In my last post (here) I looked at some of the tradeoffs for data offload between radio spectrum, backhaul and core network available with current technology. None of them hit the sweet spot on all three. This time, I’ll start looking at advances in the standards arena over the last couple of years that allow technology combinations to address the whole problem.
A big talking point at this year’s Mobile World Congress has been how WiFi can combine with cellular (large and small cells, 2G, 3G, and 4G) in a Heterogeneous Network or HetNet for short. In this scenario the various radio links are aggregated together in order to deliver a better Quality of Service and data rate to the user, whilst allowing the operator to optimise their spectrum usage. The Small Cell Forum released a White Paper yesterday looking at the benefits of integrating small cell technology with WiFi. It can be found here and I was privileged to be one of the contributing authors.
From a small cell perspective, one advantage of a 3G HNB over a macrocell is that it incorporates much of the RNC functionality and so user plane data may be separately ciphered on the backhaul link, decrypted at the HNB and the HNB then ciphers the data for sending over the radio link. The same is true for LTE cells and WiFi. So we can have the happy vision of just sending the user plane data to a combined cell + WiFi AP and then the cell selecting which radio technology to deliver the data to the mobile over based on current local interference, capacity, desired Quality of Service and so on – a VoIP call could go over cellular whilst a file download might be allocated to WiFi. Here at ip.access we’re very enthusiastic on this distributed RAN concept.
To do achieve this there are a number of hurdles to overcome. Amongst these is identifying that the WiFi part of the UE attaching to the AP comes from the same device as the 3G modem attaching to the HNB. This can be done via EAP-AKA authentication techniques to get the UE identity over the WiFi link and then matching them up. However, the harder part is being able to choose how to deliver a particular radio bearer, since they are set up tied to a particular IP address, and each of the modems on the mobile will have a unique IP address. A technique developed in 3GPP Rel-10 to help is called IP Flow Mobility (IFOM) and it works by having a Home Address (HA) in the network and a Care of Address (CoA) sitting in a client above the modems in the mobile, and the applications on the phone use the CoA. Using DSMIP (dual stack Mobile IP) techniques the modem IP addresses can be bound to the CoA and then the HA has a choice on which route to deliver data over. In the case of a residential deployment, the HA can be put in the integrated femtocell+WiFi . This allows quite a simplification of the network architecture, indicated in the two figures below:-
Figure 1 below shows a simplified 3G network with a WiFi extension based on the 3GPP I-WLAN standard and focusing on a home use scenario, as used in my last post
In the figure the TOF is a Traffic Offload Function that separates external internet traffic from operator services data and routes it to the internet via a Packet Data Gateway (PDG), whilst the Tunnel Termination Gateway (TTG) handles access control, authentication and adding or removing wrappers to the user data to put it into 3GPP format.
Figure 2, on the other hand, shows a network with IFOM enabled in the handsets and the network for an operator offering a cellular + WiFi service, and significant economies can be seen, including a much cleaner single logical backhaul for an IFOM UE after integration in both the cell, and the RAN edge. It allows agile use of the radio spectrum and data to be taken back to common routers for those devices supporting IFOM for cellular+WiFi
Recently, the Korean operator SK Telecom has in fact launched a network with LTE+WiFi small cells, and an aggregation of the carriers providing over 100Mbps data rates. SK Telecom has been an innovator in the design and roll out of self-organising networks and aggregated networks, and whilst the power consumption of a mobile using WiFi + cellular together might be quite high, I believe that they may be deploying an LTE-based architecture designed on similar principles to IFOM.
Whilst IFOM allows cleaner use of the radio spectrum and efficient use of the backhaul, it doesn't strongly address operator-supplied backhaul load, so I hope to look at that next time, discussing local direct offload in the home or enterprise - LIPA - and to the internet - SIPTO.