FAQ: How data becomes insights

In the following post, we explain how mobile data can be turned into privacy-compliant insights.

What does Invenium actually do?
We, a team of 20 mathematicians, data analysts and software developers, research human mobility. We want to understand how people move and thus improve the world in which they move. Our data helps to provide more targeted public transport, to optimize traffic flow or to distribute tourist flows in a way that improves the experience of tourists and locals. All this is done by analysing large amounts of mobile signalling data, which is provided to us anonymously by our partner, A1 Telekom.

What are mobile signalling data?
Apart from an excellent word for the next round “Hangman”, mobile signalling data is technical data that a mobile operator (in Austria this is A1, Magenta and Drei) collects to maintain their infrastructure. Each provider uses monitoring systems to know how many cell phones are connected to a cell tower at any given time. This data, which is exchanged between the cell phone and the cell tower, guarantees optimum network coverage and prevents individual towers from being overloaded. Mobile signalling data is not only useful for users to make a high-quality call or to surf the web quickly, but also for us to analyse and understand human mobility. This data is exclusively technical data, not customer data. The SIM card identification number (IMSI), is replaced by our provider A1, in a TÜV-approved anonymization process, by a randomly generated ID, before the data is shared with us. Thus, it is not possible for us to establish a personal reference at any time.

Which information do you have from the users?
As already mentioned, the data is anonymized before we receive it. This means that we can neither establish a personal reference nor trace any telephony or surfing behaviour. For us, each cell phone is a randomly generated number.

How do you get the position of a phone from the mobile signalling data?
As a child, we all learned how far away a thunderstorm was – as soon as there was a flash of lightning, we counted the seconds until the thunder, multiplied it by 300, and that was roughly how far away we were. Cell towers do something similar. The approximate distance of a device is calculated from the time it takes for a signalling message from a cell phone to reach a tower. Combining the information from multiple towers gives the approximate position of the phone.

How precise is such information?

Somewhat more precise than the thunderstorm calculations, but still not perfect. The calculated positions can vary a lot and depends on the topographic conditions. The inaccuracy of the data ranges from a few hundred meters in densely populated, urban areas to a few kilometres in rural areas.
How do apps that know my position to “meters” work? These apps for navigation systems such as Google Maps, or activity trackers such as Runtastic use the GPS module built into the phone. GPS (Global Positioning System) is much more accurate and can narrow down a location to within a few meters. Mobile phone data, as already mentioned, is only accurate to a few hundred meters. To give an example – Google Maps recognizes at which hot dog stand on the main square in Graz you had lunch, but the cellular data only tells us that a device has been in the inner-city district for a certain period of time.

You claim not to use personal data, but isn’t it possible to identify me by my location data alone, especially if I’m the only person in the area?
No, it is absolutely impossible. Our baseline data is too imprecise to determine the exact location of a device. In addition, the randomly generated ID is deleted by our provider every 24 hours and replaced with a new ID, again randomly generated. Depending on the project requirements, people are output in groups in the course of the analysis procedures. This means in detail, that for example in the legend, the smallest unit “is smaller than 20 persons”. Thus, this can be one person in the defined area, or 19.

If you don’t know the exact locations, aren’t your analyses worthless?
Quite the contrary. Our intention is not to track individuals, but to analyse the mobility behaviour of groups of people. The patterns and currents of movement – the flow, so to speak, rather than the individual drop of water. This allows interesting insights to be gained:

  • Interaction between communities, cities, districts.
  • Where significant movements take place
  • Where do people come from, how long do they stay in an area and where do they go afterwards.

In order to be able to answer these questions using our analysis methods, mobile signaling data is an optimal data basis.

Is the data from A1 sufficient? Wouldn’t we need data from all mobile operators and thus all devices to be able to produce meaningful analyses?
A1 Telekom with Bob, Yesss! and Red Bull Mobile, has an overall Austrian market share of around 43%. This results in a representative cross-section of the population in each municipality. Since it can be assumed that the movement behavior of people from different mobile operators does not differ significantly, the available data can be extrapolated to the rest of the population based on the regional market shares.

Do you work DSGVO compliant?
Yes, of course. We also spent two years having our product and the algorithms used tested by all relevant institutions. Both the Austrian data protection authority and TÜV Saarland have confirmed our anonymization process as compliant with data protection regulations. The same applies to a study by leading legal experts at the University of Vienna (ISBN 978-3-903035-12-6).

Even if all this is true, how can it be guaranteed that your customers will not misuse the data?
All our customers, including the Austrian government which uses our data in the fight against COVID-19, only receive analysis results from us – never the raw data. Depending on the question, significant movement flows at the community level, for example, can be found in an interactive web application as a textual report or as a graphic. As already mentioned, the analysis is carried out in groups – between 20 and 40 devices have moved between the two selected municipalities in the defined time window.

Das könnte Sie noch interessieren

Highlights 2023: Research & Development

Continuous development - Invenium has been involved in several research projects this year. The results also improve various aspects of Invenium's products.

Transportation Insights Webinar

Webinar: Transportation Insights

The utilization of the road system, the efficiency of public transport, the economic evaluation of company locations as well as tourist destinations - human mobility influences almost every area.

How data becomes insights – Part 4: Answering questions

11.10.22 #General

In the last part of our mini-series "How data becomes insights", we look at the questions and use cases of our customers.