Cellwize takes wraps off revenue-aware, big data SON product

An Israel-based company that is just emerging from stealth mode says it has solved one of the most pressing requirements in the mobile network today - by developing a SON engine that can automatically flex a network in real time in a revenue-aware manner.

Cellwize, which markets a set of SON-associated functions under the banner “elastic-SON”, says it it can perform real time analysis of network level data that can automatically optimise network performance, but do so in a way that the revenue implications of network performance.

The company is making all the normal steps associated with a tech start-up, attracting multi-million investment from an un-named Private Equity company, appointing a high profile Chairman with top contacts in the mobile operator space, and bringing in a mobile industry veteran from an associated field as CEO.

The difference is, according to CEO Ofir Zemer, most recently of customer data analysis company Pontis, that Cellwize already has a developed, ready-for-market product, 56 employees and three live operator customers. One of these, Cellcom, it can name. The others are still under wraps, although an operator in the central Asian republics will shortly be named.

So what does Cellwize do? Its data collection model is familiar enough to those with knowledge of centralised SON engines. It access 2G, 3G and LTE network data from across the network: log files from radio elements, RNCs etc., counters from inside the network equipment vendors’ own products. At the moment, supported vendors include Ericsson, Huawei and NSN – the three vendors most commonly seen in Cellwize’s customers’ networks. Other vendors – Al-Lu, Samsung, ZTE – will be added when they need to be, Zemer says. Analysis of this data drives SON functions such as ANR, load balancing and dynamic antenna calculations.

So far, so normal. It’s what comes next that Zemer claims is the company’s “secret sauce”.

In order to take into account not only what is happening in the network, but how valuable those subscribers are who are affected, it is necessary to process and correlate a large volume of data at very short intervals.

“The amount of data we can process in real time is humungous because we have built a very nifty software architecture that can process an amazing amount of data, in real time, in a very small hardware footprint,” Zemer claims. The company, he says, is in one current trial supporting SON use cases for 18 million subscribers on just two non-remarkable servers.

“Our algorithms take valuable customer data in a specific area and we are doing optimisation according to the value of the customer, in real time,” he added. In another trial, this time with a major Western European operator, Cellwize was given 100,000 customer id’s to monitor and asked to optimise the network to provide the best QoE for these specific users.

So how is Cellwize processing these “humungous”, “amazing” amounts of real time data? Has it truly cracked the nut of real time, customer-aware, automatic network optimisation?

Zemer explains what he terms the “secret sauce” of the company’s data processing architecture: “We don’t pile all the data inside and then start to query it, instead our algorithms sit on streams of the data and look for specific patterns. Once they are there they can start acting. Later on, we may query the database to produce reports to show what we doing. But the way the algorithms work is that we do not start by querying piles of data to work, we sit on the stream. That’s the nature of the architecture, there’s a lot smart engineering.”

On TMN we’ve written a fair bit about the amount of acquisitions of companies providing network optimisation and/or SON technology. These have included Arieso to JDSU, Amdocs buying Actix and Cellsite, Aircom going to TEOCO and Tektronix buying Newfield Wireless – all pretty recent examples. Going back further the two big marquee deals led by network equipment vendors have been Ericsson buying Optimi and Cisco dropping half a billion dollars on Intucell. We’ve also seen Infovista buying Aexio and joining a PE stable with Empirix.

All these acquisitions have been about taking SON and making it part of something bigger than merely planning, configuring and tweaking a network. Obviously, you can still do all that, but the greater game at foot here is to take that network level information and also use it in a customer-centric way: that might be in a direct manner such as feeding customer care agents with network level information so they can see what might be affecting a user’s experience. Or it might be sending subscriber data in the other direction – that is feeding SON systems with subscriber data so that networks adapt automatically and can target resources based on the revenue profile of subscribers on the network.

So far, an awful lot of money has been spent on trying to plug in RAN and network-aware optimisation platforms into subscriber data platforms and, most crucially, do something with that information in real time, or as near as possible. Why? Because operators would love to be able to do something meaningful to enable them to differentiate on quality of experience parameters at a customer level.

So when one company says it can do this, within a product platform that can be integrated on a couple of servers, is it too good to be true?

Going forward SON will be too narrow to hold on just by itself as a function in the network, there are a lot of peripheral systems and demands that make it sensible for SON to be part of a bigger offer. We’re well aware of that, and we’re preparing for the day when SON is embedded as part of a wider value proposition

“It is quite a claim,” Zemer agrees. “but [the proof of] the claim boils down to having a real operational system inside a given network, performing some of the value-driven use cases we can do. Sometimes that means we need to integrate with other another analytics platform the operator has, of course. Our claim is that that is there, and it is working.”

As for the strategic implications – does Zemer think he offers something in a productised way that major NEPs and “big data” customer intelligence companies are trying to stitch together through acquisition? What of SON in a wider strategic role within the network?

“SON is an extremely active market right now, with lots of RFIs and RFPs out there. Going forward SON will be too narrow to hold on just by itself as a function in the network, there are a lot of peripheral systems and demands that make it sensible for SON to be part of a bigger offer. We’re well aware of that, and we’re preparing for the day when SON is embedded as part of a wider value proposition,” Zemer said.

If that sounds like the company is looking to “do an Intucell” and become the latest SON acquisition target, Zemer claims that for now, he is just focussing on running the company, and is nowhere near actively pursuing an exit.

Originally forming as a services company (called Triangulum) in 2007, with customers such as AT&T and CellCom buying RF optimisation services on a consultancy basis, the company’s founders saw the opportunity to move to a more “productised” model that would act more dynamically in network optimisation.

They started operating a platform in Beta within Cellcom in 2010, achieving a full national rollout in 2012. 2012 also saw the full productisation of the platform and the foundation of a newco as Cellwize, attracting “multi million” private funding for expansion.

The founders include Daniel Dribinski, CTO, who has background at Schema and Orange, and Sasi Geva, VP Business Development, ex-Motorola, Orange and Alvarion.

Ofir Zemer joined as CEO in 2013, having previously been a co-founder of Pontis and worked at Comverse as General Manager of the Instant Communication Division.

Former Telekom Austria and Vimpelcom CEO, Boris Nemsic, joined as board chair in January 2013.

The company has 56 employees, has been deployed nationally by three operators, is in for current trials and expects to add another 2-3 imminently.