CC2531: How to Force Delete specific entry in Association Table?

Perhaps it would help if I also specify that I am using HA 1.2.2.a version of the stack...

Any suggestion from anyone on how to remove entries from the Association Table then?

Hi YK,

Thanks for the suggestion. I believe there is a single NVM parameter that I need to set to true to enable child aging? Is there anything else I need to do?

I briefly tested it by writing to this one byte parameter (NVM location $3e) but didn't see any changes at all.

thanks!

The default timeout value is set as 256 minutes. Do you wait 256 minutes to see if it works? If you don't want to wait so long, you can revise END_DEV_TIMEOUT_VALUE to 2, which means it will age out not connected end device after 4 minutest according to the following timeoutValue array.

CONST uint32 timeoutValue[15] =
{ 10, // 0 10 seconds
2, // 1 2 minutes
4, // 2 4 minutes
8, // 3 8 minutes
16, // 4 16 minutes
32, // 5 32 minutes
64, // 6 64 minutes
128, // 7 128 minutes
256, // 8 256 minutes
512, // 9 512 minutes
1024, // 10 1024 minutes
2048, // 11 2048 minutes
4096, // 12 4096 minutes
8192, // 13 8192 minutes
16384 // 14 16384 minutes
};

Thanks for the additional info YK. So I have waited ~24 hours after enabling Child Aging and the results is a bit surprising - all the active devices that are still in the network and are sending messages to the coordinator disappeared from the association table. In fact, all but 3 devices are left and those 3 devices were deleted at least 6 months ago (based on notes / logs I kept when testing those devices).

So now the coordinator claims that it cannot talk to the "active" devices. And only some messages from the devices are received by the coordinator. So the entire network is not communicating / functioning properly anymore :(

Would forcing an active scan with the ZDO Network Discovery request help? And does Permit Join needs to be enabled for the scan to re-discover the devices?

thanks.

Actually, the default age timeout is 256 minutes which is not 24 hours and I had replied you can revise END_DEV_TIMEOUT_VALUE to 2, which means it will age out not connected end device after 4 minutes.

Well this is the issue... so I expect that after 24 hours, which is way longer than 4 or 256 minutes, all non-existent devices should have been aged out and "live" ones would be in the table. But that is not the case.

After setting NVM item 0x3E to 1 (from 0), I lost all entries in the association table (except the 3 non-existing/long-deleted ones) so I think it worked. But not sure how do I make it re-learn the valid associations.

When device starts and send beacon requests, it should be able to rejoin.

But that means I need to enable Permit Join on the coordinator?

Actually, you don’t need to enable permit join.

umm... that's interesting. Device is communicating - I can see some messages from the device but their addresses are not in the coordinator's association table. And if the coordinator wants to send them a message, I get a status=0xdc (no route).

What if the device is not re-started?

Is there a way for the coordinator to ask all devices in the PAN to rejoin?

In general, this depends on device implementation. A usually device would do beacon requests when it is power cycle or loses connection to the network.