Unable to ping using ML-EID

[Rohith R]

I have created a Thread Network on  SubGHz(868MHz) using EFR32MG13.
My network have only 2 nodes- leader and child or router. Thread network
is hardcoded into 3 devices.(sleepy demo ftd on leader and router and mtd on child)
Leader and child kept at a particular distance and they are able to ping using ML-EID
and Link Local Address. When ping from Leader response is sent form Child.

> ping fdde:ad00:beef:0:b28c:6d22:6777:a98d
ping fdde:ad00:beef:0:b28c:6d22:6777:a98d

16 bytes from fdde:ad00:beef:0:b28c:6d22:6777:a98d: icmp_seq=21 hlim=64 time=11ms
1 packets transmitted, 1 packets received. Packet loss = 0.0%. Round-trip min/avg/max = 11/11.0/11 ms.
Done
> ping fe80:0:0:0:b01c:9f93:6f4a:4db7
ping fe80:0:0:0:b01c:9f93:6f4a:4db7

16 bytes from fe80:0:0:0:b01c:9f93:6f4a:4db7: icmp_seq=22 hlim=64 time=10ms
1 packets transmitted, 1 packets received. Packet loss = 0.0%. Round-trip min/avg/max = 10/10.0/10 ms.
Done

When the Child is replaced by Router it is unable ping using ML-EID but ping
through Link Local address

> ping fdde:ad00:beef:0:e63a:82f5:d378:e89a
ping fdde:ad00:beef:0:e63a:82f5:d378:e89a

1 packets transmitted, 0 packets received. Packet loss = 100.0%.
Done
> ping fe80:0:0:0:88db:3b4b:cc38:7785
ping fe80:0:0:0:88db:3b4b:cc38:7785

16 bytes from fe80:0:0:0:88db:3b4b:cc38:7785: icmp_seq=16 hlim=64 time=8ms
1 packets transmitted, 1 packets received. Packet loss = 0.0%. Round-trip min/avg/max = 8/8.0/8 ms.
Done

What will be the reason for this ping issue

[Jonathan Hui]

One reason could be that the Active Operational Datasets are not consistent between the different devices. Can you print the Active Operational Dataset on all the devices and make sure they are all exactly the same?

--

Jonathan Hui

--
You received this message because you are subscribed to the Google Groups "openthread-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openthread-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/ae0f4b67-3b1a-48a1-a30e-649e6e5be712n%40googlegroups.com.

[Rohith R]

Thank you  for your reply. Both the devices are hard coded with the same Thread Network. Both are loaded with sleepy_demo_ftd   example code. 

Do they have the same active operation dataset ?

For Leader 

> dataset active
dataset active
Active Timestamp: 1
Channel: 0
Ext PAN ID: c0de1ab5c0de1ab5
Network Key: 1234c0de1ab51234c0de1ab51234c0de
Network Name: SubGHzThread
PAN ID: 0x2222
Done
> 

For Router

> dataset active
dataset active
Active Timestamp: 1
Channel: 0
Ext PAN ID: c0de1ab5c0de1ab5
Network Key: 1234c0de1ab51234c0de1ab51234c0de
Network Name: SubGHzThread
PAN ID: 0x2222
Done
>
>

--

Rohith R 

[Jonathan Hui]

It appears your Operational Datasets are missing a number of key parameters, including the Mesh-Local Prefix, which is why your devices are not able to communicate using the ML-EID.

I would suggest initializing the Operational Dataset using the `dataset init new` command, as is done in the Quick Start of src/cli/README_DATASET.md

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/db78881f-b73a-44a0-81f4-eff791398903n%40googlegroups.com.

[Rohith R]

Dear Jonathan

I am attaching the  image of my Thread Network consist of a Leader, Router and a child.

 The problem I am facing is that in the Thread network Child can ping the Leader through ML-EID. 

But the router can not ping the Leader through ML-EID

(Also Router cannot ping Child )when they kept a particular distance.

But when these 3 nodes are kept nearer to each other , they are able to ping  through ML-EID.

(child- Router, Router - Leader and Leader - Child).The Thread network is hardcoded into these 3 nodes 

(sleepy_demo_ftd and sleepy_demo_mtd). Mesh local prefix is also hard coded.

I am attaching the Network creation function.

The operational dataset of the 3 devices are same as follows

> dataset active
dataset active
Active Timestamp: 1

Channel: 2
Ext PAN ID: c0de1ab5c0de1ab5
Mesh Local Prefix: fdaf:ce27:4bdc:5026::/64

Network Key: 1234c0de1ab51234c0de1ab51234c0de
Network Name: SubGHzThread
PAN ID: 0x2222
Done
--

Rohith R

[Jonathan Hui]

When the router is "kept at a particular distance", does the router actually retain connectivity with the rest of the Thread network?

You can use the output of the `state` and `router table` CLI commands to gain visibility into the Thread topology.

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/018f351c-0a87-479c-b04a-db940d82bc09n%40googlegroups.com.

[Rohith R]

The router ping the leader via  Link Local address but not using Mesh-Local EID when it   kept at a particular distance. 

Still the router is in the Thread Network. I am attaching the console details. Please find the attachment.

--

Rohith R

[Jonathan Hui]

I noticed that the LQ In and LQ Out values are both zero for the neighboring device, which indicates that the link quality is very poor. When the link quality is poor, the Thread router will not establish a link with the neighboring router, which is necessary for non-link-local communication.

Can you provide the output of `neighbor table` command? I'm wondering if the RSSI values are very low (which could be due to a radio driver issue or the fact that the link quality is actually low).

--

Jonathan Hui

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/079625ec-6f0f-417a-a437-b4a871bb09b5n%40googlegroups.com.

[Rohith R]

The neighbor table result of the router is showing below

> neighbor table
neighbor table
| Role | RLOC16 | Age | Avg RSSI | Last RSSI |R|D|N| Extended MAC     |
+------+--------+-----+----------+-----------+-+-+-+------------------+
|   R  | 0xf000 |   6 |     -103 |      -101 |1|1|1| 5a61bba83888af1e |

Done

"Thread router will not establish a link with the neighboring router" -

But the Router ping the Leader through Link Local Address

The position of nodes are same as that in the image I shared.Router and  child are approximately the same distance from the 

Leader. The Child can ping the Leader via ML-EID but the Router  cannot via ML-EID but with Link Local Address.

--

Rohith R

[Jonathan Hui]

For routers to forward messages, they need to establish links and routes. The RSSI values are lower than the receive sensitivity reported by the radio driver (see src/src/radio.c). You could try adjusting the radio receive sensitivity to something like -120, since OpenThread currently defaults to requiring a link margin of 10 dB.

Link-local messages do not require routes because they are only to neighboring devices.

--

Jonathan Hui

[Rohith R]

Thank you for your quick response. I changed the Txpower to 10dbm.(Initially it was 0dbm) Now the Router can ping the 

Leader through ML-EID.  

Now is it necessary to change the Receiver sensitivity to -120 ?

--

Rohith R

[Jonathan Hui]

The receive sensitivity should be set to a value that accurately reflects the platform.

+Sagar Chinchani , do you have a suggestion for an appropriate receive sensitivity value?

--

Jonathan Hui

--
You received this message because you are subscribed to the Google Groups "openthread-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openthread-use...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/9329ce49-f8f1-4074-b538-1cc600201331n%40googlegroups.com.

[Rohith R]

Dear Jonathan

Thank you for your reply.  From our discussion I understand RSSI is important for routers to route information (communication based on

ML-EID)and is not as much important for link - local messages(communication based on LLA). Am I correct ?

I have some doubts based on that

What will be the  required  range of RSSI for routing ? is it -70dbm to -80dbm ?

RSSI below than -80dbm is a poor signal?

 Should  RSSI  be greater than Receiver sensitivity for good routing ?

"The receive sensitivity should be set to a value that accurately reflects the platform" - Sorry sir I

did not get what you mean.

" radio receives sensitivity to something like -120, since OpenThread currently defaults to requiring a link margin of 10 dB." -

What is the logic behind this ?

 Can we do anything to  improve the Link quality (bcz when i check the router table link quality is 1,poor) ? 

Please clarify my doubts

--

Rohith R

[Jonathan Hui]

On Fri, Mar 4, 2022 at 3:26 AM Rohith R <rohith...@gmail.com> wrote:

Dear Jonathan

Thank you for your reply.  From our discussion I understand RSSI is important for routers to route information (communication based on

ML-EID)and is not as much important for link - local messages(communication based on LLA). Am I correct ?

RSSI is important to establish stable links that can be used to route packets.

I have some doubts based on that

What will be the  required  range of RSSI for routing ? is it -70dbm to -80dbm ?

RSSI below than -80dbm is a poor signal?

 Should  RSSI  be greater than Receiver sensitivity for good routing ?

What matters is link margin (difference between RSSI and radio receive sensitivity). OpenThread requires a link margin of 10 dB to establish a link.

"The receive sensitivity should be set to a value that accurately reflects the platform" - Sorry sir I

did not get what you mean.

A given hardware platform can be characterized by its receive sensitivity.

" radio receives sensitivity to something like -120, since OpenThread currently defaults to requiring a link margin of 10 dB." -

What is the logic behind this ?

Radio link quality varies over time. The link margin threshold avoids unstable links that can lead to an unstable mesh topology.

 Can we do anything to  improve the Link quality (bcz when i check the router table link quality is 1,poor) ? 

Link quality is based on properties of the hardware platform and the surrounding environment. There's not much you can do in software to affect link quality.

--

Jonathan Hui

[Sagar Chinchani]

Hello Rohith,

Jonathan has correctly pointed you to all the important pieces. (Thank you, Jonathan!)

At an onset, -100 dB for Rx sensitivity seems to be in the right range (although I am trying to get a confirmation at my end). That being said, can you confirm if you are using a whip antenna (external antenna via the SMA connector) for the 868 MHz band, for all the nodes in your setup? Not using an external antenna when operating on the Sub-GHz band will impact RSSI values in turn impacting the link margin, resulting in poor link quality.

Thanks

Sagar

[Rohith R]

Dear   Sagar Chinchani

Thank you for your reply. I am created this Thread Network by taking the example code sleepy_demo_ftd and sleepy_demo_mtd 

from the simplicity studio and edited some portions (phy layer is changed by the code that you shared, changed  the channel settings in 

openthread-core-efr32-cofig.h and  set the radio PTI configuration  in radio.c file) to make it as working on Sub-GHz. I expect it is working in the 868MHz Sub - GHz region.

(Not Tested using spectrum Analyzer).  I connected the external whip antenna using SMA connector  in  all the 3 nodes. 

Any software changes is needed for the whip antenna ?

Link margin = RSSI - Receive sensitivity . For me RSSI is some time in the range of -95 to 110.

Is it necessary to change   the  Receive sensitivity to  -120dBm. so that the Link margin is  > 10bBm

[Sagar Chinchani]

Hi Rohith,

I had a further discussion with the team in this regard. Sensitivity depends on the exact PHY and you could get better or worse the -100 depending on the PHY. As such, even the -100 dBm value may not be applicable for all our parts at 2.4 GHz.

Typically, for official PHYs, the sensitivity value is specified in the datasheet. For a custom PHY (like this one) one would have to measure it. Since this PHY was provided for experimentation purposes, I do not have an exact value for you to use. That being said, a range between -100 to -104 dBm seems reasonable.

Thanks

Sagar

[Rohith R]

Dear Sagar Chinchani

Thank you for your reply. In the datasheet of EFR32MG13 specify the Receiver sensitivity upto -120dBm for 868MHz. 

When I changed the receiver sensitivity to -120dBm to keep the Link margin > 2 , for an RSSI of -104 , link is maintained

 and the nodes ping each other at a particular distance. 

Is there any problem keeping the Receiver sensitivity upto max?

Thanks

Rohith R

--
You received this message because you are subscribed to the Google Groups "openthread-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openthread-use...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openthread-users/b765504e-ba73-4863-b484-4c7c74ca4b69n%40googlegroups.com.

[Sagar Chinchani]

Hi Rohith,

I had a further discussion with the team at my end.  The theoretical sensitivity for the custom PHY that was provided for the EFR32MG13 part on 868 MHz, is around -96dBm. Accordingly, the recommendation from our end will be to use this value and not modify it to -120 to satisfy the link margin values. That being said, there could be other factors in your setup that may impact these parameters, for example, a poor antenna match resulting in a degraded RSSI. Verifying these may be a good debugging step at your end.

Thanks

Sagar.

[Rohith R]

Dear Sagar Chinchani

 I will look the factors that degrade the RSSI.
Please find the attachment, portion of EFR32MG13 Datasheet..

Please give valuable in suggestions or your opinion in the question.