VTrack is a system for estimating travel time using sensor data from mobile phones, addressing energy consumption and sensor unreliability. The system uses a hidden Markov model (HMM) to match position samples to road segments and estimate travel times. It can use GPS or WiFi sensors, with WiFi being less energy-intensive but noisier. VTrack is designed to handle noisy and sparse data, accurately identifying delay-prone road segments and providing travel time estimates for route planning. The system was tested on real-world data from 25 cars, showing that it can tolerate significant noise and outages in location estimates. VTrack's travel time estimates are accurate enough for route planning, even with poor individual segment estimates. The system also demonstrates that using WiFi for localization can effectively detect hotspots, despite the limitations of WiFi data. VTrack's HMM-based approach is robust to noise and provides accurate travel time estimates, making it suitable for real-time traffic monitoring and route planning. The system's performance is evaluated on a large dataset of GPS and WiFi location samples, showing that it can achieve good accuracy with both WiFi and sparsely sampled GPS data, saving energy in both cases. The results indicate that VTrack's travel time estimates are sufficient for route planning and hotspot detection, even with significant errors on individual segments. The system's performance is further validated through simulations, showing that it is robust to significant amounts of simulated Gaussian noise.VTrack is a system for estimating travel time using sensor data from mobile phones, addressing energy consumption and sensor unreliability. The system uses a hidden Markov model (HMM) to match position samples to road segments and estimate travel times. It can use GPS or WiFi sensors, with WiFi being less energy-intensive but noisier. VTrack is designed to handle noisy and sparse data, accurately identifying delay-prone road segments and providing travel time estimates for route planning. The system was tested on real-world data from 25 cars, showing that it can tolerate significant noise and outages in location estimates. VTrack's travel time estimates are accurate enough for route planning, even with poor individual segment estimates. The system also demonstrates that using WiFi for localization can effectively detect hotspots, despite the limitations of WiFi data. VTrack's HMM-based approach is robust to noise and provides accurate travel time estimates, making it suitable for real-time traffic monitoring and route planning. The system's performance is evaluated on a large dataset of GPS and WiFi location samples, showing that it can achieve good accuracy with both WiFi and sparsely sampled GPS data, saving energy in both cases. The results indicate that VTrack's travel time estimates are sufficient for route planning and hotspot detection, even with significant errors on individual segments. The system's performance is further validated through simulations, showing that it is robust to significant amounts of simulated Gaussian noise.