On Monday, I read Bernard Herscovich, CEO, BelAir Networks saying the following in RCR Wireless:
Wi-Fi is obviously a way to offload data to alleviate congestion, but it also contributes to overall network profitability by delivering data at a lower cost per megabit that traditional macrocells. ABI Research estimates that carrier Wi-Fi can deliver data at 5% the cost of adding cellular capacity. Perhaps the most important driver, though, is the fact that, properly designed and architected, a carrier Wi-Fi network will deliver a consistently great user experience. The implications of that on attracting and retaining subscribers are obvious.
We've also seen cable operators taking advantage of their broadband HFC infrastructure to mount Wi-Fi APs throughout their coverage areas, offering free Wi-Fi as a sticky service to attract and retain home broadband subscribers.
At the GSMA Mobile Asia Congress, back in mid-November, 2010, KDDI's president and chairman explained that while they would be migrating to LTE, which would double their network capacity, data demand in Japan was forecast to increase by 15 times over the next five years. So LTE alone, he admitted, would not be enough. A few weeks before that, European operators, including Deutsche Telekom and Telefonica, were making similar statements at the Broadband World Forum in Paris.
It is clear that LTE alone will not be sufficient to meet ongoing mobile data demand. Technical innovation has resulted in huge capacity gains, but we're now at a point where additional bandwidth is more of a by-product of incremental spectrum. And, we all realize the finite nature of that resource. So, based on this new spectrum, LTE macrocells could deliver a 2 – 4X capacity increase. Meanwhile, ABI estimates that data capacity requirements are increasing 150% per year.
So, it's pretty clear that carriers are going to need more than just an LTE swap out to keep delivering a great user experience. They need to, as many already realize, augment their licensed spectrum with Wi-Fi. KT, the second largest mobile carrier in South Korea, claims to be offloading 67% of their mobile data traffic onto Wi-Fi. There may also be additional unlicensed spectrum made available, at least in the U.S. and the U.K., through the release of so-called white space spectrum, freed up through the switch from analog to digital TV.
It is obvious from the technology point of view that Multiple PDN connections would need to be supported when the UE is using LTE for part of data connection and Wi-Fi for other part. In fact these two (or multiple) connections should be under the control of the same EPC core that can help support seamless mobility once you move out of the WiFi hotspot.
One of the items in 3GPP Release-10 is to do with supporting of multiple Packet Data Networks (PDN) connections for a device. A Release-9 network and the UE can only support 3GPP access based connection via EPC. In Release-10 support for upto 1 non-3GPP access has been added.
FMC100044 specifies the following requirements:
- The Evolved Packet System supports the following scenarios: a single Operator offering both fixed and mobile access; different Operators collaborating to deliver services across both networks.
- The Evolved Packet System shall support the access of services from mobile network through fixed access network via interworking.
- The Evolved Packet System shall be able to support functions for connectivity, subscriber authentication, accounting, Policy Control and quality of service for interworking between the fixed broadband access and Evolved Packet Core.
- The Evolved Packet System shall optimize QoS and Policy management meaning that it shall offer minimal signalling overhead, while interworking between the fixed broadband access and Evolved Packet Core.
- The Evolved Packet System shall be able to provide an equivalent experience to users consuming services via different accesses.
The Rel-10 work item extends Rel-9 EPC to allow a UE equipped with multiple network interfaces to establish multiple PDN connections to different APNs via different access systems. The enhancements enable:
- Establishment of PDN connections to different APNs over multiple accesses. A UE opens a new PDN connection on an access that was previously unused or on one of the accesses it is already simultaneously connected to.
- Selective transfer of PDN connections between accesses. Upon inter-system handover a UE transfers only a subset of the active PDN connections from the source to the target access, with the restriction that multiple PDN connections to the same APN shall be kept in one access.
- Transfer of all PDN connections out of a certain access system. A UE that is simultaneously connected to multiple access systems moves all the active PDN connections from the source to target access, e.g. in case the UE goes out of the coverage of the source access.
This work also provides mechanisms enabling operator's control on routing of active PDN connections across available accesses.
The scope of the work is restricted to scenarios where the UE is simultaneously connected to one 3GPP access and one, and only one, non-3GPP access. The non-3GPP access can be either trusted or untrusted.
The design of the required extensions to Rel-9 EPC is based on TR 23.861 Annex A, that provides an overview of the changes that are expected in TS 23.401 and TS 23.402 for the UE to simultaneously connect to different PDNs via different access systems.
3GPP TR 23.861: Multi access PDN connectivity and IP flow mobility
3GPP TS 22.278: Service requirements for the Evolved Packet System (EPS)
Old Blog post on Multiple PDN Connectivity