In this thesis, the possibility to use event correlation tech- niques in the C2 monitor systems is analyzed. C2 is a monitoring software built by Ericsson Broadcast & Media Services for their television broadcasting center in Stock- holm. C2 enables the operators in playout to easily moni- tor events and alarms generated by devices responsible for producing each channel.
Previously archived events that are saved in the database was analyzed to achieve an understanding regarding what events that can occur within the system. C2 utilizes a compression correlation technique for events to limit the number of records in the database. This means however that information regarding the time aspect for each event often is lost which made the problematic to draw any con- clusions from the data.
The most common correlations techniques today are compared and analyzed in comparison to the requirements set up for an implementation into C2. The dependency graph technique was chosen as the most promising tech- nique. A prototype with a depth first search and a limit on depth level was then tested. The results indicate that this technique was not efficient enough to correlate all events in real time. The main reason for this is because C2 is built on Node.js, which is an single threaded framework. When events can occur within a few milliseconds between each other, extensive computations would make it impossible to handle new events quickly enough.
After discussions with operators a none real-time cor- relation implementation was done. This makes it possible for them to quickly find devices with corresponding events that are masked out and hidden from the current view.