Abstract: With the continuous expansion of the scope of traffic sensor networks, traffic sensor data becomes widely available and is continuously being produced. Traffic sensor data gathered by large amounts of sensors shows the massive, continuous, streaming and spatio-temporal characteristics compared with traditional traffic data. How to provide intergrated support for multi-source, massive and continuous traffic sensor data processing is becoming one key issue of the implementation of diversified traffic applications. However, due to the absence of support for spatio-temporal traffic sensor data, it is difficult to develop corresponding applications and optimize the data transfer among different nodes in currenent distributed computing platforms. In this paper, we propose a traffic domain-specific processing model based on spatio-temporal data object. The spatio-temporal data object is treated as the first-class object in the distributed processing model. According to the model, we implement an intergrated processing platform for traffic sensor data based on the share-nothing architecture of cloud computing, which is designed to combine spatio-temporal data partition, pipelined parallel processing and stream computing to support traffic sensor data processing in a scalable architecture with real-time guarantee. Applications of the platform in real project and experiments based on real traffice sensor data show that our platform excels in performance and extensibility compared with traditional traffic sensor data processing system.