Data extraction from central storage systems is the easiest way to obtain data for national level statistics. There are several places where such data can be found, collected for different purposes and containing different kinds of information:

  • The billing domain stores call detail records (CDRs) which are outgoing/incoming calls, short message service (SMS) and internet protocol detail records (IPDRs) for charging purposes. Data will be stored here after a successful charging record generation procedure.
  • Customer databases contain information about users, which can add extra value to CDR or IPDR data – e.g., socio-demographics and place of residence.
  • Data warehouses host collections of data from different sources but might not always be easily accessible due to the huge amount of data stored there.

Databases in the billing domain are generally considered most easily accessible. Data stored in the billing domain is gathered by a mediation system from different network entities responsible for providing various types of services.

There are two types of probing for monitoring and acquiring data from telecommunications networks: active and passive mobile positioning. Active probing is used by the MNO to generate traffic to monitor the overall network performance. In other words, active positioning locates the owner of the phone in real-time and requires the owner's consent. In contrast, passive mobile positioning data do not require personal contact with the people involved and yield a large amount of anonymous data. Passive probes monitor data flows between different network entities, extracted data represents information about historical locations of the phones from the log files of MNOs. This latter method allows for a longitudinal view of all subscribers and is usually the preferred option.

The simplest method for passive mobile positioning is "a billing log" that is recorded for call activities. Any active use of a mobile phone (call and SMS messages in and out, GPRS, IPDR, etc.) is deemed to be call activity.

Capabilities of passive probing are usually achieved by deploying licensed software together with the required hardware to the network. There are several possible locations for such probes, but they are most commonly as local as possible in the system. These probes can also query databases, such as the Visitor Location Register (VLR), that store relevant user data.

In terms of increasing the granularity of observations in data, there are many benefits to using data from probes which are as follows:

  • It gives access to data types that will usually not be stored in billing centres and data warehouse systems.
  • It increases the number of records per subscriber helps to minimize diurnal and daily differences in record counts which are the result of subscribers' calling patterns.

However, using data from probes (signalling data) has its drawbacks:

  • Expects monitoring or data acquisition systems to be "online" in order to constantly gather and store the acquired data that is temporary in nature;
  • Additional network load for MNOs;
  • Not all information is mandatory for an MNO to store during probing (e.g. location area identity (LAI) is mandatory, but the more precise cell global identity (CGI) is not);
  • Increases the number of events per subscriber considerably, meaning that more resources need to be dedicated to the storage and processing of such data;
  • Involving system-generated data in statistical analyses might result in high levels of background noise that needs to be addressed in the data processing phase.

The cost of installing the probing systems in the MNOs' systems is usually high and it is not implemented simply for generating statistical indicators for the NSIs. But if the MNOs have already implemented some sort of a probing system and the required data storage, it is a good source of data for statistical indicators as it involves much more data compared to "traditionally" collected CDR from a billing system of a data warehouse. Probing data is limited to inbound roaming and domestic data, normally signalling data is not exchanged between MNOs, which is why it is excluded from outbound roaming data.

Signalling data generally refers to directly obtaining transmission signals from the radio network and storing it in a database. Similar to probing data, signalling data is also very voluminous and can be limited to inbound roaming and domestic data, excluding outbound roaming data. Signalling (or other data from probes in the network) is not critical for detecting internal migration, since it concerns long-term patterns, which can be studied with billing data. Signalling data is useful to detect international migration patterns because of the cost of international roaming. In that case, signalling data can be limited to inbound roamers, until they switch to a local SIM card. This can lead to an optimal balance between high and low volume data.


  • No labels