What is REDS?
The “Real time Events and Decision Support” Laboratory is hosted by the Department of Informatics, Systems and Communication (DISCo) of the University of Milano-Bicocca, building U14, room T1009, ground floor.
The lab is dedicated to the research in the field of software systems, aimed at understanding and controlling the behavior of complex environments evolving in real time. We explore the technologies needed to deal with the massive streams of events generated by such environments.
This kind of problems arises in a wide variety of application areas, which range from innovative recommendations systems to the integration of physical and logical security, from urban traffic management to the realization of smart environments providing advanced caring services. The importance of these scenarios is growing at a very fast pace, thanks to the diffusion of technologies enabling the collection of event streams from people (wearables) and objects (IoT sensors).
The laboratory also includes a soul devoted to communication networks, which must evolve towards more flexible solutions in support to the collection of massive data coming from IoT sensors.
The technical solutions we are working on include, among others:
– Layered Knowledge Artifacts for domains and problems understanding, which integrate ontologies, probabilistic models and knowledge based systems;
– Bayesian networks and machine learning algorithms;
– Methods for event collection from wearables and other kinds of sensors;
– Special purpose (virtualized) network platforms for the collection of real time data, based on the SDN paradigm and on virtualization technologies;
If you are a student and you are interested on Bachelor and Master Theses, check the dedicated section of the website.
Research topics
We are involved in several research topics, among which:
Context-Aware
Data Analysis
The research on Context-Aware Data Analysis exploits the data coming from multiple sources of information in order to characterize the context where events occur. Among these sources we can quote ambient sensors, wearables, smartphones and videocameras. Some of the most important applications pertain to the realm of Smart Environments and eHealth.
Time-Dependent Decision
Support Systems
The research on Time-Dependent Decision Support Systems deals with those cases where the observed system and its reference environment change in time, passing through a series of macroscopic states, each one characterized by a specific information model. Moving from one state to another, the meaning and importance of some events can change drastically, therefore the applicable inferences and decisions must change accordingly.
Artificial Intelligence
for Society and Economy
AI can significantly support societal and economic development due to the rapidly emerging demand for innovative technologies in many application areas. The main aim of this research topic is to bring together heterogeneous competencies from different research fields, offering a wide picture of the best practices, of the challenges, and of the expertise that can be shared among researches and practitioners, exploiting case-based reasoning, ontologies, rule-based systems as conceptual paradigms and/or enabling technologies.
Communication Networks
Our research on communication networks deals with innovative approaches that massively rely on Virtualization, on Edge Computing and on Artificial Intelligence capabilities. We explore technologies and paradigms such as Edge Computing, Network Function Virtualization (NFV), Software-Defined Networking (SDN), SD-WAN, Serverless Computing, also in conjunction with Machine Learning, for an increased network flexibility and automation. We deal with scenarios related to the optimization of data flow management and security.
People
Dr. Fabio Sartori
Fabio Sartori is an Assistant Professor at the Department of Informatics, Systems and Communication of the University of Milano-Bicocca. He teaches “Algorithms and Programming” and “Laboratory of Mathematics and Informatics” courses at the Department of Mathematics and its Applications of the University of Milano-Bicocca.
His research activity mainly concerns knowledge-based systems and knowledge management. In these contexts, he has taken care of both methodological and computational aspects in the development of decision support systems, exploiting the Case-Based Reasoning (CBR) paradigm. Recently, he has applied the CBR paradigm in Artificial Intelligence for Society and Economy research field, taking part in the many research projects, funded from both public and private institutions, and studying how the CBR problem solving method can be used in forecasting and, possibly, avoiding, enterprise bankruptcy.
He is involved, as Programme Committee Member, in the organization of many International Conferences in the Knowledge Management and related research areas: in particular, he is one of the founders of the Metadata and Semantics Research Conference Series (MTSR) which aims to bring together scholars and practitioners that share a common interest in the interdisciplinary field of metadata, semantics, linked data and ontologies. He is Associate Editor of Program, an inter-disciplinary journal covering Computing and Information Science and also the Social Sciences in general and any other discipline that is concerned with digital data. Moreover, he serves as reviewer for many journals related to different topics of Computer Science.
He belongs to the Committee of International Experts of the doctoral programme “Knowledge and Information Engineering” of the Computer Science Department at the University of Alcalà (Spain).
Dr. Marco Savi
Marco Savi is a Tenure-Track Assistant Professor (RTD-B) at University of Milano-Bicocca (Milano, Italy).
He received my BSc (2010) and MSc (2012, summa cum laude) in Telecommunications Engineering (Networking area) both from Politecnico di Milano (Milano, Italy). His master thesis investigated users’ privacy concerns in a Smart Grid scenario.
In June 2016 he pursued a PhD degree in Information Technology at Politecnico di Milano. During his PhD studies he worked on Content Delivery and on the emerging Network Function Virtualization (NFV) paradigm.
From March 2016 to March 2020 he worked with Fondazione Bruno Kessler (Trento, Italy) initially as Postdoc Researcher, being promoted to Expert Researcher in 2018. He was involved on several research topics including Fog Computing, Network Monitoring, SDN Multi-layer Networks and NFV.
In April 2020 He joined the Department of Informatics, Systems and Communication (DISCo) of University of Milano-Bicocca as Non-Tenured Assistant Professor (RTD-A), becoming Tenure-Track Assistant Professor (RTD-B) in November 2022. He is currently working on various topics related to computer networking, including Serverless Computing applied to NFV, Programmable Data Planes for Network Monitoring and Machine Learning applied to Network Security. He also carries on teaching activities.
He has published more than 60 peer-reviewed papers in reputed international journals and conferences.
Prof. Riccardo Melen
Riccardo Melen was a Full Professor of Computer Science at Università di Milano Bicocca (formerly he was at Politecnico di Milano and Università dell’Insubria). His research interests include mobile applications and services, event management, security and authentication systems. He is the author of more than 60 papers published on international journals or conference proceedings.
In 2000, as a consultant for the Italian Regulation Authority for Telecommunications (AGCOM), he defined the technical specifications for the local loop unbundling regulation in the Italian network.
In 2003, as a consultant for the Ministry of Innovation, he has been responsible for the definition of the architecture of the new network for the interconnection of Italian Public Administrations (SPC).
In 2009 he was the co-founder of NFC Alias, a university spinoff company dedicated to the development of NFC applications.
He has been working as a consultant for various companies: among others for Telecom Italia, Fastweb, Cisco Systems and HP. He has also performed evaluations of start-up companies for venture capital funds and other investors.
In his previous experience in the industry, as a researcher and laboratory head at CSELT (the former research unit of Telecom Italia) he was involved in the design and performance evaluation of multiprocessors and was the main system designer of an experimental broadband packet switching systems realized in CSELT in the period 1987-90.
Jacopo Talpini
Jacopo Talpini is a PhD student in Computer Science at “Università degli Studi di Milano-Bicocca”.
He received his BSc in Physics from “Università degli Studi di Milano” in 2018. Then, he obtained a MSc in Astrophysics at “Università degli studi di Milano-Bicocca” with a grade of 110/110. His master thesis project was focused on applying deep learning methods to characterize transient noise in gravitational-wave detectors.
His research interests include machine learning and its applications to solve real-world problems, in particular related to cyber-security.
Emanuele Petriglia
Emanuele Petriglia is a PhD student in Computer Science at the University of Milano-Bicocca (Milan, Italy).
He received his BSc in Computer Science from the Sapienza University of Rome (Rome, Italy) in 2021 with a grade of 110/110 cum laude. Then he received the MSc in Computer Science from the University of Milano-Bicocca in 2024 with a grade of 110/110 cum laude. His Master’s thesis project deals with the use of multi-agent reinforcement learning for workload distribution in FaaS edge computing systems.
From 2023, he is working as an adjunct professor for the Distributed Systems course of the BSc Computer Science degree at the University of Milano-Bicocca, and from 2024 also as a laboratory tutor for the same course.
His research interests are in the area of edge computing combined with reinforcement learning.
Available Theses
We have a number of available theses for both Bachelor and Master students.
Check them out and contact us!
Dynamic Decision Networks for Modelling and Supporting Training Programs
Responsible: Fabio Sartori
Analysis of Data Collected from Wearables
Responsible: Fabio Sartori
Serverless Computing / Function-as-a-Service
Serverless Computing is an innovative event-driven cloud computing method based on the execution of ephemeral functions, often paired to the Function-as-a-Service model. A thesis in this area can pursue
one between the following two objectives:
1) Enhancing an existing platform for decentralized Function-as-a-Service at the edge (DFaaS);
2) Studying the applicability of the method for virtualising network functions, in the context of Network Functions Virtualization.
Responsible: Marco Savi
ML-based Network Intrusion Detection at the Edge
Data-driven models based on Machine Learning (ML) are promising to analyze and classify network traffic within Network Intrusion Detection Systems (NIDSs). A thesis on this topic aims at improving
ML-based network intrusion detection along one (or a few) of the following possible directions:
– Enhancing the trustworthiness of the ML model;
– Defining a method to label only relevant data (Active Learning);
– Re-training the ML model without need of storing old data (Continual Learning);
– Collaboratively training the ML model (Federated Learning or Peer-to-peer Learning);
– Unlearning obsolete attacks to keep the number of attack classes low.
Responsible: Marco Savi
Programmable Data Planes and P4
In Software-Defined Networking, programmable data planes enable switches to perform complex operations on packet headers and to collect advanced statistics on the network traffic, going beyond simple frame forwarding. One of the most adopted domain-specific programming languages in this context is the P4 language, which can be adopted to program the switches’ data plane behavior to support novel protocols and features. The objective of the thesis is to define and implement P4-based strategies for performing advanced network monitoring (e.g. by adopting sketches) or assisting ML inference.
Responsible: Marco Savi
Satellite Routing in Low-latency Satellite Networksd
The problem of data routing in satellite low Earth orbit (LEO) networks must be carefully taken, as satellite network topology varies fast and deterministically. The objective of the thesis is to study the problem of satellite
routing by means of Machine Learning techniques (e.g. Graph Neural Networks, Deep Reinforcement Learning) or heuristic algorithms. Two different approaches can be investigated:
1) Assuming that global routing decitions are taken in ground stations;
2) Assuming that local routing decisions are autonomously taken by satellites.
Responsible: Marco Savi
Service Provisioning in SD-WAN
SD-WAN is a very promising technology to establish reliable and cost-effective connectivity across the different sites of a company, in substitution of existing (and expensive) MPLS tunnels. Many SD-WAN products are
currently available on the market, but they rely on very simple strategies with respect to service creation and differentiation. The objective of the thesis is to improve the existing SD-WAN technology along one of the
following directions:
– Designing and implementing distributed learning techniques (e.g. multi-agent reinforcement learning) to understand how to distribute traffic on output interfaces;
– Using network tomography techniques to reconstruct the (unknown) topology of network operators.
Responsible: Marco Savi
Teaching
Here you can find the courses we teach.
Related to the area of Communication Networks: “Telecommunications Systems and Services”, Master Degree in Computer Science, 1st year (Teacher: Marco Savi)
Related to the area of Communication Networks: “Networks and Operating Systems”, Bachelor Degree in Computer Science, 2nd year (Teacher: Marco Savi)
Papers
Here you can find the list of our joint papers
J. Talpini, F. Sartori, M. Savi, “Enhancing Trustworthiness in ML-Based Network Intrusion Detection with Uncertainty Quantification”, to appear in Journal of Reliable Intelligent Environments, Aug. 2024
F. Sartori, M. Savi, G. Tarrini, J. Talpini, “Towards a Knowledge Diversity Notion to Identify Intrusions in Industrial Contexts”, in International Conference on Enabling Technologies: Infrastructure for Collaborative Enterprises (WETICE), Jun. 2024
J. Talpini, F. Sartori, M. Savi, “Hierarchical Multiclass Continual Learning for Network Intrusion Detection”, in IEEE Conference on Network Softwarization (NetSoft), Jun. 2024
N. Di Cicco, J. Talpini, M. Ibrahimi, M. Savi, M. Tornatore, “Uncertainty-Aware QoT Forecasting in Optical Networks with Bayesian Recurrent Neural Networks”, in IEEE International Conference on Communications (ICC), May 2023
J. Talpini, F. Sartori, M. Savi, “A Clustering Strategy for Enhanced FL-Based Intrusion Detection in IoT Networks”, to appear in International Conference on Agents and Artificial Intelligence (ICAART), Feb. 2023
F. Sartori, M. Savi, K. Shala, A. Moglia and J. Talpini, “Knowledge Artifacts to Support Dietary: the Diet Module of the PERCIVAL Project”, in International Conference on Metadata and Semantics Research (MTSR), Nov. 2022
Fabio Sartori, Marco Savi, Jacopo Talpini, “Tailoring mHealth Apps on Users to Support Behavior Change Interventions: Conceptual and Computational Considerations”, in Applied Sciences, vol. 12, no. 8, pp. 1-19, Apr. 2022
Fabio Sartori, Marco Savi, Riccardo Melen, “The Role of Data Storage in the Design of Wearable Expert Systems”, in International Conference on Metadata and Semantics Research (MTSR), Dec. 2020
Marco Savi, Fabio Sartori, Riccardo Melen, “Rethinking the Design of Wearable Expert Systems: The Role of Network Infrastructures”, in IEEE International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), Aug. 2020
Fabio Sartori, Riccardo Melen, “Wearable expert system development: definitions, models and challenges for the future”, Program 51(3), 235-258, 2017
Dario Baretta, Fabio Sartori, Andrea Greco, Riccardo Melen, Fabio Stella, Letizia Bollini, Marco D’addario, Patrizia Steca, “Wearable devices and AI techniques integration to promote physical
activity”, MobileHCI Adjunct 2016, pp.1105-1108, 2016
Riccardo Melen, Fabio Sartori, “Employing Knowledge Artifacts to Develop Time-depending Expert Systems”, Communications in Computer and Information Science, Springer-Verlag, 2015
Riccardo Melen, Fabio Sartori, “Time Evolving Expert Systems Design and Implementation: the KAFKA Approach”, KEOD 2015, Lisbona (Portugal), 2015
Riccardo Melen, Fabio Sartori, Luca Grazioli, “Modeling and Understanding Time-Evolving Scenarios”, Journal on Systemics, Cybernetics and Informatics, Vol. 13, No. 5, 2015
Contact Us
For further information you can contact us at the following email addresses.
Fabio Sartori: fabio[dot]sartori[at]unimib[dot]it
Marco Savi: marco[dot]savi[at]unimib[dot]it
Riccardo Melen: riccardo[dot]melen[at]unimib[dot]it