Cyber-physical system dependability enhancement through data mining

The research presented in the current paper addresses the use of data mining techniques for enhancing the dependability characteristic in the case of cyber-physical systems. A cyber-physical system for environmental monitoring was considered as a case study. In this context, the main task of data mining is to predict the missing sensor values caused by hardware and software components malfunctions. Different versions of a data mining regression algorithm were tested on the case study system, based on a well known data mining methodology. The test results show that the algorithms taken into consideration can predict sensor data with satisfactory accuracy, leading to a decreased failure rate of the system.