WP1. Process mining and behaviour profiles
Under WP1 the main goal was to study the issues of process mining, process discovery and process learning, with the outcome of designing algorithms for automatic, adaptive, computer assisted generation of models for information systems for workflow process management (business process management systems) and decision support systems. For validation of methods and algorithms in WP1 several reference processes both from public and private sector were used.
WP2. Energy Efficient Intrusion and Error Prediction
Behavior detection is challenging due to its energy requirements. In this work package, we implemented an Android application, “FunFi”, that allowed us to collect network and location information. Based on the collected information, we designed and analyzed an algorithm for user behavior prediction with the focus on user trajectory prediction. Deviation of a user from a predicted trajectory can be perceived as an error that needs to be detected in a timely and energy-efficient way. Unlike other prediction approaches, our approach is based on a one-time trajectory probability computation; these probabilities remain unchanged over extended time period, what simplifies implementation of this approach as a web service. Other than trajectory prediction, this approach also allows for mapping WiFi routers to logical locations. This allows us to acquire user location with respect to a WiFi router nearby.
WP3. Mobile Information Security and Privacy
Mobile devices are still gaining popularity and transforming the whole field of information and communication technology. With the positive effects of these changes come hand-in-hand the emerging new forms of cyber-crime and misuse of data. Sensitive data stored on mobile devices lacks adequate protection. Furthermore, mobile devices themselves generate sensitive data, such as location and its history. All such sensitive data may leak from the devices due to unintentional reasons, such as poor development practices, as well as due to intentional malicious activities – malware, targeted malicious attacks, tracking in advertising or marketing.
WP4. Biometric Recognition
The project dealt with the issue of creating a biometric recognition system and several areas of research were involved. It was necessary to implement a complex recognition system composed of the input image pre-processing module, feature extraction module, and classification module. The methods chosen for these modules included sophisticated machine learning methods. Parameters of each method were optimized for increasing cognitive success and ability to work in real time.
WP5. Monitoring of Bio-physical Parameters
Modern sensor networks based on the latest technology of wireless data transmission help to collect data from many sources and maintain human health. Sensor networks, with their wide range of options, provide a wide spectrum in the range of biometrics and health care, the environment, smart buildings, the automotive industry and other economically important sectors. In data collection great emphasis is placed on ensuring a clean and error-free input signal. Basic level data collection between sensory units of networks currently represents one of the key points of information transmission integrity. Dynamic advances in the area of sensor elements and systems are driven by new materials, powerful microprocessors, great progress in the area of organic materials and the use of known semiconductors for new applications, etc. The significance and security of collecting a large volume of data and achieving network