As more and more accurate pedestrian localization systems emerged, one of the next important issues is how to make them into real lives. Providing localization with mobile phones which are indispensable for urban pedestrians would naturally embed localization into their daily lives. For smart phones start to include GPS, WiFi, accelerometers and digital compass, we propose a hybrid system to leverage WiFi and the sensors to avoid GPS's urban canyon problem. The default low-end sensors on the smart phones are so noisy that the hybrid process is not intuitive. We adopt a confidence-based hybrid algorithm to solve the problem.