search

Koenkk / zigbee2mqtt

Zigbee 🐝 to MQTT bridge 🌉, get rid of your proprietary Zigbee bridges 🔨
https://www.zigbee2mqtt.io
GNU General Public License v3.0
12.05k stars 1.67k forks source link

Inaccurate Link Quality Value in Networkmap #2890

Closed myiotcenter closed 4 years ago

myiotcenter commented 4 years ago

Bug Report

What happened

I'm setting up a new Zigbee network at home, my current setup is using CC2538 + CC2592 Coordinator with two DL-22 CC2530 + CC2591 Routers. The Coordinater is connected to a Wemos D1 Mini through UART and I connected the zigbee2mqtt hassio add-on through tcp connection.

I connected few Xiaomi sensors and everything seems to be ok but whenever I generated a network map I can see the linkquality of few sensor is far below the value I can see in the mqtt messages or in HomeAssistant sensor value.

I connected 1 sensor directly to Coordinator and the value is more accurate compared to sensors connected to router.

This is my networkmap : https://imgur.com/a/pwENLPs I highlighted the linkquality values which are far below the values shown in HomeAssistant and mqtt messages. There are some value differences compared to the log due to time difference in generating map and logs.

These are the values showed in HomeAssistant Sensors: https://imgur.com/IxXlUVf https://imgur.com/YIhdmUu https://imgur.com/gMmZ6R2

I set "zigbee_herdsman_debug": true in add-on configuration and then generate raw networkmap and can see that LQI succeeded for Coordinator and two routers.

Here's the log when requesting raw networkmap and zigbee_herdsman_debug set to true : https://pastebin.com/HQUhDR5F

Here's the generated raw networkmap only: https://jsoneditoronline.org/?id=4bd436cb721e4897b1fdba2091128f01

What did you expect to happen

Generated Network Map is showing linkquality value closed to what I can see from mqtt message or sensor value in HomeAssistant.

Debug Info

zigbee2mqtt version: hassio add-on 1.9.0 CC253X firmware version: I flashed the coordinator with the firmware for CC2538 + CC2592 by @reverieline (MODKAMRU_V3_UART-no-flow-control.hex)

and the router is using firmware by @ptvoinfo https://github.com/Koenkk/zigbee2mqtt/issues/1362#issuecomment-499373874

Questions

  1. Is this an issue on zigbee2mqtt side or issue in coordinator / router firmware?
  2. Any other things I can set / try to identify the root cause?

I've been doing fresh install of the add-on, reflash the firmwares of coordinator & routers, etc. but couldn't make it right.

Appreciate your input and advice on this issue. Thanks.

Koenkk commented 4 years ago

To figure out where the values are coming from, sniff the traffic while executing the network scan: https://www.zigbee2mqtt.io/how_tos/how_to_sniff_zigbee_traffic.html

nanobra1n commented 4 years ago

The linkquality in the mqtt message (the ones homeassistant uses) is only the linkquality for the last hop (between router and coordinator). Which explains that once you connect your sensor directly to the coordinator, you get values that agree with your map.

myiotcenter commented 4 years ago

To figure out where the values are coming from, sniff the traffic while executing the network scan: https://www.zigbee2mqtt.io/how_tos/how_to_sniff_zigbee_traffic.html

Let me setup the sniffer and see what I can get...

The linkquality in the mqtt message (the ones homeassistant uses) is only the linkquality for the last hop (between router and coordinator). Which explains that once you connect your sensor directly to the coordinator, you get values that agree with your map.

I was previously testing my zigbee network setup using CC1352P2 coordinator + 2 CC2530 with antenna routers. I can see low linkquality between coordinator <-> router probably due to location / placement of the routers to coordinator or tx/rx power of CC25350 but end devices which I placed near the router are showing high linkquality values, close to what being showed in HomeAssistant sensors & mqtt messages.

myiotcenter commented 4 years ago

Hi @Koenkk, This is the sniffed traffic while executing raw network scan... https://pastebin.com/KLxiJC82

Koenkk commented 4 years ago

@myiotcenter the textual form is a bit hard to read, but if you search for: ZigBee Device Profile, Link Quality Response, Status: Success you see that the showed values of zigbee2mqtt are correct (LQI = 1):

        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Addr: 0x0000
            .... ..00 = Type: Coordinator (0)
            .... 10.. = Idle Rx: Unknown (2)
            .000 .... = Relationship: Parent (0)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 0
            LQI: 1

@ptvoinfo do you know what can cause this?

myiotcenter commented 4 years ago

@Koenkk Thanks for pointing out the important parts of the sniffed traffic... I searched specific for that parts and found 4 occurrences of ZigBee Device Profile, Link Quality Response, Status: Success.

Tried to make a summary in form of table for those 4 occurrences

sniffedPacketSummary

Looking at the pattern I would expect that there are records saying that a router is a parent of end device but from the log they are showing that Coordinator is the parent for all end device? I would also expect that relationship between router and coordinator is child but somehow both router showing the Relationship as None. Are these values all as expected?

My house is 2 storey building and quite wide around +/- 17m wide so I need to make sure that I'm placing Coordinator & Routers in the effective positions and expecting network map can help me to achieve that. I assume having CC2538 + CC2592 (or my CC1352P2) as coordinator and 2 CC2530 + CC2591 would be powerful enough to cover the whole house, but I need some reliable measurement.

This is a non production setup, more to a proof of concept exercise before I deploy sensors, switches, led driver, etc. So far sensors work as expected, Button & Window Sensor are triggering mqtt messages as expected and status in Home Assistant also changing in a timely manner.

  1. Any idea what will be the best approach to measure the coverage of my zigbee network other than expecting accurate LQI in network map?

  2. Is there any other thing I can check to narrow down the root cause? not sure if this is an issue in zigbee2mqtt, coordinator or router firmwares?

  3. I realize that temp sensors are not providing reports regularly, some sensors are reporting in more frequent than others. Is this normal? Will they only report if detecting changes in Temperature or Humidity?

Appreciate your feedback @Koenkk @ptvoinfo @reverieline

I removed other lines and assume these lines are the most interesting parts and move the values to the table above..

ZigBee Device Profile, Link Quality Response, Status: Success
    Sequence Number: 22
    Status: Success (0)
    Table Size: 3
    Index: 0
    Table Count: 3
    Neighbor Table
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Addr: 0x0000
            .... ..00 = Type: Coordinator (0)
            .... 10.. = Idle Rx: Unknown (2)
            .000 .... = Relationship: Parent (0)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 0
            LQI: 1
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:03:3a:e0:fa (00:15:8d:00:03:3a:e0:fa)
            Addr: 0x4836
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 26
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: TexasIns_00:1c:03:a6:f9 (00:12:4b:00:1c:03:a6:f9)
            Addr: 0x599f
            .... ..01 = Type: Router (1)
            .... 10.. = Idle Rx: Unknown (2)
            .011 .... = Relationship: None (3)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 255
            LQI: 1

ZigBee Device Profile, Link Quality Response, Status: Success
    Sequence Number: 23
    Status: Success (0)
    Table Size: 7
    Index: 0
    Table Count: 3
    Neighbor Table
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Addr: 0x0000
            .... ..00 = Type: Coordinator (0)
            .... 10.. = Idle Rx: Unknown (2)
            .000 .... = Relationship: Parent (0)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 0
            LQI: 27
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:03:21:f5:b9 (00:15:8d:00:03:21:f5:b9)
            Addr: 0xd715
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 1
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:01:87:0b:61 (00:15:8d:00:01:87:0b:61)
            Addr: 0x22e6
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 3

ZigBee Device Profile, Link Quality Response, Status: Success
    Sequence Number: 24
    Status: Success (0)
    Table Size: 7
    Index: 3
    Table Count: 3
    Neighbor Table
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:03:6c:24:9a (00:15:8d:00:03:6c:24:9a)
            Addr: 0x15b4
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 1
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:02:c9:07:ad (00:15:8d:00:02:c9:07:ad)
            Addr: 0x6367
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 1
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: Jennic_00:03:3a:f8:a6 (00:15:8d:00:03:3a:f8:a6)
            Addr: 0xa859
            .... ..10 = Type: End Device (2)
            .... 00.. = Idle Rx: False (0)
            .001 .... = Relationship: Child (1)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 2
            LQI: 1

ZigBee Device Profile, Link Quality Response, Status: Success
    Sequence Number: 25
    Status: Success (0)
    Table Size: 7
    Index: 6
    Table Count: 1
    Neighbor Table
        Table Entry
            Extended Pan: TexasIns_00:05:a9:90:1d (00:12:4b:00:05:a9:90:1d)
            Extended Address: TexasIns_00:1c:03:48:32 (00:12:4b:00:1c:03:48:32)
            Addr: 0x9d9c
            .... ..01 = Type: Router (1)
            .... 10.. = Idle Rx: Unknown (2)
            .011 .... = Relationship: None (3)
            .... ..10 = Permit Joining: Unknown (2)
            Depth: 255
            LQI: 1
Koenkk commented 4 years ago

Any idea what will be the best approach to measure the coverage of my zigbee network other than expecting accurate LQI in network map?

Unfortunately not, however I wouldn't worry to much about the LQI of the end devices, just make sure all of them have a router within +- 5-8 meters. For router check below

Is there any other thing I can check to narrow down the root cause? not sure if this is an issue in zigbee2mqtt, coordinator or router firmwares?

This is the strangest thing about your network map, the routers show a linkquality of 1 to each other. Can you try power cycling the routers and see if this solves the issue?

I realize that temp sensors are not providing reports regularly, some sensors are reporting in more frequent than others. Is this normal? Will they only report if detecting changes in Temperature or Humidity?

In order to save battery, they will only report when there is a significant change

The following "rules of thumb" can be applied in order to achieve a reliable and scalable network:

I think you already meet all of the conditions above, and as you don't have problems I wouldn't worry too much.

myiotcenter commented 4 years ago

Hi @Koenkk , Thanks for your explanations. I will try to power cycle my routers and see if things are getting better. I have few more DL22 stocks I ordered from Taobao so will see if I can make a good mesh with them,

Will provide further update. Thanks.

myiotcenter commented 4 years ago

Hi, I tried to power cycle coordinator and routers but things are remain the same. Since everything works as expected I guess I need to move on and proceed with the proof of concept and eventually roll out my prod setup. Thanks for your help @Koenkk !