ABSTRACT
Digital contact tracing is being explored in all areas of the world now, even more so recently due to the deadly and widespread COVID-19 pandemic. Currently, new vaccinations are being developed and pushed out rapidly, but cases where infection is spreading are still occurring. Therefore, contact tracing is still a much-needed transmission prevention method in this critical period. Long afterward, contact tracing shall remain an important role in human society to prepare and combat the spread of even deadlier pandemics that may arise in the future.
We propose Pathsim, a highly efficient contact tracing method that utilizes path generation and analysis to determine user proximity based on the existing cellphone LTE log data. Pathsim builds on the foundations laid out in by Mukherjee et al. called Vecsim, which proposes a signal model and a method to compare two points and determine their proximity without knowing the actual user locations. While Vecsim was very effective, there were several shortcomings including comparing points receiving signal data from different cell-phone towers as well as locations where the signal was highly discontinuous. Pathsim aims at designing a better system that looks at path information that is automatically logged by a network carrier to build a more robust system, capable of real-world use. In addition, Pathsim preserves user privacy and requires very little infrastructure cost to implement. Our results show that Pathsim exhibits a 75% improvement over Vecsim, when it comes to identifying locations that are connected to different base stations and also improves the overall distance estimation accuracy over its predecessor.