Researchers in the computer science department at NC State undertook a study to investigate robocalls. The study is the first step in a broader effort to find tools to reduce robocalls. The premise for the study is that we can’t easily fix the robocalling problem without first understanding how robocallers work. The first findings of the team were published in August in a paper, Who’s Calling. Characterizing Robocalls through Audio and Metadata Analysis.

In order to study masses of robocalls, the researches worked with Bandwidth, Inc. to set up 66,606 telephone numbers. They monitored these lines for eleven months ending early in 2020. During that time, they received almost 1.5 million unsolicited calls.

The team’s next goal is to be able to identify the source of robocalls so that some of the bigger offenders can be shut down. The team says they have made progress towards that end and will be publishing more results as they reach conclusions. Ultimately, the team wants to develop tools to help combat robocalling.

The first round of research did reveal some facts about robocalling that contradicts commonly held beliefs about robocalling.

First, they found that the volumes of robocalls received were nearly identical month to month and were not increasing as the public believes – at least during the eleven-month period.

The team answered and recorded incoming calls and made an unexpected finding that 62% of robocalls included practically no audio. I’ve received many calls over the years that had silence at the other end and always wondered why such calls were made. It’s a mystery why somebody is taking the trouble of making huge volumes of calls that are delivering no message. Perhaps it’s from faulty audio technology at the robocaller’s end, but the percentage of empty calls is surprising.

The only other explanation I can think of is that somebody is somehow profiting from the call volumes of empty calls and that the calls are generated to pump telecom settlements. But the amount of compensation between carriers for calls has been so greatly reduced or eliminated that it’s hard to picture somebody making much money from even huge volumes of short calls.

They also found phones that answer robocalls did not see an increased volume of robocalls. The popular wisdom is that people shouldn’t answer calls from numbers they don’t recognize because that leads to even more robocalls. The team found that phones that always answer robocalls didn’t receive any more calls than phones that never answer.

The team did witness the event they labeled as a call storm. A few of their phones received huge volumes of calls for a day or two. They figure that a robocaller had mimicked the caller ID from one of these numbers and that the huge volume of calls came from people calling back to ask who had called them.

Finally, the most promising research is the ability to identify robocaller campaigns. These are events when a given robocaller is making a lot of calls in a short period of time. During the research, they identified 2,687 specific robocalling campaigns where the same robocaller was calling multiple numbers in the pool. The researchers believe that it is these campaigns that account for a large percentage of the robocalls being made. Their next research direction is looking for ways to identify a robocaller campaign early so that carriers can shut it down.

I was hoping that the research might explain why I get one robocall every month from the 202 area code. The recording is in Chinese and the best I can tell it comes from a telecom company. I’ve always figured this was Huawei telling me that they are no threat and that there is no 5G race.