A couple of weeks ago, we saw a huge furore erupt about the fact that femtocell-offloaded data traffic is sometimes charged against a user's quota, while most WiFi offload traffic is not. AT&T and Vodafone are both among the guilty here.
[A discussion for another day, but I think there's a great business model for someone to work out how to charge back to operators, for their use of *my fixed broadband pipe*, in this scenario, as I become the backhaul service provider].
But beyond the billing argument, there is another tier of discussion about policy. If I use (say) a Vodafone femto over a BT ADSL line, or an AT&T femto via a cable modem, then we essentially have a situation where "policy stacking" occurs. The mobile core network policy engine, and whatever is on the fixed network before it breaks out a peering point.
In the base case, this means that the user receives the lowest common denominator between the two sets of policies - although obviously the femto offload data is buried inside a VPN tunnel, so the fixed DPI will find it hard to see exact details. (If you have local breakout via LIPA / SIPTO it becomes more visible).
But that may just mean that the mobile core looks just like an encrypted connection to a web proxy server, from the fixed network's point of view. And so may therefore fall foul of some policy rules itself.
I guess there might also be scenarios where you get some sort of "ping-pong" effect as the two policy engines try to adapt to each other's presence unwittingly. Added to some sort of intelligent connection manager client, there are all sorts of possibilities I'd imagine.
With WiFi offload, another possibility occurs - if the mobile operator has sold some sort of "content control", perhaps a parent buying a device for children, or just the default settings on many devices - that ceases to work in many cases when attached at a hotspot. Something like I-WLAN could bring all the traffic via the mobile core for secondary policy enforcement, although that requires it to be supported in legacy WiFi access points and devices, which seems unlikely.
It may also be very easy to develop applications which split packets *between* WiFi and femtocell accesses on the device, thereby making it impossible for either policy engine to reassemble a whole bit-stream and work out what the user is doing.
One solution is the type of "managed offload" I discuss in the Broadband Business Models report, in which the fixed/WiFi operator is somehow able to "import" the mobile policy rules and enforce them, instead of its own ones.
A topic to be continued, I suspect.