Cell Tower Data Analysis Is Junk Science? You'Re Doing It Wrong.
Many of the calls I get from attorneys are requests for the analysis of cellphone records to determine the location of a phone at some given time. The analysis can provide useful information, but poor application of standards is threatening to ruin a valuable type of evidence.
HOW IT WORKS
When done correctly, the location process work like this:
1. A cellular network is organized into “cells”. A cell is usually a pie-slice-shaped area, or “sector”, on one side of a tower site. A tower site typically has several cells, pointing in different directions.
2. Call records will usually give the tower location and general direction of the cell where a where a call started. They may also give the tower location and sector direction of the cell where a call ended. (This information may also be available for SMS and even for data sessions, depending on how promptly the request is made. In criminal cases, other types of data may also be available, like registration events, handovers, and timing advance, if they were requested ahead of time.)
3. When the logged event happened (the call started, for example), the phone must have been in the covered area of the given cell. This must be true because this is the definition of “coverage”.
4. We can determine the covered area of the given cell and put it on a map. How we do that depends on time and budget and required level of accuracy, but the short of it is that we will use some combination of field-collected measurements, radio propagation modeling tools, knowledge of cells in the area, and knowledge of standard practices in cellular network design for the particular technology in question. In other words, we will use the same methodologies that the cellular operator used for designing and validating the cell.
5. We mark out the coverage on a map. That is the typical serving area of the cell.
In other words, we use the same principles and methods that cellular operators use to design their networks and verify their coverage, with known levels of accuracy.
So, what can we really say about handset location without making unsupported assumptions and straying into junk science?
* We can say that the phone was likely in the mapped coverage area. This is the intended service area of the cell, by design. But for any given call by any given phone on any given day, the actual coverage area may be bigger or smaller, depending on a lot of factors, some known and some unknown.
* We can say that the likelihood of the phone being somewhere else gets smaller and smaller as you move away from the coverage area. How much smaller how fast? That depends on the technology type, the climate, and the terrain, but it can be calculated with industry-standard methods.
* We can say that the phone was very unlikely to be in areas where other cells on the same frequency have stronger coverage, because in these areas, the interference from the stronger cell makes the serving cell unusable.
Overall, the best way to think of the location estimate for a given phone call in a given cell is like a “heat map”, not a clearly-bounded box. Just how small and tight is that heat map? It depends on a lot of factors. You need to make the analysis to know.
That’s why you need an expert.
HOW IT DOESN’T WORK
I recently saw an article challenging cell phone tower evidence in general, claiming that it should not be allowed in courts, and calling it “junk Science”. What the author describes, though, is not so much a failure of experts to provide veracious results as a failure of courts to enforce the conditions required by the Daubert standard. This blog is usually about telecommunications, not legal principles, but this particular article struck me as important, especially being as some attorneys I have talked to also read it, so I am going to respond. I will preface this by saying that I am an engineer, not a lawyer, but this is what I understand from what I read.
The Critics Have a Point
I will start by saying that the article made an important point: The closest tower is not necessarily the one serving the handset. And you will notice that I never make that assumption in the procedure that I gave. There is no basis for such an assumption.
Cells always have overlapping coverage. The overlap is intentional and controlled, usually following the best practices for the cell’s technology type. And cellular networks are often designed in layers, with “umbrella” cells filling in the coverage gaps of standard cells, and smaller cells adding capacity in more crowded areas. In urban areas and complex terrain, hills and buildings cast shadows that can obscure or weaken the signal from a nearby cell. Along roads, operators often deploy “ribbon cells” with long, narrow shapes. In dense city centers, cells might even be stacked vertically to cover different levels of buildings. Anyone who has ever understood (or made) a cellular network design understands these things. And anyone who understands these things knows that the closest cell is not always the one that gets selected for service.
Change your brand of expert
I am astonished and concerned that someone might put forward a simplistic “nearest cell” analysis and still be considered an expert. This seems like a failure in the application of the Daubert standard, substituting courtroom experience for engineering experience. Testifying in cases may make you an expert at giving testimony, but does little for your understanding of radio resource management and microwave physics. Someone who has not spent significant time studying what the industry calls “RAN (radio access network) planning and optimization”, both the practice and the underlying principles, has no business offering opinions on it. They do not meet the qualifications of the Daubert standard for cell tower evidence, regardless of how many times they have been allowed to testify about it in the past.
There is solid science if you demand it
In the first section of this article, I specifically called for the same methodologies that cellular operators use when designing and verifying their networks. These “RAN planning and optimization” methodologies are based on decades of refinement, analysis, and testing by thousands of engineers with a high motivation to produce accurate results. They form the globally-accepted foundation of project planning and performance optimization in one of the world’s largest infrastructure industries. This should satisfy “reliable principles and methods” qualification of Rule 702, or any other formulation of the Daubert standard.
The referenced article paints a picture of a system where courts are incapable of identifying true experts and valid methodologies, along with an industry of professional testifiers who are accepted as experts simply because they have been accepted as experts before. This calls into question all expert witness work, not just cell tower data analysis, because any type of evidence can be degraded in this way.
Cell tower data analysis, if done properly, is based on the same scientifically grounded and commercially verified methodologies that are used to build and manage the world’s cellular networks. If done improperly, it is “junk science”, like any other improper technical analysis. The problem is not that this type of evidence is inherently unreliable. The problem is allowing unqualified people to pose a experts. The courts will get the quality that they demand, and quality is out there if they demand it.
David Burgess has worked in telecom since 1998, in both military signals intelligence and in commercial cellular equipment. He also does expert work in legal cases.
Copyright David Allen Burgess
Disclaimer: While every effort has been made to ensure the accuracy of this publication, it is not intended to provide legal advice as individual situations will differ and should be discussed with an expert and/or lawyer.For specific technical or legal advice on the information provided and related topics, please contact the author.