For years now, cellcos have been waiting for 3G to make good on its promise of high-speed mobile data services and the revenues that - ideally - come with them. Seven years after the first 3G licenses were issued, the wait appears to be over. HSPA has catapulted 3G from a measly peak speed of 384 kbps to as much as 14.4 Mbps. More to the point, end-users are finally showing interest in accessing the web via mobile devices as familiar web content brands like Google, Yahoo, Facebook and YouTube chase the mobile dollar, and new devices like the iPhone make the mobile internet experience a sexy proposition. Indeed, the iPhone has been credited for boosting mobile internet usage in the US since its launch a year ago. And that's just the EDGE/Wi-Fi model - the HSPA-enabled iPhone arrives in Asia and elsewhere this month. Meanwhile, cellcos already know that flat-rate data access has proven a major incentive for people to use mobile data. It's easier to understand than charging by the kilobit, and users typically surf more when they don't have to keep an eye on the meter. While cellcos don't necessarily want to become simple fat-pipe providers, flat-rate is an attractive entry-level ticket to ride the mobile broadband wave. There's just one problem - mobile users might actually take them up on it. A number of cellco executives sounded out warnings at this year's Mobile World Congress in Barcelona that they faced a real risk of becoming victims of their own success. O2 CTO Vivek Dev, for example, observed that iPhone sales were 'driving unheard-of levels of mobile internet usage, and the introduction of flat-rate data tariffs is expected to increase this further. Both of these place huge capacity demands on our networks.' ABI Research senior analyst Nadine Manjaro concurs that flat-rate plans raise new challenges for cellcos. 'Operators need to pay close attention to the capacity limitations of their 3G networks and the true cost of 'all-you-can-eat' data plans.' The chief problem, in a nutshell, is backhaul. Users might be able to get a 7-megabit link to the base station, but sitting behind that base station, typically, is a couple of E1 leased lines. Up to now, that's been enough to handle 3G traffic, but only because the majority of 3G customers have been using their handsets for voice, SMS and relatively low bandwidth apps like downloading ringtones, wallpapers and games. But now, as HSPA grows and more users set their sights on 'proper' mobile internet, whether via handsets, USB dongles or MIDs, an E1 backhaul link isn't going to cut it. According to Heavy Reading, backhaul capacity for cell sites in business districts alone will need to hit at least 20 Mbps by the end of this year, and as high as 40 Mbps by the end of next year. And that's just going to intensify when LTE arrives. 'From 3G to LTE, your backhaul requirements are going from 45 Mbps to 300 Mbps per site,' says Amir Zoufonoun, president and CEO of Exalt Communications. 'It's logarithmic, that's the problem.' Fortunately, this isn't news to cellcos. They've known for some time that one day, at some point, they would have to upgrade their backhaul links. And the reason they've been putting it off is fairly straightforward - backhaul is expensive, and upgrading it in the traditional way (i.e.
adding more leased lines) makes it even more expensive. So far, the backhaul has been holding water, so cellcos have been staying the course and assessing alternative technologies to address the backhaul issue. But now 3.5G and the looming specter or 4G have forced their hand, and the window for making a decision on a backhaul migration strategy is closing fast. No more E1s, please The one thing cellcos know they don't want to do is just add more E1s. As it is, backhaul is one of the biggest single operating expenses for cellcos (75% of mobile transport costs and 25-30% of total opex, by most estimates). That's because leased lines are pricey, and as you add more, the cost goes up proportionate to the extra bandwidth, not incrementally. The other problem is this: even if a cellco was willing to spend that kind of cash on backhaul, the ARPUs they can expect to make from flat-rate internet services are unlikely to make up the difference, says ABI's Manjaro. 'The cost of increasing backhaul capacity, handling more calls to service centers, and potential increases in customer churn can quickly erode any gains from flat-rate plans,' Manjaro says. That's why cellcos are already focused on balancing the books by cutting their opex as much as possible - to include finding cheaper backhaul options that decouple capacity and cost. So what are the options‾ There are several, as it happens, though there are some basic criteria to be met, such as the ability to support legacy TDM traffic. Also, according to Heavy Reading, cellcos are keen on shifting to Ethernet backhaul, citing reasons such as reduced cost per transported bit, high capacity, flexible connectivity and IP/Ethernet transport layer convergence. Ethernet is also technology-neutral when it comes to access transport and media, says Ilan Seidner, marcoms director at RAD Data Communications. 'It can be extended to cell sites over whatever connection is available, whether it's copper, bonded copper and xDSL, fiber, PON, Wimax, microwave radios or even hybrid fiber/coax.' But the chief attraction is the cost effectiveness, he says. 'While E1 price levels are directly proportioned to the increase in bandwidth, Ethernet connections are typically offered in 10x increments - 10 Mbps, 100 Mbps, 1 Gbps, and so on - and can be charged accordingly.' So, for example, a single 10BaseT Ethernet connection costing E750 a month could replace five E1s costing E250 each. That works out to annual savings, of over E17 million in transport costs alone for every million subscribers, Seidner says. Another attraction of Ethernet backhaul is pseudowire technology that creates tunnels within Ethernet to support legacy TDM and ATM traffic carrying the cellco's 2G voice traffic. And it also works on ay transport technology that supports Ethernet. Not enough fiber As for the actual backhaul technology, fiber is the ideal choice, but as Sprint Nextel found out when rolling out its Wimax network in the US, fiber isn't always available where you want it, and when it is, it's not cheap (although it is cheaper than the equivalent number of E1s). In Asia, it depends on the market. In China, for example, says Shen Donglin, VP of wireless technology for ZTE USA, 'we have a lot of fiber, which makes it easier to deal with the backhaul issue in general, but in some areas, especially in old buildings, backhaul is still an issue.'
Tim Conboy, area product manager of wireless infrastructure solutions for Radio Frequency Systems (RFS), adds that fiber has limited applicability in Asia because of difficulties in gaining access and ensuring the reliability and security of the link. 'When fiber is run in Asia, it's often as a 'spur', rather than 'looped' to provide a redundant link, because it's less expensive,' he says. 'Fiber is also prone to damage by natural disaster such as flooding, landslides and earthquakes, not to mention the old 'backhoe fade'.' An alternative to fiber, Conboy adds, is microwave, which is already somewhat commonplace in Asia, particularly in India. 'Microwave presents four key advantages for Asian backhaul: cost, speed of deployment, access and reliability, and - probably most importantly - it is very easy to establish loops for redundancy,' he says. Another attraction of microwave backhaul, says Zoufonoun of Exalt, is that it's an opportunity to turn backhaul into a capex expense. 'A carrier can build a microwave network for what it cost them to lease backhaul capacity for a tear, and they can do all their backhaul spend in one go.' Lance Hiley, VP of market strategy at Cambridge Broadband Networks, agrees. 'Usually you're leasing backhaul capacity from a competitor, or the regulatory environment is such that the price you're paying today won't be the price you're paying in the future, and the price could go up as well as down,' he says. 'So more and more operators are looking to building their own microwave backhaul solution to take control of the situation.' Even when comparing microwave backhaul solutions, there are a number of options available. Exalt, for example, offers microwave solutions in the unlicensed 2.4/5.2/5.8 GHz bands for markets where microwave spectrum might be hard to come by, while Cambridge Broadband is pitching a point-to-multipoint microwave solution. As Hiley explains, straight point-to-point microwave radios require dedicated radio channels between each point, but a point-to-multipoint solution allows multiple base stations to share the same link. 'You could have 20 cell sites aggregated into one single point, and those sites would be sharing spectrum resources,' Hiley says. That's significant partly because it requires 40% less capex than point-to-point microwave, and up to 60% less in opex, and partly because it allows network planners to dimension their networks for peak usage rather than mean usage. 'Peaks and troughs in network traffic rarely coincide, so we can allocate from one cell to the next as needed,' Hiley says. 'We can backhaul from one cell site at peak, then when it passes, that spectrum is available for other cells.' Hiley says Cambridge Broadband has already demonstrated this on Maxis Communications' HSPA network. Part of the secret of Cambridge Broadband's solution is compressing traffic before it goes out over the microwave link, says Hiley, who adds that the multipoint approach also picks up additional efficiency gains via statistical multiplexing (under which the more traffic is running on a network, the more efficiently the network carries it). In fact, regardless of the specific backhaul technology, traffic optimization is a crucial strategy to whipping backhaul links into shape, says Seidner of RAD Data. 'Statistical multiplexing allows operators to implement overbooking strategies, which reduces the actual bandwidth required to transport multiple services and frees up additional capacity and network resources,' he says.
Mix and match Of course, traffic management isn't meant to be a standalone solution for dealing with backhaul capacity, and indeed, for the most part, no single backhaul technology is designed to solve the entire backhaul issue single-handedly. Cellcos can (and already do) mix and match depending on the needs of each site. For example, as mentioned earlier, fiber might be preferable where it's available and affordable. Otherwise, microwave can fill in fiber gaps, either permanently or as a stopgap for future fiber, says Conboy of RFS. 'A very Asian solution might be to retain the legacy microwave link, upgrade capacity as required by 3G, and to afford some semblance of redundancy to the fiber spur,' he says. Hiley of Cambridge Broadband adds that point-to-multipoint microwave makes sense in some scenarios but not others. 'Because you can aggregate lots of cells under one tower, it's more suitable for dense urban environments,' he says. 'Also, it depends on the frequency band. Usually regulators allow you to do point-to-multipoint in the 10.5, 26 and 28 GHz bands, but you're not allowed to do it in other microwave bands.' In any case, cellcos are more likely to favor a gradual transition than a forklift upgrade, says RAD Data's Seidner. 'Many operators are opting for a hybrid 'HSDPA Offload' approach, where you have two traffic-dedicated transport paths: real-time voice and video over existing SDH/SONET or ATM backbones, and UMTS and HSDPA data traffic over a PSN.' In fact, he adds, cellcos looking to make use out of existing assets as much as possible could even turn to xDSL as an interim backhaul alternative to fiber, making use of technologies like EFM (Ethernet in the First Mile), or mid-band Ethernet over bonded copper. If nothing else, it's dirt cheap compared to an E1 line, though DSL has its own limitations, such as its dependency on loop length for optimal bandwidth speeds. And if all else fails, says Shen of ZTE, there's always the power grid. 'One proposal we are working on for Clearwire is to use powerline to do backhaul for its Wimax base stations,' he says. 'This doesn't support very high bandwidth, but it does help in particular situations like old buildings in where you cannot get fiber or other technologies in them.' Fiona Chau contributed to this story Go to the source One way to ensure you've got enough backhaul capacity: keep an eye on your users If backhaul is coming under pressure from heavier mobile data usage on the access side, one way to deal with it is to put tight controls on how much traffic users can generate in the first place. SmarTone-Vodafone, for example, is managing its 14.4-Mbps HSPA network - which is also offering QVGA-resolution mobile TV - via a bandwidth mediation solution from Comptel that essentially ensures QoS by watching out for excessive usage and throttling back bandwidth hogs. 'The problem they faced was how to ensure a sufficient quality of service level when there are users with unlimited plans for downloading,' says Comptel marketing director Olivier Suard.
'Specifically, they needed to make sure no one was misusing the 14.4-Mbps connection and constantly destroying the service levels for all.' The solution: collect real-time usage data, measure that against cell capacity and the user's price plan, and throttle them back if they exceed a certain amount of usage. While the solution is targeting the access link, Suard says it can also be used in the backhaul link as well.