~ >However ... just guessing ... it would seem to me ~ >that, if we use the same SSID, that the STRONGEST ~ >should win, and, if one disconnects, it *should* ~ >(logically anyway) switch seamlessly over to the ~ >stronger signal as the person roams the home. ~ ~ ~ Fat chance. That's the way it should work. Instead, what happens is ~ that the client will remain connected to the initial access point, no ~ matter how weak or disgusting a signal it offers. Even if turn off ~ the client device, it will try to reconnect to the same initial access ~ point, even if there's a stronger/better signal with the same SSID ~ evailable. Even if you intentionally disconnect, the client will ~ retain the MAC address of the initial access point. When you try to ~ reconnect, it will try that MAC address first. ~ ~ Intel seems to have gotten the clue and offers a setting as to how ~ "aggressive" the client will act in retaining a connection: ~ ~ It's not a total solution, but does work rather well on my various ~ laptops.
I think you're being overly pessimistic, Jeff. The scenario where you have multiple APs advertising the same SSID on different non-overlapping channels (where all BSSIDs are bridged to the same L2 broadcast domain), actually works pretty well with most clients nowadays.
Our large customers often have buildings or campuses with dozens or even hundreds or thousands of APs all offering the same SSID, and most modern clients can roam throughout these coverage areas without losing more than one or two seconds of data connectivity at roam time.
Most clients do offchannel scans for other APs will associated, so they know what all other APs are out there (although the freshness of that information will vary.) They are apt to decide to roam based upon one or more factors like these:
- currently associated BSSID has dropped below a given RSSI threshold (e.g. below -80 dBm)
- there exists a better BSSID whose RSSI is more than threshold k stronger than the current one, so let's roam to it (e.g. k=15, so if the current BSSID's RSSI is -74 dBm and another AP is at -57, let's go there)
- n consecutive 802.11 retransmissions to the current BSSID have failed (e.g. where n=32)
- n consective beacons from the current BSSID have been missed (e.g. where n=10, i.e. ~1 second)
To be sure, in networks that are very large and/or have very stringent performance requirements (hospitals), exotic roaming schemes involving L3 tunneling, 802.11r, CCKM, WPA2 PMKID caching etc. can be called for. But for home/small organization networks, basic WPA2/PSK roaming across APs within a given SSID will work just fine (again, assuming that all of these BSSIDs are bridged to the same broadcast domain.)