Statistics and Runtime Verification

The importance of correctness of systems is becoming more crucial as computers control more of \n \nour everyday activities. Various approaches have been \n \nadvocated and used for the verification of such correctness, with one of the more promising ones being \n \nruntime verification. One important issue in runtime \nverification is the logic used to specify properties, \nsince this influences both the overheads induced by \nthe monitors, and the applicability of the approach \nto a particular domain. In this paper we propose \ntechniques for the expression and runtime monitoring \nof statistical properties, enabling easier manipulation \nand expression of non-functional requirements. The \nlogic is developed as an extension of the existing \nruntime verification tool LARVA, and has been applied \nto an ftp server implementation, adding a new layer of \nprobabilistic intrusion detection and system profiling.