WP3 "Adaptive reconfiguration of heterogeneous distributed applications"
T3.1 aims at designing and developing intelligent mechanisms for privacy and security meant to operate within the proposed GAUChO architecture. Here, the attacker role has to be intended as misbehaving system (i.e., one not completely compliant with the local rules) instead of a malicious node. As a consequence, T4.1 has the goal of proposing novel techniques for devices authentication, data encryption and devices identity management. In particular, T4.1 aims at defining an organic and homogeneous privacy and security framework supporting the GAUChO platform, and investigating about the possibility of resorting to current standards/solutions or developing and testing new ones whereas necessary.
T3.2 focuses on the design of intelligent model-free mechanisms for change detection. Such mechanisms are crucial in the integrated FC/FN GAUChO architecture since they enforce the analysis of the data-quality of incoming datastreams so as to detect both changes in the environment in which end devices operate (and activate subsequent reactions) and the (possible) occurrence of faults affecting the end devices –perceived as changes- that might cause erroneous decisions. In particular, the research activity on intelligent model-free mechanisms for change detection aims at: (i) designing energy-efficient an multivariate change detection tests able to exploit the FC paradigm in its location-aware context, possibly with mobile devices; (ii) identifying the minimum set of relevant features extracted from acquired datastreams for change detection purposes at the FC units level; (iii) designing distributed change-detection tests able to take advantage of relationships existing among the datastreams acquired by end devices composing a FC cluster; (iv) designing of energy-efficient reconfiguration mechanisms able to exploit both information about the change (e.g., time instant the change occurred and its type) and location-aware information to support the self-adaptation/reconfiguration of GAUChO platforms over time. All these aspects will be evaluated on the application scenarios described in WP4;
T3.3 aims at designing and developing of a model-free fault diagnosis system for the GAUChO architecture able to operate without requiring any a-priori knowledge about the monitored phenomenon. This aspect is crucial in highly complex and dynamic application scenarios like the ones considered in this project (see WP.4) where faults/malfunctioning must be promptly detected. In particular, this task will focus on: (i) the design of energy-efficient fault detection and diagnosis mechanisms able to jointly operate both on the end devices and the FC/FN units of GAUChO. Such mechanisms will rely on and extend the change detection methods proposed in task (i) of T3.2; (ii) study novel location-aware mechanisms able to isolate the end device/FC clusters affected by the fault by analyzing both temporal and spatial relationships existing among acquired datastreams; (iii) design model-free mechanisms able to distinguish between time-variance and faults affecting sensor/actuators in end devices. Here, solutions able to deal with multiple faults and characterized by evolving fault-dictionary learning have to be considered. In addition, once a fault has been isolated in a specific sensor/end device, the fault identification phase yielding type, magnitude and temporal evolution of the fault must be characterized; (iv) study of intelligent mitigation mechanisms following a detected fault (e.g., by activating location-aware reconstruction mechanisms or virtual sensors) to maintain, whenever possible, the requested QoS of the application. All methodologies and methods provided in this task will be assessed, validated and ported on the application scenarios described in WP4.