WiMAX Directions: LTE can only dream as WiMAX starts to deliver the flat IP network Jan 20, 2010 – By Rethink Research
There are many benefits to having an all-IP network, and the wireless
carriers are finally to reap many of them, as they move to new generation
platforms like WiMAX and LTE. In theory, the greatest of these should be the
ability to deploy a flat, distributed network, just like the wireline
broadband providers have – open to any IP devices, and of maximum
efficiency in handling traffic. But how viable will such a network be for
many mobile carriers, given the huge legacy of the older wireless networks?
Even LTE is starting to look rather less ‘flat’ than was
intended, as it struggles to cope with non-IP challenges like voice.
In this situation, WiMAX carriers can benefit in two ways. Their networks can
get closer to the decentralized ideal, being built, generally, with no need
to integrate 2G and 3G structures, except in terms of hand-off. And as those
hand-off techniques improve, they will find a new source of revenue among the
very cellcos that face such hurdles moving to IP, many of which are looking
to offload traffic to a partner’s wireless system.
One of the most daunting challenges for mobile operators is handling the
explosion of data traffic, including multimedia and peer-to-peer, on their
networks – while delivering an acceptable user experience and making a
profit. The additional capacity of a new ‘4G’ network, and the
higher bandwidth of the modern technologies, is not enough – only a
tiny handful of carriers, Clearwire among them, will have anything like the
amount of spectrum needed to deliver the capacity and wide channels for
wireline-class broadband.
Superior spectral efficiency is important, to eke more out of the frequencies
an operator does access. And network efficiency is vital too, to ensure that
data stays for as little time as possible on the cellco’s system
– either because it is travelling swiftly through a flat, decentralized
structure, or it has been offloaded before it even hits the carrier’s
core network.
These twin approaches – flat networks and offload – are the best
hope of supporting 4G services efficiently. They also give the operator a
chance of making a profit on its offerings, and seeing some ROI from its 4G
investments. Offload reduces the amount of infrastructure that is required in
the hard-pressed urban metro areas, while efficient traffic routing allows
the mobile infrastructure to handle greater volumes of data.
The flat network:
The flat network promises many benefits, compared to the 3G approach, where
there are at least four stages between the base station and the PSTN. Most
commonly, these are the base station
controller, the class 4/5 switches and the core elements (SGSN, GGSN, mobile
switching center and home location register, and sometimes more). By
contrast, 4G networks look to reduce these steps to just one gateway and a
simplified IP core. A streamlined structure promises to lower opex, because
there are fewer elements to buy and support; decouple the cost of delivering
service from the volume of data; minimize latency; and allow the core and RAN
to be evolved and managed separately, for greater flexibility.
This leads logically to the same sort of approach taken long ago in the fixed
IP world. Here, most of the network intelligence is at the edge, not the core
as in mobile systems. Costs are reduced and efficiency increased with the use
of plug-and-play, commoditized equipment throughout the chain. All this has
the knock-on effect of supporting an open device model, with standard IP
clients, and their data, handled in a uniform way, without the need for
complex carrier testing. Barry Hill, VP of sales and marketing at mobile
broadband gateway start-up Stoke, puts it well: “The fixed IP guys
realized they must have a flat distributed network, with edge technologies,
content data networks – in other words, get the traffic on and off the
networks as quickly as possible, and do not have centralization in any
form.”
So how close to WiMAX and LTE get towards that goal? Although the two
platforms have similar DNA, including IP and OFDMA, the network structure
continues to be an area where they are quite different. This is more a matter
of history and politics than technical intent. WiMAX and its supporters tend
to come from a background in IP broadband and the open PC/IP device model. So
it is natural that the network has evolved in a way that reflects the
decentralized norms of fixed broadband.
Mobile WiMAX is very flexible in how it can be deployed, with options such as
partial mesh that help keep traffic away from the center. Even when it is
based on a central core, this is highly simplified. It has a standardized ASN
gateway to connect its access services network to its CSN (core services
network), which may be owned by different entities to ease open roaming and
wholesaling. Within the CSN, there are two critical elements, the Home Agent
and the AAA server for authentication and billing.
Compromises in LTE:
LTE gets some way towards the flat network too (as does HSPA+ when a direct
tunnelling option is implemented). The standard also includes partial mesh
options, and the possibility for a non-hierarchical design that eliminates
many of the elements of the 3G network. In a pure LTE roll-out, there would
be a similar structure to that of WiMAX – the RAN connected to an SAE
Gateway, linking to the IP-based Evolved Packet Core (EPC). However, this is
where the flat,
distributed network dream starts to break down for many carriers –
because most the majority of LTE operators will need to support legacy 2G and
3G systems and services for years to come.
Because of this, standards work has been heavily geared to integration
between old and new networks. While WiMAX can work alongside 3GPP/2 networks,
handing off traffic as required and supporting integrated billing and user
experience, the LTE community has aimed for a far tighter degree of
convergence. This has advantages for existing cellcos in terms of bringing
all their networks and services together within a single core, and allowing
for flexible roll-out choices like overlays. But in terms of the nirvana of a
flat all-IP broadband network, this sets them back many years behind WiMAX
carriers – with the consequent impact on their business models and
efficiency.
The LTE system, even before it is commercially deployed to any extent, has
come across several of the burdens of legacy systems. Continuing to support
2G and 3G, especially when it comes to non-IP (but vital) services like
circuit switched voice and SMS, involves major compromises on the purity of
the IP network structure.
Already, it is very noticeable that diagrams of LTE networks often look
rather less flat than they used to. The number of elements in the chain is
indeed reduced – to the base station or eNode B and the SAE Gateway or
core – but other elements appear around these, such as the MME for
mobility management and various specialized traffic management products. This
is because mobile IP just isn’t as simple as fixed, and additional
resources are genuinely needed if everything is to be managed in the core
rather than the edge.
For HSPA+, the solution is direct tunnelling, which reduces some elements in
the chain, but does not allow for offloading of GGSN capability, and still
leaves a lot of complicated protocol conversion going on. Vendor solutions,
notably Nokia Siemens’ Flat Ran, are more advanced, but lock the
customer into one supplier.
To make things worse, 3G operators are increasingly introducing advanced
techniques to stretch their current networks and spectrum further, such as
deep packet inspection (DPI), which can identify commercially valuable, or
mission critical, traffic and offload the rest. This is one example of how
the core network – a fairly simple affair in the voice-dominated days
– has become highly strategic, carrying out many of the functions that
enable the carrier to offer a wide variety of well targeted data
applications. But the downside is that, when so many activities take place
within the core, it becomes bloated and expensive, prolonging the
disadvantages of the traditional, centralized thinking on cellular networks.
Yet very few cellcos are in the position where they can cheerfully abandon
their old hierarchical networks and obsese cores, and move quickly to a
‘pure’ mobile IP system based on an IMS (IP Multimedia Subsystem)
core, or even a modern distributed architecture. IMS solves all kinds of
problems, but it is a major investment and migration exercise that the
majority of cellcos will wait as long as a decade to implement fully. In the
meantime, their hybrid networks look ever more complicated, and less and less
likely to deliver a profit, when compared to the flat, distributed, all-IP
systems of their WiMAX challengers.
WiMAX – solution, not rival, for cellcos:
Of course, there are many advantages to having a legacy 3G network –
brand awareness, customer relationships, control of national roaming
networks, spectrally efficient voice delivery, SMS-based
delivery of user services. But these benefits are reduced in scale, the more
operators’ actual profits become driven by mobile broadband data. In
the early days of 3G, national coverage and cost efficient voice were still
the USPs; a decade later, a carrier is more likely to please its investors
with high capacity networks in a range of key metro areas, delivering premium
services and advanced quality to the most demanding user bases. In this
scenario, the flat network and lack of legacy become the key competitive
advantages. In most cases, these type of deployments are being based on WiMAX
(Wi-Fi mesh has also been used, and was a useful testbed for some open IP
concepts, but has the disadvantage of uncontrolled spectrum and QoS).
Providers like Clearwire, UQ and Yota are seeing huge growth in data revenues
– their networks can handle the traffic without degrading the user
experience, and their cost of delivery is low. For national coverage and
voice, they can rely on partnerships with cellcos, sometimes their own
investors (like Sprint in Clearwire).
The response to this dilemma by the 3G cellcos could come in two forms (not
necessarily mutually exclusive). They could aim for the same cost and
performance benefits as their WiMAX counterparts by deploying LTE as a
separate, greenfield system from 3G, focused on high value metro areas, and
with just simple roaming and hand-off onto their legacy networks to provide
coverage. This puts them on a level playing field with WiMAX carriers –
apart from the cost of building and supporting two entirely separate
structures, with none of the cost efficiencies of the overlay or converged
core approaches. Given that many of the competitive advantages of established
cellcos lie in their ownership of deep customer data, this needs to be
integrated into the 4G system too.
Some operators will go for this dual approach, despite the cost and
trade-offs. The other option is to rely on a third party for the IP data
services, in order to reduce that upfront investment and risk, and cut
ongoing opex, sharing the burden with the network owner and potentially other
MVNOs. This is clearly the approach taken by Sprint at Clearwire (its costs
spread among four partners, with more likely to join); or KDDI at UQ. These
carriers do not get the total control of building their own 4G networks, but
they get a major headstart on the market and reduced risk.
Even for mobile operators that are not looking for a full joint venture with
a true mobile broadband network owner, the WiMAX players can be alluring
partners for the cellcos. We have seen that one important approach to
handling the mobile data boom is to offload as much traffic as possible from
the overstretched 3G systems, preferably at the edge of the network to avoid
stress on the core. Some cellcos are relying on Wi-Fi hotspots and metrozones
for this, notably AT&T, but a WiMAX partner brings all the same openness
benefits plus greater QoS and control. As WiMAX finds its way into many
multimode devices from notebooks to smartphones, the trend will intensify.
For instance, Italian WiMAX operator Aria is building a national, high
capacity network and already has Telecom Italia signed up as a reseller for
fixed wireless services, and plans to add TIM, the mobile arm, this year,
focusing on metro area offload. TIM plans to move to LTE over a few years,
but recognizes that this will be a long process, while the problem of
relieving pressure on the 3G networks needs to be addressed now.
Proponents of mobile data offload see the benefits of moving intelligence and
data routing to the edge and integrating with flat, efficient IP networks.
Manish Singh, VP of product line management at Continuous Computing, wrote
recently: “The solution is to move the traffic shaper out of the core
network into the same unit as the traffic offloader in the radio access
network. Not only does this vastly improve traffic management, it also allows
traffic shaping to become adaptive. This means that shaping can be carried
out locally based on activity at a specific base station, allowing operators
to improve the quality of their networks in a much more targeted manner. For
instance, the user experience can be monitored and tweaked dynamically and
adjusted for different locations and times of day, when cells may experience
different usage levels.”
Using such methods, up to 90% of traffic can be offloaded to WiMAX, Wi-Fi or
the internet. Unsurprisingly, such methods are finding favor in the Asian
markets where use of mobile broadband is at its most intense. UQ provides an
invaluable second network for KDDI’s more demanding applications; and
Korea Telecom is using Stoke’s mobile broadband gateway to integrate
its CDMA, HSPA, WiMAX and Wi-Fi networks, to create a ‘pool’ of
spectrum and wireless resource. In this way, KT says, it can leverage full
value from all its network investments by eking as much performance as
possible out of the older technologies, and enabling all the systems to
behave, in effect, as a giant pool of data capacity. This will also postpone
the day when the carrier needs to make heavy investment in yet another
network, such as LTE.
Such approaches are in tune with the broader IP/internet world’s
thinking, rather than grounded in the specific heritage of cellular networks.
While this cannot be ignored, the industry needs to move more quickly towards
flat, decentralized, multinetwork systems. WiMAX operators are showing the
way, and their efficiencies will goad more traditional carriers into action.
Ironically enough, that action may consist of partnering with the WiMAX
network owners themselves, at least for the next few years, rather than
seeking to match them megabyte for megabyte with an early move to LTE. In
this way, return on investment in 3G can be maximized and the life of those
systems prolonged, while LTE build-out can wait until this is less beset by
compromises, and the 4G standards really are promising to deliver the flat
network dream.
WiMAX Directions: LTE can only dream as WiMAX starts to deliver the flat IP network
