One of the most common complaints I hear about the Wi-Fi Alliance is their lack of enforcement and testing around the 802.11 roaming process. The general consensus is that our WLAN’s would work much better if everyone played by the same rules.
Coming from a client vendors perspective, I can say this is far easier said than done!
I recently talked about the “stickiness” of clients at the WLAN Pro’s Conference (#WLPC) in Prague 2018. Thanks to WLAN Pro’s making all those videos freely available you can watch that ten minute talk here.
The crux of the issue is that not all clients are the same. They don’t all perform the same job. And therefore they have very different roaming and connectivity needs.
Do you think an IoT thermometer such as Nest or Ecobee needs to roam very often? And do they care about being disconnected for 60 seconds while this happens? Are there more than one access points in the home where this device is deployed anyway?
How do you think that compares to a VoWLAN client that is highly mobile and needs to maintain a constant stream of very low latency traffic?
They are worlds apart, and there are hundreds of other device categories in between with different needs again. So many so that it might be unreasonable to ask any single organisation (be that the WFA or an infrastructure vendor) to know what every device type needs in order to “own” the roaming process.
This is why I believe we have seen such poor client adoption of 802.11v compared to other amendments like 802.11k and 11r – because the client knows best what it needs and when it can roam, and it doesn’t make sense to hand that decision over to the infrastructure.
To further compound the problem, a LOT of client device manufacturers don’t even go through the WFA’s certification process. There is no incentive for them to go through the time and cost involved, especially if you’re trying to make a sub-$50 product. So if the WLA did have some way of testing/forcing roaming it might drive even more client device manufacturers away.
What we really need is a better roaming framework from IEEE for the 802.11 standard. LTE devices are far better at roaming than 802.11 because it is written into the standard. Remember though, it is the wide variety of client types that makes this challenging. You speak to any connected vending machine manufacturer about GSM/LTE roaming and they’ll tell you about clients at the cell edge flip-flopping between towers. And I’d predict that there is far more client diversification in 802.11 than LTE.
If the infrastructure is to start dictating roaming, which I don’t disagree is needed, then we need categories of clients, much like with WMM, where each one has different transition parameters. And far better (standardised) sharing of information between the client and the network, so that the right decision can be made.
Yes, testing by the WFA could be good, as then we’d have somewhere to direct purchasers when choosing client devices. But what we really need is a proper set of rules, carefully thought out, and then applied across the 802.11 standard so that all devices know and follow the rules. The WFA don’t set the rules, the IEEE do.