Cell-Site Analysis - Forensic Data Science Laboratory

	Forensic Data Science Laboratory
Home	Cell-Site Analysis

Selected Publications

Tart M., Brodie I., Patrick-Gleed N., Edwards B., Weeks K., Moore R., Haseler R. (2021). Cell site analysis: Use and reliability of survey methods. Forensic Science International: Digital Investigation, 38, article 301222.
https://doi.org/10.1016/j.fsidi.2021.301222

Bosma W., Dalm S., van Eijk E., el Harchaoui R., Rijgersberg E., Tops H.T., Veenstra A., Ypma R.J.F. (2020). Establishing phone-pair co-usage by comparing mobility patterns. Science & Justice, 60, 180–190.
https://doi.org/10.1016/j.scijus.2019.10.005

Tart M. (2020). Opinion evidence in cell site analysis. Science & Justice, 60, 363–374.
https://doi.org/10.1016/j.scijus.2020.02.002

Tart M., Pope S., Baldwin D., Bird R. (2019). Cell site analysis: Roles and interpretation. Science & Justice, 59, 558–564.
https://doi.org/10.1016/j.scijus.2019.06.005

Grant Application:

Improving Forensic Interpretation of Cell-Site Data (IFICD)

Team members:

TBA.

Abstract:

Forensic cell-site analysis addresses questions related to the location of mobile telecommunications devices. It is used to investigate and prosecute various types of crime, including organized crime and terrorism, and can be used in searches for missing persons. Call-data records (CDRs) indicate which cell ID a device connected to. A cell ID is associated with an antenna at a particular location pointing in a particular direction. Devices over a range of locations can contact to the same cell ID, and devices at the same location can connect to different cell IDs. In widespread current practice forensic practitioners present CDRs and results of radio frequency surveys as facts, and leave interpretation to police investigators or legal-decision makers. This can be misleading and prejudicial: whether, during a survey, a connection is made to a particular cell ID from a particular location is probabilistic, not deterministic. We will develop interpretive methods and TRL-7 tools that, using statistical models, address the following use cases:

1. At the time of a crime, a device connected to a particular cell ID. What is the probability of connecting to that cell ID if the device had been at the crime scene versus if it had been at an alibi location?

2. Given a cell ID, where is a device most likely to be?

3. Given CDRs from two different devices covering the same relatively-long period of time, what is the probability of obtaining the two series of cell IDs if the devices had been co-located during that time period versus if they had not been co-located?

2 is investigative only. 1 and 3 can be investigative or evidential, and our methods will output likelihood ratios as quantifiers of strength of evidence. We will also develop training for forensic practitioners on application of the likelihood-ratio framework to interpretation of cell-site-analysis data, and will establish an international network for researchers and practitioners working in cell-site analysis.

Status:

Application in preparation. Submission deadline 2024-11-20.

Project:

Pilot Database of Cell-Site Survey Data

Team members:

Matt Tart, CCL Forensics (Project Lead)

Geoffrey Stewart Morrison, Aston University (Funding Aquisition)

Description:

In preparation for the proposed research, we have developed a data-collection methodology, and in summer 2021 conducted a pilot data collection.

For the pilot, data were collected in multiple types of area:

urban (high street);

suburban (residential streets);

business park; rural (farmland and hills);

intermediate and wider areas (including highway).

Our methodology involves traversing the same areas multiple times travelling in different directions at different times of day over the course of several days/weeks, and pausing for several minutes at specified locations. Data are automatically collected at intervals of a few seconds or less.

Four different types of data are simultaneously collected:

Logs from multiple SIM-based engineering handsets – connected mode (EHS-c). EHS are adapted telephones that log their activity, including details of outgoing and incoming calls and texts, data downloads, and cell used. These activities were repeated many times while traversing the survey areas and while paused at specified locations.

Actual CDRs related to the activity of the engineering handsets described in 1. For UK data collection, these are obtained from each of the four mobile network operators. This is possible because, while planning the pilot study, a relationship was established between researchers, practitioners, law-enforcement agencies, and the mobile-network operators.

Logs from multiple SIM-based engineering handsets – idle mode (EHS-i). These log the details of the cell that they would select if they were required to make a connection as in 1.

Software-controlled radio (SCR). This scans and logs the details of all cells it can find on all networks including received signal strengths (each EHS in 1 and 3 selects a single cell at a time on a single network).

EHS and SCR also log the date and time (calibrated against an external reference clock) and their GPS location (calibrated at known locations). Traffic and weather conditions are manually logged.

To our knowledge, no dataset of size and scope similar to that of the pilot dataset has previously been published.

Download:

The database will be made available to other researchers and practitioners after we have published our first paper using its data.

That paper will include a full description of the data-collection protocols and the resulting data.

Acknowledgements:

The collection of this database was supported by Research England’s Expanding Excellence in England Fund as part of funding for the Aston Institute for Forensic Linguistics 2019–2024.