In what is probably the first published and peer-reviewed research paper of its kind, the title of this blog is, essentially, what I and my colleagues have argued.
Digital Investigation Journal has published our research in its December 2011 edition under the title: “Historic cell site analysis – Overview of principles and survey methodologies”. In the paper, we make a number of scientifically-justified recommendations on how cell site analysis surveys should be carried out.
It is important to note, though, that there will never be a perfect survey technique; survey methods are constantly evolving as technology moves on, and our analysts are continually searching for the best way to gather the appropriate evidence. Not to mention the fact that it very much depends on the case in question and what is required from the survey.
We outlined a number of advantages and disadvantages of several cell site survey techniques and explained the problems that can be encountered when interpreting survey data. These problems can cause practitioners to draw inappropriate conclusions from their results and can end up causing arguments in court.
A five-minute static location sample would enable cells with a timing offset of less than five minutes to be selected over others. It is also a quick and efficient method for data collection and analysis.
As with spot samples, five-minute location samples showed significant variability in results with small changes of position and between equipment, but to a slightly lesser degree. Even with large amounts of data from a single location, it is still less likely that other legitimately-serving cells will be reselected.
Again, any piece of equipment may not monitor all “valid” cell IDs as serving, and conversely many neighbour cells were detected that never provided service. It would be inappropriate to exclude a cell from serving at a given location using this method because it had not been detected in the survey.
These showed the least variability in results between pieces of equipment.
This method optimises the chance of cell reselection, minimising the effects of restricted BA lists and non-dominance. It is not infallible, but there is a much clearer indication of which cells genuinely provide service at and in the immediate area of a location or property.
Area surveys provide a wider picture of the general network configuration around the area of the location, enabling comment as to possible network changes if further work is required at a later date.
It takes longer to generate data and there is more data produced, making it more complicated to analyse – and this adds time (and therefore cost) to the examination.
Large quantities of measurements are not routinely obtained at a single specific location (although this can be done) for cells with a timing offset to be selected.
More possible cells will be identified (some may be false positives for that specific location, especially if they are “small” cells at the edge of the sampled area). The experiments we conducted suggest that there would be an estimated increase of around 20 per cent because of this. However, they may become relevant to an investigation later because they are detected in the local area even if not at the specific target location.
It is rare that a specific point is highlighted as being where a call was believed to have been made from.
The advantage of this type of survey for a “normal” voice call or text is that the size of the area served by the cell can be demonstrated, which can be extremely useful in highlighting the limitations of the cell site evidence (if the cell provides service over a large and relevant area) or in emphasising its importance (e.g. if the service area is very small).
With hundreds of data points, absence of a specific cell from the serving and neighbour information can be a good indicator a given cell would not be expected to provide service at the location.
So what’s the best way?
There is no perfect way to conduct cell site analysis surveys, but we need to be sure we can rely on survey data to indicate whether a cell does, or – more importantly – does not, legitimately serve at a location. Different investigators will have their own way of interpreting call record data from phone companies, and we have successfully challenged some of these in court.
For example: if a suspect claims that he was in a particular area at a particular time, and was using his phone at the time, this can be verified by analysing the coverage area of the mast that he claims he was using, or visiting the location to see if it serves there.
However, by simply turning up at the scene and taking a “spot sample” as to whether the mast serves that point is not sufficient. We are aware of investigators who do this, but it is quite feasible that by carrying out a more comprehensive analysis including, for example, approaching the scene from different directions, a different picture could be revealed.
The implications of this could be far reaching in terms of the verdict delivered during a criminal court case – the difference between “guilty” and “not guilty”.
We reviewed a number of survey techniques to determine the most reliable method for collecting RF survey data for historic cell site cases. Results from experiments have demonstrated that area surveys around a location of interest provide the most consistent method for detecting serving cells at a location.
Area surveys were also more reliable for excluding cell IDs from a location, and for assessing possible network changes if further surveys take place later.
We hope that this paper will set the standard for surveys in cell site analysis, and ensure that the criminal justice system is built upon a sound base of scientific evidence. Only by keeping up with changes in technology, and reviewing what other cell site experts are doing, can practitioners ensure that they undertake the most appropriate analyses. There is enormous scope for more research, and we hope to see much more published for peer review.
Cell Site Analysis Expert