The use of cellphone records to place suspects at or near crime scenes is coming under attack in courts nationwide, challenging an established practice by federal and local law enforcement that has helped lead to thousands of convictions.
Cellphone records are often used as evidence, relied upon to trace which cell tower was used to make or receive a call and then determine a caller’s whereabouts. But experts say that using a single tower to precisely locate where someone was at the time of a crime has severe limitations. And while defense lawyers are gradually recognizing the problems with such evidence, the FBI continues to rely heavily on the data in its investigations. The agency wants to expand its full-time team of 32 agents dedicated to the analysis of cell-site data, and it has trained more than 5,000 state and local police investigators in the basic methodology.
The conflict between the growing awareness of cell records’ limitations and the FBI’s desire to expand their use is increasingly forcing the nation’s judges to parse the technical evidence and determine if it’s being used fairly. In recent federal cases in Portland, Ore., and Chicago, judges have ruled that the analysis of cellphone records was not scientifically valid or reliable in locating people, in part because investigators have overstated its accuracy. In the hundreds of trials where judges have allowed the evidence, some defense lawyers have persuaded juries to acquit defendants of kidnapping and murder.
Telecommunications and forensic experts point to the case of Lisa Marie Roberts, wrongly imprisoned for nearly 12 years after both Portland prosecutors and defense lawyers misunderstood cellphone evidence, as an example of how the methodology can be misused.
“They’re using this stuff to put people in jail,” said Michael Cherry, a former Bell Labs and NASA consultant whose Falls Church, Va., company is pushing law enforcement and the defense bar to understand cellphone technology. “Lisa Roberts went through a nightmare,” said Cherry, who assisted the defense in both the Roberts and Chicago cases. “Complicated telephone technology is frequently oversold and under-defended in the courtroom.”
“As well-intentioned and completely honest as some of the prosecution experts are, I don’t think they have that deep understanding of how the [phone] network systems operate,” said Edward J. Imwinkelried , a University of California at Davis law professor and widely acknowledged expert in the use of scientific evidence.
When both prosecution and defense have an equal understanding of cellphone technology — issues which are now much disputed in the courts — then “the end result will be a much better understanding of cell- tower technology and better use of cellphone evidence,” Imwinkelried said.
In 2012, federal and local law enforcement agencies made more than 1.1 million requests for the personal cellphone data of Americans for a variety of investigative reasons, according to an annual privacy-related survey by Sen. Edward J. Markey (D-Mass.). But that may become more difficult. This month, a federal appeals court ruled for the first time that law enforcement agencies must obtain a search warrant for cellphone location data.
Experts said it’s hard to know exactly how often these records are used to convict defendants because the information is not always used, and sometimes other evidence is involved.
“In every major case in Los Angeles, they use cell-tower information,” said Jennifer Friedman, chief of forensics for the Los Angeles County public defender’s office. “It’s like fingerprints, it’s that common.”
At the heart of the controversy over cellphone data is a debate about how cellphone calls are routed and the range of the cell towers with which the phones connect. The FBI and local police officials maintain that they can place a suspect in a particular area because a cellphone, when making or receiving a call, usually selects the closest tower with the strongest signal and that most towers have a range of no more than two miles.
Bureau officials are confident that their methodology for interpreting single-tower records is sound. “All we are representing is what the analysis shows,” said Special Agent Robert Bornstein, supervisor of the FBI’s Cellular Analysis and Survey Team. “We are not biased for the prosecution or against the prosecution. We’ve used this investigative technique to recover victims of crime, missing children, deceased individuals. It’s overwhelming that the analysis conducted by the team assists law enforcement throughout the country.”
The FBI says their experts can trace the whereabouts of a suspect by returning to the crime scenes and testing the strength and range of individual towers as they put together their cases. FBI Special Agent William Shute said agents drive around the area near a cell tower, “using the same equipment cellular providers use themselves,” to determine a tower’s range.
“Judges constantly rule it’s admissible,” he said, “but tell jurors it goes to the weight of the evidence. That’s when we explain that it is reliable. Because we do it every day.”
But numerous experts and telecommunications workers say the FBI analysis techniques are wrong: Cellphone signals do not always use the closest tower when in use but instead are routed by a computerized switching center to the tower that best serves the phone network based on a variety of factors. In addition, the range of cell towers varies greatly, and tower ranges overlap significantly, and the size and shape of a tower’s range shifts constantly, experts say.
“It’s not really junk science, it’s misinterpreted science,” said forensic expert Larry Daniel of Raleigh, N.C., who has consulted and testified for the prosecution and the defense in numerous cases, including a capital murder case in Fayetteville, N.C., where police claimed the cell-tower data showed a man was at the crime scene. “It is useful and can be used. But in the hands of a novice, this is dangerous science.”
Verizon, AT&T and Sprint all declined to answer questions about how their systems work or discuss what they share with law enforcement. Experts said the algorithms for routing calls are a trade secret that the wireless carriers are inclined to keep secret.
But telecommunication experts are increasingly testifying in court about how the systems actually work. For instance, in a 2012 murder case in California, AT&T radio frequency engineer Trin Lopez testified that cellphones first connect with the mobile switching center before they are routed to a cell site and that towers in the Los Angeles area have ranges of zero to 20 miles, depending on the wattage of the tower and aim of the antennas.
“It is not possible,” Daniel said, “for anyone to reliably determine the particular coverage area of a cell-tower antenna after the fact based solely on historical cell-tower location data or call-detail records.” He said weather, time of day, types of equipment and technology, and call traffic all affect an antenna’s range.
Jeff Fischbach, a forensic expert from Los Angeles who assisted the defense in the Roberts case, said, “There are so many different factors [involved] that two cellular devices stationed next to each other making phone calls at the same moment could still get different towers. . . . I’ve seen proof that two individuals, subscribed to the same cellular provider, standing next to each other — on surveillance — can still get different towers.”
“We have never seen courtroom evidence,” Cherry said, “that authenticates an antenna’s range.”
The use of historical cell-site locator data is different than real-time triangulation of three cell towers to locate a phone, or GPS technology using satellites. The accuracy of those technologies is not in dispute, but phone companies do not save GPS or triangulation data for an individual phone — so that information is not used as evidence.
In the summer of 2002, Lisa Roberts was thrown in jail in Portland and charged with murder. Her girlfriend had been found strangled and dumped in a park. Roberts strongly maintained she was innocent and demanded a trial. On the eve of her trial, though, Multnomah County prosecutors revealed they had new evidence: Roberts’s cellphone records, showing she used her phone near the park where the body was found.
Without examining the evidence, Roberts’s attorney advised her to plead guilty, saying the records could “pinpoint” her near the crime scene. Reluctantly, Roberts took a plea bargain for manslaughter and a 15-year sentence.
But years later, new lawyers and phone experts who examined the case saw that the prosecution’s cellphone evidence was thoroughly flawed — claiming that a tower with a 360-degree antenna and a range of more than 300 square miles placed Roberts at the scene.
In April, U.S. District Judge Malcolm F. Marsh threw out Roberts’s guilty plea, virtually unheard of in federal appeals, ruling that if Roberts’s original attorneys had presented expert testimony on “the reliability of cell-tower evidence to pinpoint a caller’s location, [it] likely would have changed the outcome of the trial.” Last month, Portland prosecutors decided not to retry her, and Roberts was released from prison after nearly 12 years behind bars.
Roberts clearly remembers the day in 2004 when the county prosecutor told her about the cell-tower evidence, and her attorney telling her, “You need to take the plea bargain.” Roberts said, “I just bawled. I didn’t do it.”
The lack of full understanding of cellphone technology has consequences every day, experts say, even in the private sector.
For instance, in a 2012 case in Massachusetts, a company that provided cellphones to a three-man crew wrongly believed, due to cell-tower data, that the crew had not been in its assigned area. The company fired them, said Nicole Decter, the attorney for the men. Nine months later, after the company was shown that it had erroneously interpreted the cell-tower data, the men were rehired, Cherry said.
Insurance companies use cell-tower data when investigating claims and are able to obtain private phone records because of a clause in most policies mandating that customers cooperate with their investigations. In a case in Washington state, an insurer was able to properly use cell-tower data to show that a client was not 200 miles away from an arson fire, as he claimed, said Tim Gosselin, attorney for the client.
But it’s in the criminal defense world where the misuse of the technology has concerned experts the most. In Chicago in 2012, U.S. District Judge Joan H. Lefkow ruled that an FBI agent could not testify to the specific whereabouts of an alleged kidnapper based on his cell records because the “theory has not been subject to scientific testing or formal peer review and has not been generally accepted in the scientific community . . . [the] chosen methodology has received no scrutiny outside the law enforcement community,” Lefkow wrote. The ruling was the first by any court in the country to find that the cell-data analysis did not meet the Supreme Court’s legal test of scientific validity.
A jury then found the defendant, Antonio Evans, not guilty.
In cases where judges have allowed the evidence, some juries have figured it out on their own. In Connecticut, prosecutors claimed that they could place an alleged kidnapper at the scene of the crime by his cellphone records, as well as testimony from his six co-defendants or informants. “Due to the junk science that the state introduced,” defense attorney Aaron Romano said, “it called into question the other witnesses against him.” His client was acquitted.
In a 2009 murder case in Ocala, Fla., prosecutors tried to use cellphone records to show the defendant was near the crime scene. “We were just able to take on the science behind it,” public defender Nicole Hardin said. “You could be standing on one street and be picked up by a tower not in that radius or footprint. If there’s too much interference or it’s overloaded, it’s relatively unreliable.”
The jury found her client not guilty.