Re: L2 trigger interface
Jean Tourrilhes <[email protected]> Fri, 27 Sep 2002 15:28:26 -0700
| Newsgroups | gmane.ietf.pilc |
|---|---|
| Message-ID | <[email protected]> |
On Fri, Sep 27, 2002 at 03:15:38PM -0700, Rajeev Koodli wrote: > > Hello, > > I read your posting on the [pilc] list about L2 > triggers and found it relevant for the current > on-going discussion on fast handovers in > [mobile-ip] list. > I also scanned through the airo.c file in our > linux tree (2.4.19) looking for the place where > the driver might make a decision to switch from > one AP to another. But, I could not really find > it. Perhaps it's done in firmware (?). Correct. We only get notified at the end of the procedure, and the driver just pass the event up. Simular setup in the other driver/cards (at least, 802.11b - Orinoco, HostAP). > Extrapolating from your posting below, could > your interface provide an "up call" to a "client" > such as Mobile IP when SNR falls below say 50% > (taking hysteresis into account) ? You need to poll SNR/RSSI yourself, do the averaging and define your own threshold. iwconfig or iwspy will give you the SNR values. The threshold you use may have to be set depending on the card/application. > And, could it > also supply the BSSID of a prospective AP whose > SNR is better ? You need to perform scanning to know what AP is better (you can't be on two frequency at the same time), and while you perform scanning your IP traffic is on hold. The card will trigger scanning automatically as needed (in firmware, cf. your first question). The NEWAP event (end of L2 handoff) contain the MAC address of the new AP, which is also its BSSID. You can also manually trigger scanning via the new scanning API (iwlist scan) and see the full list of APs. Same deal, traffic is interrupted while scanning is happening. > I think such an API would be > very useful for Mobile IP. There is already versions of MobileIP that use the Wireless Extension SNR stuff to trigger handoff (see link on my web page). But, we are talking of actual code that is working with actual cards, so you just need to try it out in your own setup ;-) > Best Regards, > > -Rajeev Have fun... Jean _______________________________________________ pilc mailing list [email protected] https://www1.ietf.org/mailman/listinfo/pilc http://www.ietf.org/html.charters/pilc-charter.html http://pilc.grc.nasa.gov/