This post is by Zhi-chun Honkasalo from Nokia Solutions and Networks.
Imagine we are able to learn about the entire mobile broadband network in real-time. Imagine operators offering a personalized network experience every time a subscriber’s device interacts with the network. And now imagine telecom network big data as footprints…
Colossal amount of data – trails and stories emerging from bits
Big data is older than ancient history, dating back to when man started tracking and hunting animals.
A trace can tell a story of past events as well as offer insight to the near future – presuming we have the skills to observe, read and interpret. Traces can be comprised of minute variations of the surroundings – grass, sand, mud, snow – and quite often they are hard to weave together into a story. This is how the role of scouts and army trackers came about.
To understand the traces, you need to learn how big they are, how far apart, how deep, how symmetric and so on. We can determine where animals have come from and envision where they are heading. We can also be aware of a passing wolf and prevent it from catching our prey. Take a look at this paw print in the snow. How can we tell it’s a wolf? A specialized hunter can likely spot if it’s too big (11 cm) to be a dog (8 cm), but for thorough analysis, he or she needs big data consisting of e. g. snow conditions and the skills to analyze it.
Some 10,000 years ago, when man hunted for a living, he soon noticed the limitations of eyesight in tracking and he domesticated the wolf to help him. The canine nose works like an efficient big data analyzer of tiny amounts of chemical substances forming scent traces. They can “see” a trail which is completely invisible to the human eye.
Drawing the parallel to the telecom world, we need a tool to analyze the bits affected and left behind by various events happening in the network. Every time a subscriber’s device interacts with the network, it leaves traces at different network interfaces. This is not about “seeing” the private messages of end-users, but about understanding and managing the experience and services for those users in real time. To understand them and make use of the information, we need to learn how to see, read and interpret every trace.
Just to put it into perspective, let’s consider an average network of 10 million 3G subscribers during a busy hour. Here, we need a processing machine capable of performing complex analysis over at least 1 million transactional messages per second, roughly 288 GB per hour in data volume. “Rush hour” in a typical mobile broadband network lasts for 8 to 10 hours per day, and this amount of data repeats itself from Monday to Friday, and partially on weekends. This volume as well the speed is an order of magnitude higher than the current social media platform on the internet. So that’s a lot of footprints…
Recently, something remarkable happened on Twitter: On Saturday, August 3 in Japan, people watched an airing of “Castle in the Sky”, and at one moment, they took to Twitter so much that we hit a one-second peak of 143,199 Tweets per second. (August 2 at 7:21:50 PDT; August 3 at 11:21:50 JST)
To give you some context of how that compares to typical numbers, we normally take in more than 500 million Tweets a day, which means about 5,700 Tweets per second, on average. This particular spike was around 25 times greater than our steady state.
Real-time big data – enabling actions at the speed of the user
Traditionally, network-originated volume data such as element-specific performance indication counters and subscriber charging data records have been collected and processed on an hourly, daily or weekly basis using offline database technology by OAM systems. Up to now, the retrieval (or “tracking”) of data has been slow, infrastructure-intensive and non-real time.
New technologies such as in-stream parallel processing combined with scalable real-time databases and cloud storage mean we’ve found our tracking wolf – capable of acquiring and processing massive amounts of data as it occurs. And because its real-time in the network, the information can be used online within seconds.
This means its possible to analyze everything that is happening inside a mobile broadband network in real time, which opens up a world of new opportunity, from real-time traffic steering, automated network modifications, to intelligent pricing and immediate individual marketing. It allows us to go from the world of “one size fits all” to a truly individualized personal experience.
We may, for instance, use the timely information of subscriber tracks to make intelligent predictions, improving customer satisfaction by solving problems before they appear. Or we may implement automated anomaly detection to improve network throughput by solving hidden network problems. We can apply root-cause analysis in real time to cut down operator reaction times and minimize network outages.
Learn more about big data and NSN’s FutureWorks cognitive network development at our upcoming webinar on 24 April: Technology Vision 2020 ─ Personalize network experience to enable innovative business models.
You can read more on cognitive networks here.
Share your thoughts on the topic and join the discussion with @NSNtweets using #CSPCX #mobilebroadband #NSNnews #TechVision2020 #Bigdata #CEM #FutureWorks #Innovation #NSN #NSNperformance.