Tracking and mobile data in use against the corona pandemic
In recent weeks, data on people’s movement behaviour has become a key factor in the fight against the pandemic. We present this data and rank it against emerging criticism coming from data protectionists.
Transaction data in times of standstill
‘Social distancing’ is the sad buzzword of recent weeks and will probably accompany us for some time into the future. Physical distancing and avoiding travel are the core measures to contain the effects of the Covid-19 pandemic. It can be formulated as a binding decree or as an invitation to which citizens can respond in the manner they consider to be appropriate. The latter case, in particular, poses the question as to how effective these measures really are. A quick and accurate estimation of the remaining contacts, but also the evaluation of large movement patterns within and between cities, can save many lives.
Mobility data and tracking service providers are currently offering their data and software solutions. The aim is either to record the current intensity of social exchange or to retrospectively identify possible points of contact for those who have tested positive.
The provision of mobile data to the Robert Koch Institute (RKI) has been frequently discussed, often critically, in various press reports. Data protectionists have also voiced concerns in this context, with scenarios of digital surveillance quickly being circulated. The concern is that in this time of crisis this would set a precedent for the future use of data by the state and related institutions.
Where does movement data come from?
Since misleading assessments of movement data and its potential have been made time and again, we would like to present here a compilation of movement data which could be used in the fight against Covid-19:
1. Mobile phone data: ‘Events’ generate data via mobile phone masts, which are used to evaluate movement profiles and flows within cities or between larger spatial units. Events can be the registration and de-registration of devices, changing mobile phone masts, a phone call, sending a text message or using the mobile Internet.
Precision here depends on the size of the radio cell and ranges from a few hundred meters to several kilometres. These data sets converge at the mobile phone providers and are independent of active participation on the part of mobile phone users.
2. In the case of smartphone tracking, the position data is recorded at the level of the mobile phone itself. Tracking this data requires the installation of corresponding apps by users.
This kind of data is often generated, for instance, when using map and routing services as well as ‘fitness trackers’.
During installation, the operating systems indicate that an app wants to access the mobile phone’s location service. Technically, the sensors built into the mobile phone (GPS, wifi, Bluetooth) determine the position. Tracking data is therefore much more detailed in terms of space and time than mobile phone data and can thus provide information about users’ exact routes and small-scale whereabouts.
3. Approaches are also currently being discussed in which special apps use Bluetooth to send out an ID only which apps on nearby devices store. If one of the participating persons subsequently returns the information ‘Corona positive’ to the system, all potential contact persons are informed retroactively. The corresponding systems therefore do not need to store the exact location.
4. In areas where cars are used predominantly, Floating Car Data (FCD) can provide a good picture of movement patterns. To generate FCD, the vehicle’s satellite positioning (e.g. GPS or Galileo) is accessed. Depending on the provider, the vehicle navigation systems transmit the data via a mobile radio connection, or fleet operators, e.g. logistics service providers or taxi companies, provide data from their own fleet management systems. Certain data can also be aggregated here to form movement flows. The spatial and temporal resolution is similar to smartphone tracking.
5. To evaluate movement patterns in public transport, which has unfortunately seen very bad press in recent weeks due to its alleged role as a ‘virus slingshot’, use of the Electronic Timetable Information Systems (EFA) could be a helpful addition. This data can be used to make statements regarding the concrete relations between source and destination. These data sets can now also be easily evaluated and conclusions can be drawn regarding people’s spatial and temporal movement patterns.
The potential and limitations of mobile data
In recent weeks, the media has frequently equated mobile phone data with smartphone tracking data. However, the two kinds of data are completely different: Mobile phone data is generated at mobile phone masts and is therefore only as fine-meshed (or as perforated) as the mobile phone network itself. Data from smartphone tracking, on the other hand, is generated directly on the device itself, thus achieving a significantly higher tracking quality and is therefore also more problematic with a view to data protection.
Handling mobile radio data is a science in itself. Together with Telefónica, Fraunhofer IAIS and MotionTag, civity is working in the xMND project to explore ways in which this data can be used for public transport. The coverage and alignment of masts are changing continuously and dynamically. Extrapolating the data set from the population (the respective contract customers of the mobile network operators) to the total population calls for a differentiated calculation, because this incorporates data on the market share of the operators by population group and region. What’s more, the two service providers active in Germany have a sophisticated data anonymisation system: Data points are always available in minimum sizes of five observations, i.e. no data sets are played out per source-destination relation below five movements with the same attributes.
To assess compliance with lockdown and travel restrictions, the providers’ standard data can supply a good evaluation. For example, it would be possible to say whether the postal code area 10997, where Görlitzer Park in Berlin is located, has had a constant volume of traffic in recent days or whether fewer people from other postal code areas were present there. Corresponding evaluations are now also used to plan transport services and can help institutions, such as the Robert Koch Institute, when it comes to assessing the situation.
On the other hand, any analysis of personal trajectories would not offer any added value from a methodological point of view, even if providers were to provide non-anonymised data to the authorities in this case. The reason for this is that coverage by mobile phone masts is too coarse to accurately determine direct interactions at individual level. Besides, the trajectories often oscillate between mobile phone masts.
The potential of other transaction data
Data from smartphone tracking can provide more precise position data here, however, this would require the informed consent of users as a basic prerequisite for this procedure. It is conceivable for individual data subjects to download their Google history data and make it available for analysis, in much the same way as map service provider Ubilabs described in its blog.
In applications like these, real-time tracking is also ultimately technically possible.
Similar to mobile phone data, FCD can also be obtained as a standardized data product or via dashboards. Analysing this data can be useful for rural regions up to medium-sized cities, enabling them to identify movement patterns and to quantify the potential for propagation. In big cities, however, the informative value of this data declines due to the lower modal split share of private cars.
Conclusion
Some of the technical solutions for supporting analyses to contain the Covid-19 pandemic are already available, and some new ones are currently being developed – for altruistic reasons, but certainly also with the addition of a pinch of marketing. The use of mobile radio data passed on to the Robert Koch Institute especially calls for a certain degree of data expertise on the part of those using the data, provided that the data transfers communicated so far are correct and are not in violation of basic data protection regulations.
Over the weeks to come, we will see what other solutions will be established and whether they can help to contain the pandemic.