Macro network offload has been one of the hot topics at the recent Mobile World Congress in Barcelona. 3G and LTE femtocells are one interesting way to move voice and data traffic from cellular macro networks to fixed line connectivity in homes and offices. Another equally interesting possibility is Wi-Fi and it has some advantages as well.
Wi-Fi is the air interface of choice today for office- and home networking and is widely supported on smartphones as well. However, it still misses some features for the purpose:
First, what's still missing are intelligent clients on mobile devices that automatically switch over to Wi-Fi whenever available. In an ideal case the IP address and all open TCP and UDP connections would be kept when switching between the 3G / LTE macro network and Wi-Fi. For seamless switching, an IP sec tunnel or similar would be required on the Wi-Fi side. It would also be a good remedy for the inherent security issues of non-encrypted hotspots. Not impossible but also not straight forward to implement either.
Another thing that is missing is voice capabilities over Wi-Fi. One the one hand this would help to offload further traffic from the macro network. On the other hand, Wi-Fi networks could be used as an extension in areas where the macro network can't reach.
And finally, what comes to my mind is that Wi-Fi networking is very diverse. While the phone could be pre-configured for a network operator's Wi-Fi and the private Wi-Fi at home, using a wireless network at a friends house or at another office requires some user interaction. Not the best thing to make this work.
While built in intelligent “switching applications” are likely to be only a matter of time before they appear, the other two things are a bit more tricky to pull off. On the voice front, VoLGA and its brother technology GAN (Generic Access Network) might hold an interesting answer. Kineto demonstrated their Voice over LTE via Generic Access solution over a real LTE network in Barcelona and since it's based on GAN, it can easily be adapted for Wi-Fi as well. VOLGA for voice calls while the mobile device has LTE coverage, GAN while the Wi-Fi network is near and standard circuit switched voice calls while under 2G or 3G coverage, all seamlessly integrated into the device without too much effort. From application layer down to the protocols, all is the same, independently from the access technology. Only the lowest layers require adaptation. A straight forward solution.
And for using Wi-Fi at a friends house seamlessly, Bzeek might hold the answer. The software from an Israeli start-up company transforms your PC at home into an access point to share your Wi-Fi network with friends.
Interesting possibilities, but agreed, it's still early days when it comes to intelligent off-load.
But GAN requires a switch on the phone’s voice stack to switch between “trad” and IP-based GAN access. If the phone vendors don’t do it (as they really haven’t), it’s not going to matter much. Now that VoLGA has been put aside in favour of IMS based solution, what’s the future of GAN? I know it’s a nice idea and all (I’ve even written a course on it), but if it’s not adopted it’s a mute point.
When talking about offload, voice should not be considered. Voice is not an issue for current macro networks and therefore smart offload solutions should focus on transparent data switch (your first point) and not on voice support (your last point). This is the technology direction 3GPP is heading with TS 23.261. A prototype of this technology was also shown at MWC in Qualcomm booth… I guess we are not as far as you assume for intelligent clients.
Hi Gerardo,
Thanks for commenting and the link to TS 23.261. I was not aware of it so far. After having had a closer look I can see the merit of the approach. To me it looks like it is still a bit of a thing for the mid- to long-term future as it depends on IPv6 and MIP. Sure, you can run IPv6 over IPv4 and then the original IPv4 over IPv6 again in scenarios were for example the WLAN only supports IPv4 and your bearer is only IPv4 but at some point having 4+ IP headers per packet makes things look a bit strange… I like that the TS mentions untrusted WLAN and the ePDG as an access method so the solution can potentially be used with any Wi-Fi access point.
Kind regards,
Martin