Backhaul means priority and quality, not just more capacity says Annie Turner. After one of longest run-ups in telecoms history, the mobile internet has taken off. 

Reassured by flat-rate data packages – rather than bewildering and expensive per kilobyte or per megabyte charging – everyone is beginning to enjoy internet access including email, from wherever they are.

An astonishing fact: three-quarters of all mobile web browsing is now done on the iPhone, despite its market share being a fraction of those Windows Mobile, BlackBerry/Java or Nokia/Symbian devices. The kick-start administered by Apple has been followed by an outbreak of dongles bundled with data packages, which are also proving very popular.

So, have operators finally arrived at where they want to be regarding mobile data usage, the reason for their fabulous investment in 3G? Not exactly.

Lance Hiley, VP, marketing strategy for microwave specialist Cambridge Broadband Networks (www.cambridgebroadband.com) explains: “The big issue is that mobile will become the primary means of accessing the Internet – and not just in developing economies, where there is already more mobile broadband than fixed. This is driving packet data traffic but not revenue. It’s a diverging graph – traffic up, revenue flat or down.”

Worse, there have been numerous complaints about service quality from those with dongle packages, including much slower than advertised speeds and interrupted sessions, which raises another spectre for operators – a rerun of the churn nightmare of the late 1980’s/early 90’s when users constantly switched suppliers in response to poor service.

The central conundrum is how to finance the necessary investment in backhaul capacity to cope with surges in data traffic but no corresponding rise in revenue. It is an acute problem for operators and equipment vendors alike.

Nokia Siemens Network’s (www.nsn.com), CEO Simon Beresford-Wylie didn’t make light of the challenges facing vendors at a recent press conference. He hammered home the point that businesses need to make sustainable profits and underlined the difficulty in doing so when, in mature markets, margins are typically at 30-35% and in high tech growth areas, operating expenses (OPEX), runs at 27-33%. This leaves earnings before interest and tax at nought to five percent making it very hard to run a sustainable network equipment business in the current climate.

Last year NSN pulled out of a GSM infrastructure deal with India ’s Bharat Sanchar Nigam Limited (BNSL), on the grounds that it would be impossible to make a profit.

NSN’s chief technology officer, Stephan Scholz, said at the same press event: “We expect a hundredfold increase in traffic in the next few years but we won’t see a hundredfold increase in ARPU. There will be a huge increase in the number of subscribers, but the next billion will have much less disposal income.”

Laurie Spiegel, director of product marketing at Telcordia (www.telcordia.com), adds: “The situation today is very tough indeed and prices are being driven down hard but it has always been the telecom challenge to build out a lower cost network to reach more people and offer more services. Carriers need to increase revenue by using the new capacity to offer more services, not simply by providing more of the same.”

Which of course is why operators are moving towards an IP core and away from more expensive and complex ATM systems.

Deploying technologies like Gigabit Ethernet equipment is simpler and less expensive to manufacture so mobile carriers can build their own infrastructure rather than relying on hugely expensive leased lines continues Spiegel.

However, cheaper boxes are not the whole story. Spiegel elaborates: “It’s much easier and therefore cheaper to manage the lower cost equipment as long as you have the right operational support system. This is because the OSS enables the devices to manage themselves more and identify themselves to the network, so the network knows the asset is there and can optimise the use of it.” The devices now have attributes like self-healing, rerouting and fault detection to reduce administrative and downtime.

Scholz thinks another essential ingredient is a new approach to the access network and how it interfaces with the backhaul infrastructure. He asserts: “We need flat network architectures in the access network and then aggregate them – it’s the only possible way to carry a hundred times more traffic without massive upheaval and expense.”

Hiley agrees: “Network architecture has to change, as does the approach operators take to designing networks and allocating resources within them. The fact is that backhaul problems are not primarily about a lack of capacity so much as a lack of planning: if the available capacity were used properly, it would work well enough.”

His company’s VectaStar solution uses point-to-multipoint (PTM), microwave links, as opposed to point-to-point (PTP), links - used in telecoms infrastructure, typically in tree and branch topologies for decades. A single access point reaches several packet microwave terminals by grouping the access points into hubs. Each access point antenna divides the transition paths into sectors of 90º, so four access points provide 360º coverage, each with a radius of six to 10km. This reduces the number of radios needed: if a customer covered 20 cell sites with a hub, they would need 24 radios instead of a minimum of 40 if they were using PTP technology.

“You can instantly share resources across radios and, as capacity and density rise, the spectrum requirement grows the total sum of the mean bandwidths of each sector in a busy hour for all sites,” adds Hiley. “Consequently PTM offers a natural trunking gain, aggregating traffic from many points into a single point and then passing through the TDM, ATM, Ethernet, layer 2 native IP or any other interface/network protocol converter to be carried across the backbone.

“Data is optimised at the cell site before it even starts its journey towards the core network: for example, we remove empty frame headers in Ethernet or by removing IMA overhead in ATM. We convert whatever we take from Node B and optimise it by a factor of between 20% and 80%, depending on the nature of the traffic and the format using our proprietary, over the air (OTA) protocol.”

Hiley claims as operators need less spectrum, less space on a tower and fewer radios using PTM, it can reduce capex by up to 44% and opex as much as 81% compared with other backhaul technologies.

Another technology increasingly making its presence felt in better management and use of capacity is deep packet inspection (DPI). Until recently, DPI was mostly used for governmental anti-terrorism surveillance and other security applications but is now being used to analyse network, application and user behaviour, service control and traffic management, as well as security, peer to peer control (such as for VoIP), and distributed denial of attack mitigation.

Peder Jungck, CTO of DPI specialist Cloudshield Technologies (www.cloudshield.com), says: “Until now, the telco market has been about vendors selling a box for each app – the Cisco approach. We’re saying it doesn’t have to be like that. Lots more is going on in Layer 7 because of SIP and it’s just not practical to do it that way.

“Our single platform CS-2000 can work around the usual RVSP mechanism for reserving bandwidth, so that instead of a VoIP call using 64kbps, it uses 16, meaning four more calls for the same bandwidth. On the other hand, it can intercept all incoming traffic and ensure each bit is routed appropriately, even altering the way data is carried if necessary – such as to accommodate a router that can’t cope with MPLS.”

No wonder Dr Elan Amir, CEO of Bivio Networking (www.bivio.net), another DPI specialist, predicts the total global DPI market would be worth between US$5 and US$7 billion (€3.2 to €4.47 billion), in the next five years, becoming the single biggest element of network spending, based on an aggregation of analysts’ figures for telcos, national security networks and other security applications.

During that same time, operators also expect to see huge increases in traffic with the arrival of femtocells, currently deployed in large scale trials and expected to take off in 2010.

Paul Callahan, VP, business development with femtocell company Airvana (www.airvana.com), comments: “With femtocells, every family or customer that installs a unit is adding capacity to the radio network and reducing the cost of the operator’s backhaul at the same time because the user pays for it in the shape of their monthly DSL subscription. This is the first time that a technology is adding capacity without a linear increase in cost, which is what happens when operators add a macro cell shared by a number of users, probably on flat rate tariffs.”

However, DSL performance varies widely, ranging from a few hundred kilobits per second up to several megabits per second, thereby causing further QoS issues for femtoells and putting the already creaking xDSL infrastructure under greater pressure. The most satisfactory solution would be to deploy fibre to the home/kerb/cabinet (FTTx). Yet NSN’s recent decision to opt out of Gigabit Passive Optical Networks (GPONs), demonstrates how difficult the company thinks it will be to justify the business case for FTTx in a world of flat-rate broadband subscriptions.And on we go…