On 1/11/2001 I obtained the position of Assistant Professor (Researcher) in Computer Science from the faculty of “Scienze dealla Fromazione” (Arts and Teaching Faculty) of the University of Turin (Italy) after a public competition won in September 2001. In November 2005 the position has been confirmed, becoming a Senior Researcher.

On 1/3/2010 I moved to the Department of “Elettronica, Informazione e Bioingegneria” of the Politecnico of Milano as an Assistant Professor (Researcher).

On 13/07/2015 I become Associate Professor the Department of “Elettronica, Informazione e Bioingegneria” of the Politecnico of Milano.

The main topics of my ongoing research activity are:

Energy optimization of computing infrastructures: study of the energy consumption of large-scale datacenters. My goal is to analyze the power consumption of the components of a single node (i.e. CPU, storage, network) under different workloads and how they change due to the effects of:
– virtualization
– introduction in a cloud infrastructure.
In particular I focus on the definition of workload distribution optimization techniques and on the study of their performances using Multi-Class Queuing Networks and Petri Networks. Results on these topics have been accepted to international conferences that are not indexed yet (july 2019).

Performance analysis of advanced computing infrastructures including Big Data applications, cloud environments and multi-core systems: application of performance evaluation techniques, ranging from multi-class queuing networks to mean field analysis and study the performance of complex hexa-scale systems. After performing accurate measurement on real systems, they are used to validate the proposed models. In particular I focus on determining the optimal number of nodes in Big Data applications, defining intelligent load balancing strategies to optimize the utilization in IaaS cloud infrastructures, as well I study the efficiency of multi-core processors to execute multi-threaded workloads.

Markovian Agents and Mean Field Analysis: both Markovian Agents and mean field analysis are analytical solution techniques that can be used to study systems characterized by a number of entities that tends to the infinity. The former considers objects (agents) spread over a space that communicate using messages while the latter considers objects that interact in way dependent on the total state of the system. Both techniques share a common mathematical foundation and can be successfully employed to approximate systems composed by a large number of interacting objects (such us sensor networks) or for simulating the physical properties of the materials.

Tools for performance evaluation: I work on the definition of the SIMTHEsys framework for the rapid prototyping of multi-formalism models. This allows defining both the syntax and the semantics of elements that a modeler can use to describe a system. I also support the development of the Software Tool “DrawNET” – a framework to support the composition of performance evaluation models and formalisms. I worked in particular on a new object oriented methodology to develop models in multi-formalism, multi-solution environment. These methodologies have been supported by the development of a generic graphical user interface (GUI), an abstract data definition language (DDL) and a set of access libraries, all written in Java language and based on XML technology.

Analysis of Computer Networks: I focus on the study of mobile telephone networks, “Peer-to-Peer” (P2P) systems, sensors and ad-hoc networks. Several analytical techniques (mainly based on fluid and hybrid performance evaluation models) have been developed and applied to various kinds of communication networks. In particular they have been used to study mobile phone networks (GPRS), the TCP/IP protocol and web services. Peer-to-Peer systems have been studied for what concern the ability to locate a particular resource and the time required to download it. Most recent studies concern the development of spatial fluid models to optimize the energy consumption in sensor networks and the study of efficient broadcast protocols in vehicular networks.

Fluid and Hybrid Performance Evaluation Models: study of numerical and analytical techniques to evaluate the performance of complex systems, using performance models described by mixed continuous and discrete components. In particular I started by extending the formalism of the Fluid Stochastic Petri Nets introduced by Trivedi and Kulkarni, adding new features and studying new solution techniques to compute both transient and steady state performance indices. Other investigations considered the efficient solution of Hybrid Bounded Fluid Models of the first and second order.

Virtual and Augmented Reality: study of low-cost solutions to visualize and enhance vision of 3D generated images, with complex human interaction patterns. Applications have been investigated in advanced training, rescue operations, fashion industry and minimal invasive surgery.

Topics of my past research activity, that are currently only partially investigated, include:

Queuing network with blocking: study of solution technique for finite capacity queuing systems. In particular I developed special solution algorithms based on the translation of a queuing network with blocking into a Generalized Stochastic Petri Nets

E-learning: study and development of a platform to create virtual reality based, e-learning shared applications based on low-cost technologies. In particular, simple and common web technologies (like Javascript and PHP) have been integrated with Virtual Reality applications (such as Adobe Atmosphere in the beginning then Unreal Technology), to recreate a Virtual Reality environment in which implement cooperative exercise to support the other common learning activities. Particular interest has been focused in Edutainment and serious gaming.

Performance evaluation of Multimedia systems: applications to the study of multimedia production systems, focusing on cinema and television productions. Analysis of the integrated content development systems, taking into account economical factors and using classical performance evaluation techniques such as Petri Nets and probabilistic models

National and International cooperation

I have visited several universities and I have been involved in many international cooperation. The most important are: the Technical University of Budapest (Hungary), the Department of Robotics of the Technical University of Berlin (Germany), the University “Federico Secondo” in Naples (Italy), the Laboratory for Foundations of Computer Science of the University of Edinburgh, the Arizona State University (Tempe – Az, USA), and the University of Twente (Enschede, The Netherlands). I have also given a seminar on optimization of computing infrastructure at CERN (Geneve, Switzerland). In October 2017 and May 2018 I have visited the People Friendship University of Moscow, where I gave three courses: “Performance Modelling with Queuing Networks” (10 hours) and “Fundamental Laws of Dependability” (10 hours), “Modelling with Mean Field and Markovian Agents Techniques” (10 hours).