Dependability is increasingly becoming a key aspect in the design of digital embedded systems, given their widespread adoption in industrial automation, transport, network infrastructures and home appliances. In particular, based on the specific mission- or safety-critical application scenario, embedded systems should be characterized by one or more of the dependability attributes, such as reliability, availability or safety. This goal is even more ambitious as the aggressive advances in technology scaling, higher frequencies and power densities have negatively affected the reliability of the components constituting such systems. In fact, the number of hard faults, as well as soft ones, is growing due mainly to the shrinking of components themselves, to the variations in the manufacturing process and to the exposition of devices to radiations and noise fluctuations.
In this scenario, the trend of building new complex systems by integrating low-cost, inherently unreliable Commercial Off-The-Shelf (COTS) components is one of today's challenges in the design, analysis and development of systems exhibiting a certain degree of dependability.
Therefore, when adopting the COTS-based design approach for the realization of modern and pervasive electronic systems, reliability has become one of the main optimization goals, together with performance and energy.
Nevertheless, in non-critical environments, reliability must be leveraged not to introduce too high costs, associated with not so stringent requirements; moreover, in many situations, the need for reliability may change during the activity depending on the specific working scenario.
For these reasons, we claim that there is a need for a new way to dynamically "tune" fault management properties based on the working scenario, thus finding a satisfying trade-off between benefits and costs at run-time.
My present research and professional interests are centered on the development of dependable systems by means of new software/hardware technologies for runtime adaptation to mitigate the effects of failures and lifetime improvement. This focus is extended to support the dynamic management of the tradeoff between dependability, performance and energy, to offer adaptive mechanisms suitable for a broad spectrum of computing scenarios. Many- and multi-core architectures are considered as reference platforms, as well as reconfigurable platforms (e.g., FPGAs) used to implement hardened systems.
Adaptiveness is also the key for another ongoing research within the dependability field, related to functional diagnosis. The interest, begun as an industrial collaboration with Cisco Systems, has flourished in a broader scenario with novel solutions based on the exploitation of data mining as a reasoning-based mechanism to incrementally and adaptively drive engineers during the diagnosis of faulty complex boards.
Master and PhD theses tackling the above-mentioned challenges are available; the following are examples of open topics:
More details on the research issues, goals and achieved results can be found at http://hermes.ws.dei.polimi.it/, hosting all information and bibliography.