Research of a real-time database intrusion detection method for smart grid

Real-time database management system (DBMS) is the one of core of the dispatching automation system; it plays an important role for the safe operation of the smart grid. With the increase of smart grid open, the data security in the real-time database, especially the integrity of the security issues has recently received considerable attention. When the database is attacked, it has to roll back the malicious transactions and the transactions that be infected to recovery. During recovery, database either stops (or greatly restricts) the database service, which causes unacceptable availability loss or denial-of-service for mission critical applications, or continues database service, which may cause serious damage spreading during recovery where many clean data items are accidentally corrupted by legitimate new transactions. To solve this problem, the improved data mining algorithm DMAR is put forward in real- time database for smart grid. The propagation path of the malicious transactions can be found from these attack leave a unique fingerprint after them, namely attack sequence. So the potentially affected transaction in the propagation path can be enforced to filter out during recovery. The recovery process of database can avoid the spread of malicious transaction. It could improve the survivability of the real-time database systems.