102_waterlevel/catena4450m102_waterlevel.ino


#7

I see in that table that you first receive data every second ? and then after 9 frames suddenly once in an hour ?
there must be something wrong in your code ?

  • why only 3 channels
  • why that high accuracy… or is that voltage the sensor output raw ?

(LoRaTracker) #8

I thought the fair access policy related to transmission time in total from the node, regardless of the number of frequency bands in use.


(Shankardevy) #9

sorry wrong data table had been added, now i have changed the new data.
|13/05/2019, 15:53:24|4.2412109375|26.7617352089704|
|13/05/2019, 14:53:18|4.24755859375|26.7649417329256|
|13/05/2019, 12:53:04|4.2412109375|26.7697515188583|
|13/05/2019, 11:52:58|4.228271484375|26.7694308664628|
|13/05/2019, 10:52:51|4.221923828125|26.7796917431193|
|13/05/2019, 09:52:44|4.208984375|26.763979775739|
|13/05/2019, 05:52:18|4.202392578125|26.7614145565749|
|13/05/2019, 04:52:12|4.208984375|26.7598112945973|
|13/05/2019, 02:52:00|4.208984375|26.7614145565749|
|13/05/2019, 01:51:53|4.21533203125|26.7614145565749|
|13/05/2019, 00:51:46|4.221923828125|26.7630178185525|


#10

and what do you see in your console… RSSI / SNR ?

you didn’t explain this accuracy 4.221923828125|26.7630178185525, why is that neccessary ?


(Shankardevy) #11

|Date and Time|rssi|snr|
|13/05/2019, 19:53:48|-121|-13.5||
|13/05/2019, 19:53:48|||865.0625|
|13/05/2019, 18:53:42|-120|-17.5||
|13/05/2019, 18:53:42|||865.985|
|13/05/2019, 16:53:31|-120|-17.25||
|13/05/2019, 16:53:31|||865.0625|
|13/05/2019, 15:53:24|-121|-15.5||
|13/05/2019, 15:53:24|||865.985|
|13/05/2019, 14:53:18|-119|-16||
|13/05/2019, 14:53:18|||865.4025|
|13/05/2019, 12:53:04|-123|-14.5||
|13/05/2019, 12:53:04|||865.985|
|13/05/2019, 11:52:58|-120|-17||
|13/05/2019, 11:52:58|||865.4025|
|13/05/2019, 10:52:51|-120|-15||
|13/05/2019, 10:52:51|||865.0625|
|13/05/2019, 09:52:44|-123|-16.75||
|13/05/2019, 09:52:44|||865.985|
|13/05/2019, 05:52:18|-120|-10.75||
|13/05/2019, 05:52:18|||865.4025|
|13/05/2019, 04:52:12|-121|-12.75||
|13/05/2019, 04:52:12|||865.0625|
|13/05/2019, 02:52:00|-119|-16.25||
|13/05/2019, 02:52:00|||865.4025|
|13/05/2019, 01:51:53|-120|-14.25||
|13/05/2019, 01:51:53|||865.0625|
|13/05/2019, 00:51:46|-120|-17||
|13/05/2019, 00:51:46|||865.985|
|12/05/2019, 23:51:39|-119|-12.5||
|12/05/2019, 23:51:39|||865.4025|
|12/05/2019, 22:51:33|-121|-15.25||
|12/05/2019, 22:51:33|||865.0625|
|12/05/2019, 21:51:27|-121|-15||
|12/05/2019, 21:51:27|||865.985|
|12/05/2019, 19:51:15|-123|-17.75||
|12/05/2019, 19:51:15|||865.0625|
|12/05/2019, 17:51:03|-118|-18.75||
|12/05/2019, 17:51:03|||865.4025|
|12/05/2019, 14:50:43|-120|-14.5||
|12/05/2019, 14:50:43|||865.4025|
|12/05/2019, 11:50:23|-121|-13.75||
|12/05/2019, 11:50:23|||865.4025|
|12/05/2019, 10:50:16|-123|-19.75||

battery voltage 4.221923828125
water level 26.7630178185525
there is no intention to have high accuracy. just a value


#12

I think the signal is a bit weak and maybe therefore some frames could get lost.
Try to increase the antenna height on the sensor side, a different antenna on the gateway with some more gain or a directional antenna on the node.


(Shankardevy) #13

can you suggest some directional antenna which will suit for catenna 4450. Thanks for your valuable support


#14

realize that if you don’t use/need that accuracy (for a battery 4.2 is accurate enough for example) you waste a lot of bandwith and energy.
So always try to achieve the smallest payload possible.

this one worked for me


(Ud Lo Ra) #15

@BoRRoZ is right, and since you only need 3kms, perhaps the antennas you have are sufficient, with just some repositioning. Are you in a urban or rural area? Where is the gateway (indoor, outdoor, height from ground…) and where the node too.


(Cslorabox) #16

Your RSSI and SNR figures are very poor, so data loss is not surprising - your packets which did get through, only barely did so.

If the gateway in use is your own, you may be able to monitor raw packets yourself (particularly look for the fixed or OTAA-assigned device address in their unencrypted header) and thus rule out any network issues, but those are less likely to be the problem anyway.

Really what you need to do is figure out why the signal quality of the packets that do get through is so weak. Is there a clear line of sight between the node and the gateway? Do you have antennas properly connected to each? If I recall, on the Catena boards (or in your case the Feather M0 LoRa hosting a Catena sensor board) you must calculate the length of an antenna for your frequency, cut a piece of wire, and solder it to the appropriate pad on board.

What is the power source for your node? Some things like USB charger batteries can be very tempting for their convenience, but generate large amounts of RF noise which interferes with radio links.

Are you sure all of your components are actually 868 MHz models, and not 915 MHz versions being operated on 868 MHz where their RF networks will be mistuned?

Finally, even if a long packet isn’t getting your data dropped by usage limiters, the longer your packet the more likely it is to be corrupted in transmission causing either a radio checksum or a MIC failure leading to it being discarded. If possible, try to transmit efficiently packed binary data fields, and render to printable text on the receiving side. For example, if your battery reading comes from a 12-bit ADC you could probably just transmit the actual 12 bits, either across two bytes or even better by using only 12 bits of those two bytes, and packing something else like status flags in the remaining 4.


(Shankardevy) #17

Gateway has been placed 40 meters above the ground level and it is an outdoor gateway.
we are in a rural area. node is placed in the ground level in between trees. both gateway and node are placed in a rural area. Thanks


(Ud Lo Ra) #18

Even if in between trees, it seems strange that you have such a poor SNR (this is what I obtain at 3km with an indoor gateway, 10m above the ground…). I would try to test the connection at the same distance but in a clear area (or putting the node at some height on the trees), but also check what @cslorabox is suggesting regarding antennas and connections (including shortening the payload…).


(Crumble1970) #19

I would suggest also taking a look at the orientation of the water meter. If it’s using an internal antenna or the antenna is not square to the gateway you may simply be in the wrong RF plane for a good signal.


(Shankardevy) #20

thanks @BoRRoZ @cslorabox @UdLoRa @Crumble1970 with your suggestions so far.

We have tried

  1. reducing the payload by sending only the values we need.
  2. confirmed that catena board is for IN band and not US model.
  3. wire antenna is intact with catena. (havent yet tried a directional antenna as suggested).

What I would like to try now is to increase the transmission power. does any one know how I can play with the transmission power value, specifically where in the code I can change it?

We are using: https://github.com/mcci-catena/Catena-Sketches/blob/master/catena4450m102_waterlevel/catena4450m102_waterlevel.ino

Any help is appreciated. thanks.