EvidenceProf Blog

Editor: Colin Miller
Univ. of South Carolina School of Law

Sunday, August 2, 2015

Murder Case is First Knox County Case to Involve Drive Test Analysis of Cell Tower Pings

Here's an interesting story about a murder case out of Tennessee. According to the article,

In a legal first for Knox County, a judge will allow testimony on a controversial FBI cellular analysis method in the case of a man accused of killing his pregnant girlfriend.

32 year-old Norman Eugene Clark is charged with two counts of first-degree murder in connection with the death of Brittany Eldridge, 25, and the couple’s unborn son, Ezekiel. Back in February, the bond that Clark could post for pre-trial release was reduced from $1 million to $500,000.

Eldridge was 81/2 months pregnant with Ezekiel when in December 2011 she was found stabbed, beaten and strangled inside her home at the Cross Creek apartment complex off Western Avenue.

According to the article,

There were no witnesses to the crime or DNA evidence tying Clark to the slayings. The Knoxville Police Department enlisted the aid of the FBI's Cellular Analysis Survey Team, a relatively new task force that has made its mark with such cases as the Boston Marathon bombing.

t is controversial because of the assumptions the team relies upon and its mix of old cellphone data and current tower signal strength. FBI Agent Kevin Horan testified at a hearing Friday the analysis is based on the belief a cellular phone will, with rare exception, choose a tower with the strongest signal near where the phone is being used.

The team then takes information about the phone's use from the cellular service provider. Agents then conduct a "drive test" to determine the signal strength of the tower at issue at the time of the test, not the crime. With that method, the FBI can show a suspect's location within a two-mile radius.

Those who listen to the Undisclosed Podcast will recall that there are serious questions about drive tests and the "rare exception" and two mile radius claims. Nonetheless, Horan claims that

he was able to show Clark was not, as he claimed, at his home at the time of the slayings and instead was within the radius around Eldridge's apartment.

Meanwhile, defense counsel

has an expert of his own who is part of a team that travels the country testifying against the FBI's CAST agents and their findings.

Those experts contend cellphone signals do not always use the closest tower but instead are routed by a computerized switching center to the tower that best serves the phone network, and a tower's range varies widely.

It will be interesting to see how this case unfolds. In particular, I'm interested in the distance between Eldridge's apartment, Clark's home, and the relevant cell towers pinged. 

One last point: Tennessee is a Daubert jurisdiction.



| Permalink


How far was Lincoln park from where adnan and jay lived/hung out? If it was not close, it's just a coincidence that the tower at Lincoln park pinged?

Posted by: Robyn | Aug 2, 2015 10:01:28 AM

Adnan's house was about 4 miles from Leakin Park. It's unclear exactly where Jay was living, but at least one of those places was really close to Leakin Park.

Posted by: Colin Miller | Aug 2, 2015 10:33:00 AM

Without knowing more details I can pass a final judgement but color me skeptical.

"Kevin Horan testified at a hearing Friday the analysis is based on the belief a cellular phone will, with rare exception, choose a tower with the strongest signal near where the phone is being used."

I'm not sure I'd use the word "rare" to describe the exception. Cell phones are programmed to connect to the tower with the strongest signal but signal strength varies a great deal, especially at the limits of the cell phone signal. Ten miles is considered to the the LOS distance of a normal cell phone tower to get a ping. If the usage is within three miles of the tower than I'd agree that it would be rare to see it ping a weaker signal. Outside that three mile radius I would not call it rare.

"Agents then conduct a "drive test" to determine the signal strength of the tower at issue at the time of the test, not the crime. "

Correct. And that's the reason it should have not been allowed in, IMO. The problem is that there is simply no reason to assume that the conditions of the drive test are an accurate reflection of the conditions on the day the crime happened. It is the functional equivalent of taking the weather of a random day and then claiming that the weather on the day of the crime was just like that random day.

Posted by: Daniel | Aug 2, 2015 2:05:12 PM

Darn it.

The first sentence should be CANNOT

Posted by: Daniel | Aug 2, 2015 2:05:46 PM

Ye Old Leakin Park pings again. I believe all the calls made around this time were to Jay's friends. Adnan even claims that he had the phone at this time, which shows his honesty (or lack of any criminal guile). As far as Adnan knew, he did have the phone, but based on the facts of the call log, Jay was off calling his buddies unbeknownst to Adnan.

Posted by: anon | Aug 2, 2015 3:19:21 PM

Actually, there was a 6:59 call to Adnan's friend Yaser, then 2 calls to "Jenn's Pager". Based on one of the Undisclosed episodes, I don't believe that it really was Jenn's pager. I think it was likely to one of Adnan's friends, but the State wanted to put Jay and Adnan together at Leakin Park. I think Adnan was taking Jay to his grandmother's house. Leakin Park was between Adnan's house and Jay's grandmother's house. It makes sense the phone would ping the Leakin Park tower if he was driving to or from Jay's grandmother's. Then he likely headed to the mosque to bring his dad dinner, just like he said. He makes several calls in the 7:00 and 8:00 hours to his friends. If you follow Rabia's blog, she posted a video of a mosque she goes to during Ramadan this year. People are milling about, talking on cell phones, in and out of prayers. He could easily have been making phone calls between prayers.

Posted by: gc | Aug 2, 2015 9:33:58 PM


Wonder if you'd comment?

Posted by: Paul | Aug 3, 2015 6:06:26 AM

Paul: As a law professor, I know a lot less than the true experts. The law professor who knows the most about these issues is Ed Imwinkelried. When i contacted him about appearing on the podcast, he directed me to the two true experts on the subject. One of them was Mike Cherry. I trust what he has to say about cell tower pings generally, and I trust what AT&T said in 1999 about incoming calls not being reliable for determining location.

Posted by: Colin Miller | Aug 3, 2015 7:16:36 AM

I'll respond to the comment in the link posted by Paul.

The problem here is that while the author understands the technology he doesn't understand the law. Take his 80% figure. Does that 80% standard represent "proof beyond a reasonable doubt". Some people would say yes because in there mind "beyond a reasonable doubt" means 75%--there is a law professor who name I cannot remember who has written several article claiming that to be so. But if you look at the data from a scientific point of view 80% is garbage. For example, with the Higgs Boson the criteria for that was 99.9999999%.

So what does "proof beyond a reasonable doubt" mean in terms of statistics? 75%? 90%? 99%?

Now in some legal cases cell phone tower data might only be probative--it doesn't mean to much without other data. But in some cases guilt or innocence might turn on cell phone data like the case in the OP.

So as I said in my post in the other thread that's what makes these decisions difficult. The odds of cell phone data are a constantly shifting matrix. We can, as the author of the link Paul posted correctly notes, only talk about them probabilistically. How probable is probable enough? That's the key question that Daubert or Frye demands of the judge

Posted by: Daniel | Aug 3, 2015 9:35:09 AM

Random rant from someone with a science background (and quite evidently a non-lawyer!): I think a lesson can be learned from the polygraph arena – research suggests certain methods can be 80-90% accurate at identifying guilty people, but only 50-80% accurate at identifying innocent people – and the courts have rightly seen this as insufficient, hence polygraph evidence is usually deemed admissible in court http://www.bps.org.uk/sites/default/files/documents/polygraphic_deception_detection_-_a_review_of_the_current_scientific_status_and_fields_of_application.pdf.

I really don’t think 80% likelihood is anywhere near enough, and falls way outside of ‘reasonable doubt’ territory in my mind. And the second you start adding up probabilities like these, you soon end up with something even less reasonable-sounding (and don’t even get me started on base-rates). I feel there should be better guidelines on explaining information to juries on probabilistic evidence, as it can be really quite complicated, yet its meaning can be hugely important. And it seems this shouldn’t be down to how good/bad your lawyer is at picking holes in expert testimony. Any ‘expert’ witness worth their salt should be required to prove expertise in the probabilistic nature of the data, and be required to explain this to juries, regardless of whether the right question is asked. Otherwise it all seems farcical.

Colin: Are specific jury instructions required when they’re being asked to consider probabilistic evidence? Are there any requirements for ‘experts’ to understand the nature of the evidence they’re talking about?

Rant over. Apologies for boring anyone.

Posted by: Cupcake | Aug 3, 2015 11:08:51 AM

So Adnan was calling his typical high school friends when he had a chance in the evening. Meanwhile Jay is calling drug dealers and criminals- Phil, Patrick and Jen. Who is more likely dealing with a crime and cover up?

Posted by: anon | Aug 3, 2015 11:13:06 AM


"And the second you start adding up probabilities like these, you soon end up with something even less reasonable-sounding"

Yes that's the other meaningful problem. What happens when the jury starts to combine the statistics of radio frequency with the statistics of gun shot residue with the statistics of tire markings to make a determination of guilt?

One of the more interesting aspects of the law is that while in the sciences these different odds would be multiplicative in the law they are additive.

Posted by: Daniel | Aug 3, 2015 1:43:06 PM

Colin thanks for responding. I guess where I personally find probability helpful is that it gives a marker 80percent, 90percent to “reason." Reason and what is reasonable are so subjective it is a wonder why anyone, with a good lawyer ;-), ever is convicted. In the end we are all going to be left with the individual circumstances of this case and in each circumstance we have to ask ourselves does this circumstance point to Adnan’s likely guilt or his likely innocence. Of course one circumstance alone, like cell phone evidence, can be discredited, to what the degree arguments can be made. The fact is his cell phone did ping that tower twice, both could be phantom pings and indicate nothing or they could indicate that at the time of the pings the phone was in the general area of Leakin park on the day that Hae was murdered. So I as a jury member I would like to know what is the probability that the phone was near there, and what’s the probability that this evidence is meaningless. Especially since Adnan claims not to be anywhere near that area. Anyway love the discussion such fascinating food for thought.

Posted by: Paul Hoen | Aug 4, 2015 9:45:15 AM

It seems FBI Special Agent Horan was about as convincing to the jury as FBI Special Agent Chad Fitzgerald was to Judge Martin Welch. The jury found Clark not guilty by 11:1. Sadly for Clark, because there was not unanimity, a mistrial was declared and he is being retried. Smells a lot like a way round the Double Jeopardy laws.

Were you able to find out any more detail about Horan's testimony Professor?

Posted by: Anonymouse | Sep 18, 2016 6:52:24 AM

Post a comment