Over the last several years, monitoring high performance data stream sources has become very important in various vertical markets. For example, in the public safety sector, monitoring and automatically identifying individuals suspected of terrorist or criminal activity without physically interacting with them has become a crucial security function. In the healthcare industry, noninvasive mechanical home ventilation monitoring has allowed patients with chronic respiratory failure to be moved from the hospital to a home setting without jeopardizing quality of life. In order to improve the efficiency of large data stream processing in such applications, we contend that data stream management systems (DSMS) should be introduced into the monitoring infrastructure. We also argue that monitoring tasks should be performed by executing data stream queries defined in Continuous Query Language (CQL), which we have extended with: 1) new operators that allow creation of a sophisticated event-based alerting system through the definition of threshold schemes and threshold activity scheduling, and 2) multimedia support, which allows manipulation of continuous multimedia data streams using a similarity-based join operator which permits correlation of data arriving in multimedia streams with static content stored in a conventional multimedia database. We developed a prototype in order to assess these proposed concepts and verified the effectiveness of our framework in a lab environment.
Read full abstract