Backhaul: Smashing the bottleneck

With voice revenues on the slide, data services have long been viewed as the operators’ APRU saviour. Last year finally saw data overtake voice in terms of network load and with many operators now touting the benefits of affordable mobile broadband there is every chance that data volumes will soon dwarf traffic generated by voice. Crucially though, revenues generated from data services show no signs of spiralling upwards.

Adding greatly improved capacity for marginally better returns is not an investment that will appeal to carriers. For years now the industry has been making huge investments improving speeds in the air interface and subsidising the rolling out of compatible higher end handsets in order to improve the quality and range of data offerings available to the consumer. Backhaul solutions have steadily improved over time, but there have not been any major breakthroughs in technology or capacity.

Backhaul vendors are claiming-more vocally than ever-that a surge in data service uptake as a result of next generation network rollouts and a greater penetration of 3G phones in the subscriber base, and no doubt the relatively recent widespread introduction of all you can eat data tariffs, will produce an undesirable bottleneck in the backhaul.

“There have been some tremendous changes in the last 24 months,” says Shahar Gorodeisky, CTO of backhaul specialist Celtro. “Mobile backhaul was a small niche market, everyone was focused on the radio technology and radio network, being able to supply more capacity over the radio interface, so backhaul was just a transport access solution and there was no capacity or service limitation whatsoever. This all changed after the transition to 3G when capacity to the end user jumped up.”

“The backhaul issue is one of the top five issues that operators are talking about right now, and they need to sort it out,” warns Jeanette Fridberg, head of product marketing for radio networks at Swedish infrastructure vendor Ericsson. “One of the challenges is that there is no standardised way of doing things. There are next generation networks being built on top of legacy transport links.”

“The central question is how quickly will mobile broadband be taken up and how dramatic will be the increase of traffic on the backhaul network?” says John Lively, VP network infrastructure, at industry analyst Ovum. “If it is incremental, then there are plenty of solutions that they [the operators] can use without doing a full swap out on the network. If it is going to be like a two orders of magnitude overnight, then it is going to be problematic for them in the near term.”

Jeremy Steventon-Barnes, director of strategic solutions at backhaul specialist Tellabs echoes these views: “If you keep building backhaul in the traditional way, then the business case is not going to work for mobile broadband. When you build backhaul for voice, you’re adding fixed units of capacity for fixed units of demand. So effectively for every busy hour of voice traffic, I’ve got incremental revenue of a fixed amount and that makes me a happy operator. The cost grows in slightly lumpier ways than the revenue, but the two track each other in a fairly linear way. The problem with mobile broadband is the nature of the traffic is very much more bursty.”

One operator that has enjoyed an impressive adoption of its mobile broadband service is Singapore’s MobileOne (M1). The carrier introduced its mobile broadband service late 2006 and according to the firm’s CEO Neil Montefiore, it has enjoyed a 200-fold growth. As part of its backhaul solution in the past, the firm had, in the main, been leasing lines from local incumbent SingTel. However, it soon became apparent that rolling out its own fibre network would be more cost effective.

According to Montefiore the capex cost of rolling out its own fibre network is around S$100m, while the leased circuits-even with discounts-were S$40m per year. “It was a bit of a no brainer,” Montefiore told MCI.

Singapore being a relatively small urban city-state is, of course, a bit of a special case. However, Australia’s Telstra employed a similar self-build strategy to accompany its nationwide HSPA roll out. So, do two carriers digging their own trenches for fibre constitute a ‘trend’? Montefiore certainly thinks so: “The traditional incumbents, which are still present in most countries, providing backbone networks are not really prepared to price it to make it attractive to networks so I think the operators will just self-provide. Because data rates are growing so rapidly there is very little else they can do about it.”

Ericsson’s Fridberg agrees: “Leased line is the most expensive, but if you have your own E1s and T1s the price comes down. Or you can go for Ethernet. But rolling out your own network pays for itself. In Europe it is sensible to build out. The pay back is fast.”

Ovum’s Lively paints a slightly different picture. Australia, he points out-like the US-is characterised by vast tracts of sparsely populated land with population centres tending to be isolated from one another, so it lends itself to a wireline solution. “It is only natural to upgrade from copper to fibre at some point,” he says.

But in other places, such as Europe, operators have been providing their own backhaul-mostly via microwave-since the beginning, suggests Lively. “It would be a newsworthy story if those operators started to upgrade via wire, which for the most part I do not believe is the case.”

Laurie Spiegel, director of product marketing at OSS firm Telcordia, reckons that competitive pressures will lead to the price of leased lines coming down: “I think this is analogous to the incumbent carriers who were charging by the minute and also by the location for long distance calls. Because of competitive demands, they had to change their business model. I think one of the things you’ll see in the short to medium term is possibly the incumbent carriers looking to change their pricing models so they don’t lose that huge business,” she says.

“I would be on the lookout for that because these wireless providers are spending lots of money on backhaul, it means someone is receiving that money and those recipients don’t want to walk away from that business,” says Spiegel.

Some sources suggest another factor that could exert downward pressure on the price of backhaul: dark fibre. “You’ll see the likes of Vodafone and Orange saying they have got their own fibre cores, and they have switched their networks in the core on to that. The problem you have with mobile broadband is, yes you’ll need more bandwidth in the core, but the main issue is down in the access,” says Ian Goetz, director of core network solutions at backhaul specialist  Aircom.

Ovum’s Lively outlines the two technology options an operator has it its disposal. “The first is to throw more bandwidth at the problem. Additional bandwidth solutions are available from a variety of vendors and range from microwave, copper, fibre, or even satellite,” he says.

The other technology approach backed by a number of vendors is optimisation. Not all of the bits in a constantly flowing data stream are filled. An intelligent processor can filter out the unnecessary bits before pushing the stream into the network. Pieces of kit like this-typically known as a switch or access router-generally sit at the base station although they can be installed at an intermediate aggregation point and receive the traffic from multiple towers. “A lot of traffic optimisers claim to offer a 60 per cent improvement, which is impressive, but then if you are looking at a two order of magnitude increase in traffic, like some operators have seen, then it doesn’t solve the entire problem,” says Lively.

Seeking an incremental increase might be wise, and it’s a strategy that operators routinely employ-albeit as a means of maintaining higher margins rather than choking back the traffic load on backhaul-for new services. “That’s what operators did with GPRS; it was hugely expensive and it was steadily dropped for certain services as the network capacity was rolled out. MMS and WAP have got cheaper over time,” points out Aircom’s Goetz.

“What you’re suggesting is that the operators will control the demand on the network by controlling the tariff,” says Goetz. “Which is another part of the balancing act that the operators have go to go through. It only takes one operator with more bandwidth to play around with its tariffs and he will take the market. It’s challenging times for the operators to get it right in terms of rolling out the network versus holding the market long enough to do it,” he says.

Ovum’s Lively agrees that maybe it is worth forcing an incremental uptake, but only if the competition do likewise: “Where it gets really interesting is if you’re in a competitive market. You’ve just spent all this money on infrastructure and so has your competitor. Now you’re desperate to get people signed up. So if your competitor starts to offer unlimited data then it would be hard not to.”

And, in effect, that’s what we’re starting to see, certainly with mobile broadband offerings. The race to sign up customers is a land grab exercise that the carriers will recognise. With voice, coverage mattered initially until it became ubiquitous, now the carriers are looking towards network sharing. There’s every chance that mobile broadband backhaul capacity might become at least an anecdotal issue among subscribers. Of course, consumers won’t know or care about backhaul, but they will care about the experience they have in comparison to their fixed internet life and their friends’ and peers’ mobile experiences.

If carriers are sharing facilities for voice, then why not for backhaul? According to Ovum’s Lively there is little evidence of that happening today: “There are plenty of companies sharing passively, using the same tower, but that doesn’t address the traffic issue. Active sharing, so using the same boxes and transport facilities, is very uncommon today. People have tried, but no one is doing it.”

Aircom’ Ian Goetz agrees: “As you get to more ubiquitous coverage, then the network ceases to be the same differentiator. We’re seeing companies coming along saying ‘yeah, we need to optimise our radio networks not just radio but also the backhaul and also some of the more esoteric aspects like how long the lease, is the landlord good, is it in the roof of a hospital?”

However, Steventon-Barnes of Tellabs says he sees sharing as a trend that could develop: “A number of our customers include consortia in different countries which have been set up as network sharing entities. It’s definitely an option that some operators are pursuing to make the economics work.”

Backhaul would appear to be a growing problem for the industry and unfortunately there is no magic solution. Celtro’s Gorodeisky sums up the current situation best: “We are all in a learning curve right now. One complexity is delivering capacity, the other is transition of technologies. Maybe the vendors are a little ahead of the operators. But the vendors are helping push the operators. Whenever there is a challenge for the operators, it is good news for the vendors.”

Singaporean self-starter

Neil Montefiore, CEO of M1, tells MCI why his firm moved off leased lines

We started the project almost the day we launched M1 Broadband,which is HSDPA 2.6 Mbps, but priced so it was competitive with fixed line broadband. We thought initially it would be the roadwarrior market. But we were completely wrong. People are using it as their personal broadband, even though at home in a lot of cases they had fixed broadband. They’d still take the wireless broadband, because they could take it wherever they went because it is more convenient and strangely enough they didn’t use it when they got home. Which is a bit like what happened with phones.

Most operators launching broadband are pricing wireless closely to fixed and they’re seeing the same sort of growth that we experienced. But you really have to rethink your whole backhaul. You need to move to an all IP core networkand the point-to-point E1 leased line structure has to go. It just gets too expensive because you need so much capacity.

Since we launched, the downlink speeds have gone up to 7.2 Mbps and the uplink to 2Mbps. E1s are only 2BMBs, so it’s an inefficient way of doing things. You need to move to an all IP network, and that puts a lot of pressure on the infrastructure manufacturers, so the Nokias, Ericssons, Huaweis and ZTEs have to move. Some of them have already. There is a lot of change going on.

We realised that we couldn’t really carry on using leased circuits. So we looked again at what it would cost to build our own backhaul and we found that it paid for itself in two years. So we went out for tender toward the end of last year and then awarded the contracts.

About 90 per cent of our links to our base stations were E1s with 2Mbps links, predominantly supplied by the old incumbent SingTel, although ten per cent were from StarHub. We would self-provide occasionally using microwave or fibre optic core. In those days it was not that much cheaper to self provide, but then when the data rates went up, we found that the cost to self provide came down dramatically compared to leasing so that’s why we changed.

We’re linking microwave hubs to base stations, we still have some leased circuits, but now 90 per cent are linked by microwave, they go to about 600 hubs around Singapore and those hubs are linked by fibre optic core. And in Singapore’s case because it is shaped like a big diamond, we’ve got a figure of eight in the diamond to give redundancy.

Ericsson provides the microwave, and for the fibre optics we’re just digging the roads, so we haven’t actually finished the tender process yet. The project at the moment is digging the roads and putting the ducts in. It all takes time because you need the permissions to dig up the roads. We think the fibre project will be complete towards the end of 2009. But in the meantime we can link our microwave hubs using much higher speed and cheaper IP connections. We are not going to see a lot of cost savings in 2008. They’ll start coming in by 2009 and 2010.