The purpose of carpool mainly lies in the reduction of user overheads. It can also alleviate the traffic jam and air pollution. The technological progress brings the population of intelligent mobile devices. Accordingly, new carpool systems tend to provide real-time online services. Current systems and studies on dynamic carpooling rarely address the issue of real-time search for ridesharing partners and this thesis thus targets the passenger search. After the driver starts his/her journey, we dynamically search the passengers whose get-on and get-off spots the best fit the journey. Moreover, with a dynamic-share payment scheme, we design the fast filtering mechanism that aims at the promotion of driver incomes and continuously retrieves the passengers near the journey. Simulation results show that our method can effectively improve the driver incomes. Through our experiments, we also analyze the impacts of different strategies on occupancy rate and computation time.