sdecoret -

How Singapore’s Niometrics is changing network analytics

By extracting granular data and emphasising privacy protection, Niometrics’ network analytics technology is being put through its paces in a variety of use cases

Social distancing measures imposed by governments around the world to combat the ongoing Covid-19 coronavirus outbreak are only effective if people stay away from each other when out exercising or shopping for groceries.

Besides deploying officials on the ground to enforce the rules, some governments are turning to network analytics to determine if their guidelines are being followed. This data may come from telcos, whose networks can offer insights on the concentrations of people at a given location.

Luis Rezende, vice-president for insights at Niometrics, a Singapore-based network analytics technology specialist, said the company has government clients that are already using its technology to ascertain if people are maintaining social distancing during this period.

Rezende stressed that the data is anonymised and processed to protect the privacy of individuals and is also used by telcos to improve service quality, as well as by retailers to identify potential locations for a new store.

He also singled out a client that is trying to understand how people are using different ride-hailing apps and forming a very granular view of usage patterns in big cities. “This information is very valuable for those [ride-hailing] players, especially if they want to understand how they compare with the competition,” he said.

Yet, Niometrics – founded by Kostas Anagnostakis when he was a researcher at Singapore’s Agency for Science, Technology and Research (A*Star) in 2009 – did not start out with those use cases in mind.

Anagnostakis’ research was primarily about extracting as much information as possible when analysing network traffic. When he saw the commercial potential of his work, he started the business with an eye on cyber security as the data could be used to detect anomalies and indicators of cyber attacks on networks.

“But there wasn’t too much traction in that market,” Rezende said. “It was a bit too soon for that.”

Today, Niometrics works with some of the world’s largest telecommunications companies, which use its technology to identify new digital businesses, improve customer experience and optimise their networks. Its technology analyses data from a combined base of over 500 million subscribers who consume and exchange over 60PB of data daily.

Rezende attributed the company’s growth in the market to three things – the granularity of information, an analytics layer that makes data easier to consume and an emphasis on privacy protection.

“When you are analysing traffic data, you’re trying to identify patterns that indicate that this person is playing this game or sending an email,” Rezende explained. “Broadly speaking, we call these applications, not necessarily in terms of applications you have in your phone, but how you make use of the internet.”

Rezende claimed that Niometrics can detect traffic data from some 27,000 applications, compared to between 3,000 and 6,000 applications typically supported by competing offerings.

But analysing data packets from a wide range of applications takes a lot of processing power, as well as time to classify them. “You need to be able to process and organise all the logs that are coming in very quickly. Otherwise, you’ll lose them.”

When you look at what we can’t do now with 4G because the current throughput and density are not high enough, or the latency is not low enough, what you are left with is not that large
Luis Rezende, Niometrics

Rezende said that’s where Anagnostakis’ research on high performance packet inspection, which underpins Niometrics’ analytics layer that quickly compares data packets against a large table of applications, comes in.

The granularity of data extracted by Niometrics’ technology puts telcos at a bigger risk of running afoul of personal data protection rules. In response, the company formed a privacy engineering team to embed processes such as encrypting personal identifiers from the outset to prevent data leakages in the data pipeline.

“Sometimes when the data gets too granular, you need to inject some noise into it to retain information but make it difficult for people to identify the owners of that data,” Rezende said.

Rezende believes Niometrics’ technology and its focus on network analytics will set the company apart from larger rivals such as Nokia and Ericsson which have established deep relationships with telcos over the years.

“We think we have something different to offer because we are specialists,” Rezende said. “Otherwise it would become even tougher for us to have conversations with the telcos.” 

Messy, directed 5G investments

By working closely with telcos, Niometrics has a good view of how operators are approaching their 5G investments.

Rezende acknowledged that enterprise 5G use cases at the moment are largely limited to the internet of things and teleoperations, given the technology’s ability to support more devices per cell and at lower latencies.

“When you look at what we can’t do now with 4G because the current throughput and density are not high enough, or the latency is not low enough, what you are left with is not that large,” Rezende said.

As such, he noted that telcos are making 5G investments cautiously and in a “very messy, directed way”.

Rezende said rather than build a nationwide 5G network, telcos are rolling out the technology in predefined areas where there may be a chance of something flourishing – even if these are non-standalone networks without a 5G network core.

“But the truth is there is no clear visibility,” Rezende said. “Our clients are doing it because it’s a natural evolution – though they are doing it cautiously because the business cases are not 100% clear.”

Read more about telecoms in APAC

Read more on Network monitoring and analysis

Data Center
Data Management