Celcite is one of at most half a dozen companies that specialise in SON (Self-Organising Networks) today – software systems that are the "auto-pilot" for mobile networks, configuring and setting parameters across multiple-vendor small and macrocell equipment to maximise system performance and efficiency.
We spoke with Chris Larmour, CMO at Celcite, to understand his perspective of the market status of SON technology generally – what do CTOs think of it, where is it deployed today and how is it evolving. We heard about the difference (and potential conflict) between Distributed (D-SON) vs Centralised SON (C-SON), discuss which functionality belongs in the small cell versus central servers and learnt about the reducing "trust gap" which needs to be crossed prior to mass deployment.
Brief recap on Self-Organising Networks (SON) technology
SON software tunes and optimises mobile network performance by setting configuration parameters in cellsites (both large and small), such as the maximum RF power levels, neighbour lists and frequency allocation. In some cases, even the antenna tilt angles are updated to adjust the coverage of individual cells.
Centralised SON (C-SON) software co-ordinates all the small and macrocells, across multiple radio technologies and multiple vendors in a geographic region - autonomously updating parameters via closed loop algorithms. Changes can be as frequent as every 15 minutes– this is partly limited by the bottlenecks of how rapidly measurement data is reported by RAN equipment and also the capacity to handle large numbers of parameter changes. Different RAN vendor equipment is driven from the same SON software. A variety of data feeds from the live network are continuously monitored and used to update system performance, allowing it to adapt automatically to changes throughout the day including outages, population movement and changes in services being used.
Distributed SON (D-SON) software is autonomous within each small cell (or macrocell) determining for itself the RF power level, neighbour lists etc. based on signals it can detect itself (RF sniffing) or by communicating directly with other small cells.
LTE has many SON features already designed in from the outset, with the X.2 interface specifically used to co-ordinate between small and macrocell layers whereas 3G lacks SON standards and requires proprietary solutions.
C-SON software is available from a relatively small number of mostly independent software vendors, while D-SON is built-in to each small cell or macro node provided by the vendor. Both C-SON and D-SON will be needed if network operators are to roll out substantial numbers of small cells quickly and efficiently, especially when more tightly integrated into the network with residential femtocells.
Celcite is one of the handful of C-SON software solution vendors. Founded some 10 years ago, it has grown organically by 35% annually to 450 employees. With major customers in both North and South America, the company is expanding from 3G UMTS SON technology and is actively running trials with LTE C-SON.
Quite a few companies are claiming to be in the SON space, but Celcite would argue that there are perhaps only half a dozen with the capabilities for credible C-SON solutions today. Few companies can point to live deployments. As with most software systems, 90% of the issues arise when something goes wrong and it's those "corner cases" which take time to learn about and deal with from real-world deployment experience.
A major concern is termed "Runaway SON" where the system goes out of control and causes tremendous negative impact on the network. It's important to understand when to trigger SON command and when not to. This ability to orchestrate and issue configuration commands is critical for a safe, secure and effective solution.
So how would you convince a CTO to adopt a SON solution?
We've found that practical examples and real-world demonstrations prove the benefits of the technology and convert sceptical CTOs to adopt SON.
One of our practical examples was conducted at a large stadium during a series of Madonna concerts. On the first day, the network was configured using traditional methods and performed adequately. On the second, our SON system took over and performed significantly better by distributing the traffic between available cellsite capacity.
A sign of the growing acceptance of SON solutions is that we've noticed a decrease in the "trust gap" for SON technology with new customers. In the past, we'd turn up with a "black box" and have to spend a lot of time educating and explaining how it worked to network gurus. There would be an extensive period of incremental testing and proving the solution over many months. By contrast, recently one of our customers allowed the system to go live within four weeks – running it in supervised mode for a short time, and then going fully automatic.
This reduced cycle time for adoption is not just pleasing for us, it's part of a wider network automation trend also found in other parts of network operation.
How do network operators measure the success of SON systems?
Ultimately, the key parameter is the end user/customer experience – so-called QoE = Quality of Experience. There are lots of ways to track this, and different operators have different views on how best to deal with it. Some of the older ways to measure QoE have become outmoded, with more sophisticated methods being introduced. A variety of primary data measurements are boiled down to a small number of secondary factors that relate more closely to what an end user would sense.
Network efficiency is also critical – if a SON solution can squeeze just a few percent of additional capacity out of the installed base of cellsites, it can quickly pay for itself.
As the number of small cells being deployed and integrated into networks grows rapidly, the challenges of configuring and managing them will also rise quickly. SON solutions are essential to facilitate the rapid deployment of large numbers of new small cells when tightly integrated into the existing network.
What are the "people issues" related to deploying SON solutions?
Change management isn't a new problem when introducing greater levels of automation – we've seen similar issues when selling automated cell planning etc. What happens in practice is that the new tools still need human operators and we still need people to run the SON systems themselves.
Where before, 80% of a network optimiser's day might be spent just moving data around, that drudgery will be removed and instead optimisers can spend their time on higher quality work. Our experience has been that resources are applied onto something else – jobs don't go away, they get re-purposed and people use their time differently.
How quickly is SON market evolving and what is the timeframe for SON systems being more widely adopted?
We've noticed a quicker uptake since Cisco bought Intucell in January 2013 – this really fired the starting pistol for the SON industry. I don't think it will be long before 9 out of the top 10 worldwide operators will have C-SON systems live and would expect the rest of the industry to follow suit within 18-24 months.
The vast majority of demand for C-SON today comes from the 3G sector, where there are currently no well-defined SON standards, and little possibility to retro-fit localized D-SON to individual macrocells. This makes a C-SON approach the natural choice for UMTS networks, and there have been some excellent performance gains in live 3G SON deployments to date. Additionally, given the wide range of network equipment installed in legacy 3G networks, a truly multi-vendor C-SON system that can effectively manage the 3G radio environment is essential.
Those who have already adopted the technology have become strong proponents and are looking to deploy nationwide. Having seen the benefits, there is a rapid push to achieve that elsewhere. While it might take 5 or 10 years before all operators adopt SON throughout, we have found that once it is in place it is quickly shows a huge economic advantage.
We'd say the 3GPP standards published today need to evolve further – not too far from today, we may be talking, within reason, about self managing networks. There's plenty more room for innovation and differentiation, built on an already strong baseline of standards and a collaborative process.