With the rapid development of positioning technologies, positioning sensors are widely used in smart phones, car navigation system and other mobile devices. These positioning systems collect data points at certain sampling rates and produce massive trajectories, which further bring the challenges of storage and transmission of the trajectory data. The trajectory compression technique reduces the waste of the network bandwidth and the storage space by removing the redundant trajectory points and preserving the key trajectory points. This paper summarizes the progresses of trajectory compression researches and proposes a road-network aware and error bounded online trajectory compression system, named ROADER. The system includes a distance-bounded Hidden Markov map matching algorithm and error-bounded efficient trajectory compression algorithm. Experiments based on real data sets show that the system is superior to similar systems in terms of compression ratio, error occurrence and running time.