Shirin Dehghan is chief executive and founder of Arieso, the provider of customer centric, location aware network management solutions for mobile network operators. She established the company in 2002, having worked within Vodafone Group as a technical expert for the company’s UMTS license auction team. That experience meant she foresaw – perhaps too early – the massive data consumption uptake the market is experiencing and started the company to meet CSPs’ needs for efficient insight into their users’ consumption. Here, she tells VanillaPlus how CSPs can make efficient use of the data their users generate, optimise network performance and improve end-user experience.
VanillaPlus: Why did you establish Arieso ten years ago? What need did you plan to address?
Shirin Dehghan: I started the company to focus on the then new 3G and the huge predicted growth in mobile broadband consumption. I based the decision on my experience at Vodafone where I was part of the team preparing for the UK’s 3G auctions, which gave me direct insight into the challenges that this new technology would pose. It was clear that it would generate masses of data that would need to continue to run over the same technical infrastructure, so I saw the need to plan for the impact that this was going to have on the network and, ultimately, customer experience. The first five years promised huge data traffic growth, although that took longer than I had envisaged to emerge into the market.
We started developing the products in 2002 but really this whole thing started in 2008 with the launch of the iPhone. Everyone knows the history and the really explosive growth and demand we have seen since that happened. Today, the highest data consumers are iPhone 4S users and the whole focus for CSPs now is around which subscribers are hogging the bandwidth. One per cent of subscribers now generate 50% of mobile data so the challenge for CSPs is in understanding where these subscribers are and efficiently managing the network in those hightraffic areas.
The capacity crunch that I predicted ten years ago finally happened. We’d spent the intervening time developing the solutions to this problem and we’re now solving the same problem across five continents. The market is coming to us.
VP: To what extent is the shortage of capacity a result of operators’ getting it wrong when it comes to network planning and optimisation?
SD: Traditionally, CSPs have been basing their network planning and optimisation on incomplete information. The light bulb moment comes when they recognise the source of data they are using is out-dated. For example, drive testing, with network centric data and key performance indicators, compromises the operators decision making when they deploy assets as the likelihood is that the network will have evolved since that data was collected and those KPIs developed.
It is really important to understand where the data hoggers are and which of them represent the hotspots of bandwidth consumption on the network because, if the hoggers are at the edge, they are causing far more damage. We estimated last year that the whole industry wasted $500 million in opex through poor siting of base stations.
A significant part of the reason for that is the incomplete nature of drive testing. Very few users consume bandwidth in the same way as drive testing. For instance, most people don’t download video while they are driving; they are indoors and traditional drive testing methods cannot monitor the network to this level of inbuilding granularity.
VP: Where did they start going wrong and how are they fixing the issues they face now?
SD: We think CSPs have been basing their decisions on poor data and that data has been costly to gather. The performance improvements achieved from conducting greater amounts of traditional testing wouldn’t justify the opex expense involved, because the cost of the equipment and processes used is so high. CSPs face increased demand on their networks and they have to address that efficiently. Many CSPs still don’t get it – they need new network performance engineering solutions.
That is only going to become more apparent as operators start to roll-out small cells to meet their users’ needs. You need a multitude of different solutions in this area because of the amount of capex involved. Small cells are of big interest as a way to offload data traffic but you can’t place them anywhere and they are not totally inexpensive. LTE is exactly the same. Now you have masses and masses of data rate but the more you give users in terms of throughput, the more they want. CSPs are starting to look at selforganising networks (SON), which will require far richer and more real-time inputs than traditional network-centric solutions can deliver.
Small cells, heterogeneous networks and LTE are ways to add more capacity than 3G is capable of delivering but what becomes important is that CSPs have to manage their network across 2G, 3G and 4G as a single network. There are a number of strategies needed to solve these problems.
VP: Have we reached a point where operators understand the scale of the problem or is there still an attitude that if the network’s operating, there’s nothing to worry about?
SD: The pressure on network teams is definitely intensifying. They shouldn’t think that just having LTE is going to save the day. At a time where there are calls for greater returns to investors you can’t have inefficiency, it’s inexcusable. There is now recognition of the issue. I believe it is now one of the top three things that CEOs worry about in every CSP I’ve spoken to . They want to know what they can use to have leaner network teams, and what they can do to have one network operation across all generations of network – 2G, 3G and 4G together. The issue is that CSPs now don’t have the solutions to address those.
VP: Does simply throwing money at the problem make it go away? If so, how will operators find the money especially when access to capital is constrained and they are faced with new network roll out expenditure with LTE?
SD: The solution is staring us in the face – the subscribers themselves. The subscribers understand their experience of the network and how much they use the network over time. No amount of drive testing or probes will provide that granularity. Every device in the network is providing vast amounts of data back from the end user which basically gives the CSP a very rich source of data that is specific to every single subscriber about when, where and how much bandwidth they are using.
This should be at the heart of every CSP’s business decision. We think this is so critical because being efficient about the capital a CSP deploys is a core issue. If a CSP doesn’t put capacity in the right place it has been inefficient, so you have to have an intelligent interface. Old-fashioned approaches are incomplete and they’re not going to give you the whole picture, they’re old school and they’re out of date.
Understanding how the customer demands capacity from the network is absolutely critical. You can use that in real-time for marketing, for assuring the customer experience and to gain a deeper understanding of corporate customers, for example.
We were the first to realize this and we’ve worked on this for a very long time. This is a problem that is not unique to particular operators and geographies; it’s everywhere. In developing countries even more focus is required because fixed data infrastructure isn’t there so users have to use mobile. We see a lot of opportunity in those markets.
VP: What business cases can be constructed in support of systems such as Arieso’s?
SD: We’ve dealt with many established tier one CSPs, and CSPs that have grown to become tier ones. You have to prove your worth. We helped mobile operators to sweat their assets – we’ve improved key performance indicators by 20% without adding hardware. We add value through creating efficiency. We make use of the engineering team’s time and cut drive test costs. You don’t need to do it because every device is doing it for you. Finally, we present a 20 times return on capital invested in small cells. The business case is obvious.
