When the phone vibrates and an unknown number appears, the caller tries to extract personal information or money. This scenario, which has become common, relies on a sophisticated criminal infrastructure that SIM card farm detection tools could finally dismantle.
SIM card farms tricking even operators
Unwanted calls do not all come from simple automated systems. According to Virginia Tech News, a large portion of these calls originate from SIM card farms, facilities where hundreds of real SIM cards are connected to specialized and inexpensive equipment. These devices make real phone calls, so the network sees them as legitimate.
The problem lies in the massive nature of these calls. Each number used is a real number associated with a real SIM card. As a result, operators struggle to distinguish fraudulent calls from ordinary ones. Furthermore, scammers quickly switch between hundreds of cards to dilute their activity and evade filters.
Setting up such a farm does not require expertise or a big budget. A specialized device called a SIM box costs a few hundred euros and can accommodate dozens of low-cost SIM cards. Anyone, in any country, can set one up. However, for every dismantled installation, several others appear, like a hydra.
In the United States, the problem is reaching staggering proportions. The U.S. Public Interest Research Group has documented 29.6 billion robocalls for the year 2025 alone. Traditional blocking tools, based on databases of reported numbers, are no longer sufficient against this industrialization of fraud.
A digital twin of the network to uncover SIM card farms
To overcome this obstacle, researchers at Virginia Tech are implementing an innovative idea. They are building a digital twin, a complete virtual replica of a real mobile network. In this simulated environment, they can replicate the behavior of SIM card farms without accessing sensitive data from real subscribers.
Yaling Yang, electrical and computer engineering professor at Virginia Tech, leads this project. Her team trains artificial intelligence algorithms to identify fraudulent patterns, such as synchronized call behaviors, unusual routings, or rapid switches between multiple SIM cards.
The value of the digital twin lies in its ability to bypass the main challenge in research. Telecommunications operators fiercely protect customer data and trade secrets. No company wants to share the internal workings of its network. The digital twin offers a realistic experimentation platform without compromising confidentiality.
This project is financially supported by the GSMA (Global System for Mobile Communications Association), representing over 750 mobile operators worldwide. The team also plans to share their framework with other researchers to expand detection solutions.
Initial results presented on the global stage
Yaling Yang and David Simpson, a retired rear admiral of the U.S. Navy, presented their initial results at the Mobile World Congress 2026 in Barcelona in March. Their discussion allowed participants to learn how digital twins model the operations of SIM card farms and assess fraud detection accuracy.
The project involves various departments at Virginia Tech. In addition to Yang and Simpson, the team includes Lingjia Liu, holder of the Andrew J. Young Endowed Chair in Electrical Engineering, and Yang “Cindy” Yi, a professor and co-director of the Multifunctional Integrated Circuits Center. This interdisciplinary collaboration blends telecommunications, cybersecurity, and data analytics.
Other industry players are also tackling the problem. AT&T is deploying autonomous AI agents that analyze network data in real-time to detect anomalies faster than traditional methods. However, currently available solutions for the general public remain limited, as blocking applications still rely on user reports and outdated databases.
Virginia Tech’s approach could revolutionize how operators combat phone fraud. If the digital twin delivers on its promises, it would provide a universal testing tool accessible to all researchers in the field. Telecommunications were not designed to face AI-enhanced adversary systems, and this gap will only worsen without a proper response.
