Research On The Query Technique Of Data Stream Over DDS Distributed System

The application of distributed system is wide with the rapid development of the technology, such as military operation system, wireless sensor network and telecommunication system. In these systems, large amounts of data from different nodes need to be handled and the data reach the system in the form of stream. As the scale of systems increase, the systems need to handle continuously updated data. As DDS is able to provide low latency, high throughput and real time, it has been a standard of distributed real-time systems and widely used in aerospace, defense and other fields. In a specific system, stream data reached in a period of time need to be queried and processed. However, at present, DDS doesn't have corresponding standard to query processing received data. Meanwhile, DDS real-time requires that distributed system based on DDS need to handle continuously arrived data. As a result of the wide use of DDS, the research on DDS data stream has been an urgent need.Firstly, this paper is a research for DDS (data distribution service) model, data filtering method based on the publish/subscribe mechanism and data stream query processing method of related theory. Secondly, in order to reduce the volume of data on the network communication, increasing the network utilization, we introduce the interest technology into DDS system structure and put forward publish-subscribe mechanism based on interest levels. The layer of the interests make communication node get really wanted data. On this basis, in practical applications sometimes it needs to query processing flow data received by distributed communication nodes, and flow data need to be process within sliding window range, for example, it needs to compare a period of time of temperature and humidity changes in the sensor applications. According to this requirement, this paper puts forward a method based on QoS adaptive window join querying method. The method based on DDS QoS policies of data divides into sliding window and establish the index, so as to improve the efficiency of query processing and obtain approximate results. Finally, this paper gives the design of DDS data stream query module, achieving continuous query processing. We perform experiments of the network communication performance of filtering data in the data transmission process and execution efficiency of data flow join query. The methods proposed by this paper are verified the effectiveness.