Tackling the mobile data tsunami
Less than two decades ago, the idea of mobile communication was a novelty and the possibility of mobile data was not much more than a dream. But within the past three or four years that dream has really become a reality, and checking email or Facebook, sending photos, and watching YouTube clips on smartphones are now normal parts of everyday life.
Such a transformation has of course come about as the result of phenomenal developments in mobile technology behind the scenes. But it’s not over yet. Data over mobile devices is slower than people are used to at home, and there is a limit to what people can do over 3G networks – or even over the new LTE networks.
According to Ruben Markus, CEO of Israel-based IP Light, ‘the objective of mobile operators beyond LTE is to deliver to portable devices bandwidth that is comparable to fibre to the home. In principle this means hundreds of Mbit/s per device.’
He continued: ‘In order to deliver this kind of bandwidth, you need to increase the number of RF transmitters and receivers,’ adding that mobile operators usually address this with a combination of increasing the number of antenna sites and increasing the number of operating bands.
However, this all comes at a price – and, in a competitive environment, mobile revenues are not growing in line with costs.
This is an issue that operators are very concerned about. In a white paper first published in 2010, with subsequent updates, the operator China Mobile set out the challenges in detail, particularly in relation to the RAN (or radio access network).
Traditional RAN architectures consist of base stations (BSs) connected to a fixed number of antennas covering individual cells. Each BS needs to be able to cope with the peak traffic in that cell, but traffic varies hugely depending on the location and time of day. For example, cells in residential areas see a traffic surge at breakfast time and in the evening but during the daytime much of that traffic shifts to where people are working. Similarly, cells near sports or concert venues see traffic surges when there is a big game, or a popular band is performing.
Having individual BSs for particular cells limits the possibility of responding to changes in traffic demands. To date, cells have been required to be able to cater for their peak traffic, meaning that, in quiet times, processing capacity, and power, are wasted.
And, as mobile internet demands increase and technology advances, the number of sites increases. For example, China Mobile said that, in the past five years, it has almost doubled its number of BSs, to provide better network coverage and capacity. As a result, the operator said, the total power consumption has also doubled.
This all has major implications for operators working in an increasingly competitive market. ‘Traditional RAN will become far too expensive for mobile operators to keep competitive in the future mobile internet world,’ reported the China Mobile white paper.
In addition, there are capital costs involved in installing BSs at each cell site. According to China Mobile: ‘In general, up to 80 per cent CAPEX [capital expenditure] of a mobile operator is spent on the RAN.’
What’s more, as cellular telecoms technology evolves, operators are required to support multiple standards and frequently update their equipment, and there are high maintenance costs every time engineers need to be sent to cell sites.
Centralising in the cloud
A proposed solution that has begun to receive much attention in the past couple of years is to separate the processing part of the system from the radio head and to centralise much of the BS technology.
This creates a so-called C-RAN architecture, where the ‘C’ stands for ‘centralised’ or ‘cloud’. In C-RANs, remote radio heads (RRHs) are positioned at the cell sites and the baseband units (BBUs) are located away from the cell towers in a central location, which might be anything from 10km to 40km away.
With this set-up, each RRH is no longer tied to a specific BBU, saving space and power at antenna sites and making it easier to allocate processing power to where it is needed.
‘Once you get rid of individual RANs you simplify cost, power, maintenance and demands on technical staff. Basically it’s win, win, win, win,’ said Markus.
A key component of this new approach is having a good link between the RRH and the BBU. Optical fibre is an obvious choice for this so-called ‘fronthaul’. However, there are stringent requirements for this fibre. Information is transmitted as unprocessed RF over fibre signals. As a result, significantly more bandwidth is required than would be after processing; fronthaul networks need around 20 or even 30 times the bandwidth of backhaul networks to transmit the same amount of data.
In addition, in order to maintain the speed of response, people are used to with traditional RAN setups, fronthaul must deliver very low latency and latency jitter, over many kilometres.
‘How to achieve a low-cost, high-bandwidth and low-latency wireless signal optical fibre transmission will become a key challenge for realisation of the future LTE and LTE network deployment by C-RAN,’ wrote China Mobile in the company’s white paper.
There are several approaches for delivering fibre fronthaul for C-RAN. First, there is the use of dark fibre, which can be a good option where there is already plenty of fibre available. This has benefits for speed of deployment, but network extensibility could be a challenge.
An alternative is to use a solution based on wavelength-division multiplexing (WDM) and optical transport network (OTN). This improves the bandwidth for transporting RF over fibre using standards such as the common public radio interface (CPRI) protocol and enables a large number of cascading RRHs on one pair of optical fibres. CPRI is a digitised RF signal that can be carried over an optical network. It supports a number of line rates from roughly 600 Mbits/s to 10 Gbits/s.
China Mobile conducted its first C-RAN trial with partners in 2010 in Zhuhai city, Guangdong province, China and has since carried out several further trials with a range of cellular technologies and in a range of locations. The operator reported the results as promising. ‘By using new technologies, we can change the network construction and deployment ways, fundamentally change the cost structure of mobile operators, and provide more flexible and efficient services to end-users.’
‘Fronthaul is quite challenging to do. Fundamentally it’s quite simple, but the reality is that the synchronisation and low-latency requirements are really quite hard to achieve,’ observed Jon Baldry, technical marketing director for Sweden-based optical systems manufacturer Transmode. ‘Early developments in fronthaul have been in Asia, but have been almost exclusively throwing fibre at the problem rather than building an optical network. OTN multiplexing today can’t meet the synchronisation requirements to support the protocol, although there are advances coming along to address this.’
For Transmode, fronthaul presents an interesting opportunity. The company already had in its portfolio equipment that could be used, and has been testing it in these new applications.
‘The technology we use for fronthaul we’ve had for very many years and use for other things. It’s early days for fronthaul technology, but our equipment is tried and tested for us so it’s business as usual for us,’ he said.
This, he explained, is an advantage for scaling quickly. ‘Some fronthaul projects have the potential to be very big and there can be a risk in scaling manufacturing very quickly with totally new products.’
Transmode offers a range of components such as WDM filters for hardened conditions and it has a range of options for fronthaul.
First, it has its TG series, which are products for passive WDM. ‘This passive range is very standard – we’ve sold more than 70,000 of these around the world and have built up a huge range of options over the years,’ Baldry explained. ‘If you are doing a fairly short-distance fronthaul this is a good option; you just plug it in and it’s a quick and easy fibre exhaust solution. The only requirement is that the optics at the RRH and BBU to support C-RAN need to be WDM.’
The company also has a semi-passive WDM option, which is passive in the transmission channel but adds per wavelength optical monitoring and link status monitoring to the passive solution. This enables more real-time monitoring of the network to, for example, detect power loss at the cell site and the need for maintenance in the fibre plant, ‘giving passive some advantages that active networks have,’ according to Baldry.
In addition, Transmode has two active options. Baldry explained: ‘The disadvantage with passive is if you are trying to span a long distance or if you have a ring architecture with many drop points you might be pushing beyond the power of the optics. Active gives more flexibility with how to build the network.’
The company has had some simple transponder-based solutions in its portfolio for many years and is currently doing field trials with operators on live radio networks using these for mobile fronthaul.
It also has a multiservice muxponder-based solution that can take anything up to 10 ports. This, he said, not only saves fibre but also increases efficiency. ‘We’ve created a new image for an operator for testing purposes supporting multiple fronthaul and multiple synchronous Ethernet signals over the same 10G wavelength,’ he said.
According to the company’s latest quarterly report: ‘Transmode’s Mobile Fronthaul solution was successfully tested in the quarter, securing type approval by France Telecom/Orange Labs in Lannion, France. This equipment has been verified against 3GPP standards for LTE and interoperability against Ericsson and Alcatel Lucent radio equipment.’
‘We’re very proud of this development. It’s a real testament to the capabilities of this product,’ said Baldry, who added that a second operator has also run successful field trials with the company’s fronthaul solution. ‘It’s early days for fronthaul over WDM. We’re pleased to be pushing it with equipment we’ve had for many years.’
Israel-based startup IP Light, which develops OTN processors for WDM communication, is also excited about the potential of fronthaul for mobile networks and is working hard to address the challenges. Markus explained: ‘Optical mobile fronthaul is unlike classical optical networks. It has unique features, so you can’t just repackage what’s done for other things. In order to carry CPRI signals, there are very strict requirements with delay, latency, and jitter and you need to transmit digitally.’
He continued: ‘In looking at effective ways to integrate CPRI signals, operators are faced with two choices: proprietary or OTN.’
He noted that the barrier to using OTN is that the current OTN standards have a binary structure. This is a challenge because cellular operators use directional antennas, each covering either 180 degrees or 60 degrees. The signals from the latter – six antennas – do not divide neatly or efficiently into a binary transmission system.
‘At IP Light, we’ve come up with an idea of how to address the electrical parameters and we are proposing modifications to the OTN standard that will be optimised for CPRI signals,’ he said.
This solution, which he says ‘takes advantage of different features of fronthaul’, will be presented to the ITU.
‘Our proposal will offer significant advantages,’ he said. ‘It’s a new type of signal so it would not only offer a much better solution for CPRI Option 3 but there is a natural evolution to other CPRI signals. The benefits are tangible whichever rate of transmission operator goes for,’ he said.
‘We realised early on that we were catching what would be an enormous market at very early stages and that there is an opportunity to help shape its course. This is very exciting for a young company,’ he continued. ‘We’ve also incorporated some capabilities to enable to operate in various scenarios, using greenfield new fibre or existing infrastructure, including relatively low-quality fibre, where operators still want to ensure very high-quality transport of, for example, real-time video.’
Weighing up the options
With any new approach there will be a range of options, and Baldry of Transmode anticipates that this will continue. ‘Operators have tens of thousands of cell sites, all in different conditions. Ultimately you’re competing against dark fibre. If that fibre is deemed to be virtually free, that can be challenging. That’s why we’ll probably end up with a mixture of solutions, passive at short distance, active at longer maybe. Having a range of options will be an advantage.’
And this includes alternatives to fibre. ‘We would love to have fibre everywhere but we are realistic that not all cell sites are fibred up and that for some it doesn’t make business sense,’ he said. ‘Because CPRI is digitised RF it is less spectrally efficient and so you need much more bandwidth for fronthaul than backhaul. If fibre exists you’d use fibre, if no fibre then you’d use other solutions, although some of these still have some challenges to address due to the bandwidth requirements and demanding technical requirements.’
Real networks could also be an interesting mixture of approaches too. ‘One of the nice things with WDM is you can build a network carrying different types of traffic, including both fronthaul and backhaul, over the same fibre. You could draw a nice, neat diagram for these networks but that’s not the reality of networks as they totally overlap each other in practice.’
Baldry noted that, so far, conversations with operators have been about macrocells but that some of the ideas could be extended to small cells in the future.
‘At the start of 2013 not many people had heard of fronthaul; 2014 will be key for taking it forward.’
Markus agreed: ‘We have crossed the line of the question mark. The market that will be created with fronthaul could be enormous. To call it snowballing would be an understatement.’
Advantages of a C-RAN architecture
A C-RAN architecture, where remote radio heads (RRHs) are sited at cell towers and baseband units (BBUs) are centralised and connected via predominantly radio over fibre links, promises several benefits over traditional approaches.
One of the biggest anticipated advantages is in energy efficiency. Centralised processing enables the number of BS sites to be reduced, which brings savings on support equipment such as air conditioning.
Centralised processing should also help traffic to be allocated more efficiently between cells and reduce interference between RRHs. This should simplify load balancing and traffic sharing between sites because all the BBUs are in one place and, in some implementations, shared across all the RRHs using virtualisation. This is a particular benefit for new LTE networks where rapid communication between BSs is an important part of the standard. Optimising the use of each BBU and use of virtualisation also lowers power consumption because at times of lower use idle virtual BSs could be selectively turned off without disrupting the service.
In addition to power savings, the C-RAN architecture promises other savings in both capital expenses and operating costs. Aggregating BBUs and site support equipment in a few central locations simplifies management and operation. In the C-RAN architecture the equipment located at cell sites is simpler and lighter so requires less maintenance and support and is quicker to install.