Using Wi-Fi infrastructure to predict locations and contacts
Inventors
Alemi, Farrokh • Wojtusiak, Janusz
Assignees
George Mason Research Foundation Inc Of Fairfax Virginia • George Mason University
Publication Number
US-11825377-B2
Publication Date
2023-11-21
Expiration Date
2041-10-06
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
Exemplary methods predict the locations of individuals in buildings over time based on Wi-Fi access point connectivity and other reference information. Predicted locations are used to predict risk of contact, for example of infected persons with others in public spaces.
Core Innovation
The invention provides methods for predicting the locations of individuals inside buildings over time by utilizing Wi-Fi access point connectivity data and additional reference information. The primary process involves extracting Wi-Fi log data that contains timestamps of mobile device associations and disassociations with access points. This location information is then processed to predict sets of likely rooms or areas where a person was present, assigning probabilities to each possible location based on AP coverage, occupancy, or movement paths.
These predicted locations are further used to infer whether multiple people were present in the same place at the same time, thereby identifying potential contact events (i.e., risk of physical contact or exposure). The system computes contact scores between individuals by combining their respective probabilities for being in the same location during overlapping time intervals, assisting contact tracing procedures such as those relevant to COVID-19 and similar contexts.
The problem addressed is the inefficiency and limitations of traditional contact tracing methods, which depend on labor-intensive interviews and are prone to errors from human observation and recall. Existing technologies such as Bluetooth-based contact tracing face deployment obstacles, require extra hardware, or pose privacy concerns. Additionally, simple Wi-Fi-based approaches fail to balance precision and recall in identifying true contact events due to overlapping AP coverage. The disclosed invention overcomes these deficiencies by integrating more sophisticated computational techniques, reducing deployment cost and complexity while maintaining user privacy and achieving improved accuracy in location and contact prediction.
Claims Coverage
There are three main independent claims in the patent, each introducing a distinct inventive feature related to methods, systems, and computer program products for tracing contact using Wi-Fi infrastructure.
Tracing contact using Wi-Fi connection logs and probabilistic location assignment
A method that: - Extracts Wi-Fi access point (AP) information for multiple mobile devices (each associated with a different person). The information includes the specific APs used, connection times, and disconnection times for each device. - Assigns, for each AP connection of a device, a non-zero probability to all locations from which the AP allows connection, and assigns zero probability to other locations. - Determines, for one or more time intervals for each individual, a contact score between two devices (the person of interest and other persons) using the assigned probabilities for being in the same location at the same time.
System for automated tracing of contact events with probabilistic location inference
A system comprising: - A computer readable storage medium with non-transitory computer-readable program instructions; - One or more processors that, upon executing the instructions, will: - Extract Wi-Fi AP information for a plurality of mobile devices, as described above (including specific APs, connection times, and disconnection times). - Assign non-zero probabilities to locations for each AP connection, with zero probability to all other locations. - For each time or interval for each person, determine a contact score between two individuals using these assigned probabilities for being in the same location at the same time.
Non-transitory computer program product enabling Wi-Fi-based contact prediction
A non-transitory computer program product comprising instructions that, when executed by one or more processors, cause the processors to perform: - Extracting Wi-Fi AP connection data for multiple mobile devices (as above); - Assigning non-zero probabilities to locations covered by the AP for each connection; - Determining contact scores between a target person and others, based on overlapping probabilities in locations and time.
The inventive features are centered on using routinely collected Wi-Fi network connection logs to probabilistically infer individual locations and contacts without requiring new hardware or specialized mobile device applications. The methods, systems, and software are designed for efficient and privacy-conscious contact tracing by maximizing precision in exposure identification while maintaining high recall.
Stated Advantages
The location prediction techniques are superior to prior approaches such as simple AP intersection methods by maximizing precision and maintaining high recall.
The invention does not require modification of existing Wi-Fi networks, specialized smartphone applications, or new hardware deployment, thus minimizing cost and enabling rapid implementation.
The system significantly reduces or eliminates sources of human error common in manual contact tracing by using computational prediction instead of relying on human observation and recall.
The process maintains user privacy by not relying on RSSI or sensitive data, addressing privacy concerns present in other systems.
Enables timely and informed follow-up for institutions conducting contact tracing using accessible, real-time network data.
Documented Applications
Institutions such as universities, schools, corporate offices, businesses, and government offices can use the system for efficient and precise contact tracing in response to public health challenges, especially pandemics such as COVID-19.
Law enforcement or government agencies may use the location and/or contact prediction approaches to identify and monitor public safety threats, criminal activity, collusion, or conspiracy.
Interested in licensing this patent?