Abstract:
This research presents a road traffic estimation using position and velocity of an in-car mobile
phone. Positioning of a mobile phone is implemented by matching its signal strength pattern with
those of the neighbor points. The location of one mobile phone under experiment is interpolated
from the positions of the nearest neighbor points. Therefore, the positions and mobile phone signal
strength are collected as a database before implementing the localization process. The K-nearest
neighbor algorithm is applied to search for the nearest points in signal matching procedure. There
are two techniques for positioning a mobile phone: linear interpolation of position from signal
strength and centroid of signal strength. Since the size of our database is rather large, we
additionally present a noise reduction algorithm, which eliminates bad data and compacts the
database. For the localization algorithm, the fast nearest neighbor strategy is applied in the database
matching to reduce computation time. The system is tested on a real road to estimate its traffic
congestion. The combination of positioning algorithm, a noise reduction algorithm, and a fast
nearest neighbor application is effective to estimate the traffic congestion. According to the
estimated velocity of the route under test, the traffic status is interpreted with an average error of 11
km/hr.