How is the IEEE 802.15.4 Extended Address constructed for a Matter over Thread device?

[olavt]

I'm trying to understand how to use the ExtAddress (IEEE 802.15.4 Extended Address) to map back to a Matter Node Id.

How is the ExtAddress constructed for a Matter over Thread Node?

This is an example of information I would like to map back to a Matter Node Id:

$ ./chip-tool threadnetworkdiagnostics read neighbor-table 4 0
[1711873407.689027][10077:10079] CHIP:TOO: NeighborTable: 2 entries
[1711873407.689085][10077:10079] CHIP:TOO: [1]: {
[1711873407.689095][10077:10079] CHIP:TOO: ExtAddress: 11678088788494768592
[1711873407.689105][10077:10079] CHIP:TOO: Age: 8
[1711873407.689112][10077:10079] CHIP:TOO: Rloc16: 3072
[1711873407.689121][10077:10079] CHIP:TOO: LinkFrameCounter: 133988
[1711873407.689128][10077:10079] CHIP:TOO: MleFrameCounter: 17422
[1711873407.689135][10077:10079] CHIP:TOO: Lqi: 2
[1711873407.689143][10077:10079] CHIP:TOO: AverageRssi: -84
[1711873407.689150][10077:10079] CHIP:TOO: LastRssi: -85
[1711873407.689158][10077:10079] CHIP:TOO: FrameErrorRate: 3
[1711873407.689165][10077:10079] CHIP:TOO: MessageErrorRate: 0
[1711873407.689173][10077:10079] CHIP:TOO: RxOnWhenIdle: TRUE
[1711873407.689180][10077:10079] CHIP:TOO: FullThreadDevice: TRUE
[1711873407.689187][10077:10079] CHIP:TOO: FullNetworkData: TRUE
[1711873407.689194][10077:10079] CHIP:TOO: IsChild: FALSE
[1711873407.689201][10077:10079] CHIP:TOO: }
[1711873407.689225][10077:10079] CHIP:TOO: [2]: {
[1711873407.689233][10077:10079] CHIP:TOO: ExtAddress: 11744755628457332730
[1711873407.689240][10077:10079] CHIP:TOO: Age: 8
[1711873407.689246][10077:10079] CHIP:TOO: Rloc16: 47104
[1711873407.689253][10077:10079] CHIP:TOO: LinkFrameCounter: 20876
[1711873407.689260][10077:10079] CHIP:TOO: MleFrameCounter: 25814
[1711873407.689267][10077:10079] CHIP:TOO: Lqi: 2
[1711873407.689274][10077:10079] CHIP:TOO: AverageRssi: -80
[1711873407.689281][10077:10079] CHIP:TOO: LastRssi: -79
[1711873407.689288][10077:10079] CHIP:TOO: FrameErrorRate: 0
[1711873407.689295][10077:10079] CHIP:TOO: MessageErrorRate: 0
[1711873407.689302][10077:10079] CHIP:TOO: RxOnWhenIdle: TRUE
[1711873407.689309][10077:10079] CHIP:TOO: FullThreadDevice: TRUE
[1711873407.689316][10077:10079] CHIP:TOO: FullNetworkData: TRUE
[1711873407.689322][10077:10079] CHIP:TOO: IsChild: FALSE
[1711873407.689329][10077:10079] CHIP:TOO: }

My hope is that the ExtAddress is somehow constructed based on a node's HardwareAdrress and that I'm able to translate it back.

This is an example on where I can find a Matrer node's HardwareAddress:

$ ./chip-tool generaldiagnostics read network-interfaces 3 0
[1711874274.122194][10087:10089] CHIP:TOO: NetworkInterfaces: 1 entries
[1711874274.122250][10087:10089] CHIP:TOO: [1]: {
[1711874274.122259][10087:10089] CHIP:TOO: Name: OpenThread-f7ab
[1711874274.122267][10087:10089] CHIP:TOO: IsOperational: TRUE
[1711874274.122275][10087:10089] CHIP:TOO: OffPremiseServicesReachableIPv4: null
[1711874274.122283][10087:10089] CHIP:TOO: OffPremiseServicesReachableIPv6: null
[1711874274.122293][10087:10089] CHIP:TOO: HardwareAddress: A80B0120B9FE0A08
[1711874274.122302][10087:10089] CHIP:TOO: IPv4Addresses: 0 entries
[1711874274.122313][10087:10089] CHIP:TOO: IPv6Addresses: 0 entries
[1711874274.122323][10087:10089] CHIP:TOO: Type: 4
[1711874274.122330][10087:10089] CHIP:TOO: }

[jwhui]

Per Thread Specification, the IEEE 802.15.4 Extended Address is randomly generated when the Thread device is commissioned into a Thread network.

[olavt]

It actually looks like the ExtAddress is a real MAC address and not just some random number. Not sure if anything changed on the Silicon Labs side, but output from commands like "ot-ctl neighbor table" have a heading named "Extended MAC" for the ExtAddress field, so I guess it's a real MAC address after all.

Correct?

[abtink]

The ot-ctl neighbor table is a Thread CLI command that lists information about the neighbors of a Thread device. As Jonathan mentioned, the IEEE 802.15.4 Extended Address is randomly generated by each Thread device when it is commissioned into a mesh network.

[olavt]

@abtink Well, an ExtAddress that is matching the Thread device MAC address can't be random.

Do you mean that the specification specifies it as a random number, but the Silicon Labs (or Matter) implementation have decided to use the MAC address?

[hemanth-silabs]

@olavt I think there is a misunderstanding here. Thread devices do generate extaddr in random. And that is used in as 802.15.4 MAC address. In Openthread you have extaddr cli to print out the extended address generated at random and being used, but you can also check the devices EUI64 (HW MAC address) using eui64 cli. The output will differ for both. The devices on the network won't have any knowledge of each others EUI64, so the info printed on the neighbor table is the extaddr of the Neighbor and not the HW MAC address.

[olavt]

Ok, that makes it clearer. That means I can actually use the ExtAddress of the NeighborTable entries to find the actual Matter Node id like I do here (the NeighborTable of the ThreadNetworkDiagnostics info used is for node 2):

Here I use the ExtAddress 8A161990E4DC52EA to find the Matter NodeId by looking up the MAC Address for the network interfaces in the GeneralDiagnosticsCluster of all the nodes. In this case it belongs to node 1.

The ExtAddress 7E3CC94BA8670D87 is not found and is the OTBR MAC address (not the physical NIC, but a generated one).

This makes it easy to present the NeighborTable of the ThreadNetworkDiagnostics info with information about what Matter node id the entries are representing.