Adaptive system for
crowding monitoring using user's devices fingerprinting

crowd SENSING solution

Overview of the crowd monitoring approach, technologies, and practical rationale for its use.

Rationale

Large crowds forming during temporary and often unpredictable events such as festivals, public celebrations, demonstrations, or touristic peaks pose increasing challenges for cities and public authorities. These situations can affect public safety, strain local services, disrupt mobility, and reduce the quality of experience for both residents and visitors. While understanding how crowds form and evolve is essential for informed decision-making, existing monitoring solutions are often intrusive, expensive, difficult to deploy, or unsuitable for short-term use. This project is driven by the need for a practical, privacy-respecting, and easily deployable crowd monitoring solution that can be used when and where it is needed, without requiring complex infrastructure or specialised technical teams.

In a nutshell

The system relies on small, portable sensors that anonymously estimate crowd levels by detecting wireless signals already emitted by people’s mobile devices, without collecting personal data or tracking individuals. Sensors can be quickly installed in different locations and adapt to the available connectivity, even in challenging environments. During deployment, calibration support mechanisms can be used to adjust the sensing configuration to the characteristics of each site, improving measurement reliability. The collected information is then presented in dashboards with real-time and historical views of crowding levels, supporting timely and informed decisions. The result is a plug-and-play crowd monitoring tool designed for real-world use by public authorities during temporary events.

Available Technologies

Public administrations are often confronted with an overwhelming range of technologies when considering digital monitoring solutions. Combining sensors, connectivity, data processing, and visualisation requires expertise across multiple domains, making it difficult to decide which technologies to use and how to configure them reliably. This complexity frequently becomes a barrier to adoption, even when the operational need, such as crowd monitoring during events, is clear.

Simplified Deployment

Deploying crowd monitoring solutions is often complex and requires technical expertise. MoniCrowd simplifies this process by enabling installations to be progressively adjusted to each location, using calibration and deployment feedback to improve reliability. This approach shifts deployment from a rigid, configuration-heavy process to a more flexible and iterative one, reducing setup effort across different environments.

Proven solution

The research team participated in the CONFRONT – Challenge ON wifi FRame fingerprinting for people cOunting aNd Tracking, an activity funded by MOST (Centro Nazionale per la Mobilità Sostenibile), Ministero dell'Università e della Ricerca, Italy. The challenge brought together seven participants from four countries (Germany, Italy, Portugal, and Slovenia) to evaluate fingerprinting algorithms on a common dataset. The solution developed by the Iscte team, which forms the basis of the algorithm currently running on the MoniCrowd edge sensors, achieved the highest score, with 295.92 out of 300.00.

Building on this result, the system has evolved significantly. A new labelled dataset was collected under controlled conditions using a custom-designed Faraday cage, enabling the refinement of fingerprinting techniques and the development of more robust device identification methods. In parallel, machine learning models were explored to complement the deterministic approach, showing improved accuracy in estimating the real number of devices, particularly in high-density scenarios.

These advances support a hybrid approach, where rule-based fingerprinting ensures reliable device differentiation, while data-driven models improve aggregate counting accuracy. Together with real-world deployments and edge-based processing, this establishes MoniCrowd as a mature, validated, and continuously evolving solution for privacy-preserving crowd monitoring.

Sensor Cases

The sensor enclosure was designed with practical deployment in mind, supporting straightforward installation in diverse real-world settings. Its compact and robust form factor facilitates placement in different environments while protecting the internal components. The unit operates on battery power and integrates GPS positioning, enabling flexible installation and precise location awareness without relying on fixed infrastructure.

This design reflects the project’s focus on delivering a solution that is not only functional, but also ready to be handled, installed, and reused in operational contexts.

Interact with the model (rotate, zoom, or click) to explore the sensor design

Page updated

Google Sites

Report abuse

Adaptive system forcrowding monitoring using user's devices fingerprinting