ANTLab Main Research Activities
The ANTLab research group consists of 2 Full Professors, 1 Associate Professors, and 3 Assistant Professors which work in close contact with about 10 PhD students, 3 Post Docs and 1 external collaborator. The group has more than 300 publications in the last 10 years in international conferences and journals and approximately 500 K€ of fund raising per year. Our research methodologies include:
- Traffic theory, queuing theory, stochastic models
- Simulation tools (system level, discrete events)
- Optimisation models and algorithms, Game theory
- Prototype implementations (Linux embedded devices, TinyOS devices)
- Machine-learning tools
The activities of the group can be classified in four areas:
- Internet of Things,
- Wireless Networks,
- Internet of Energy/Smart Grid, and
- Traffic Data Analysis.
Internet of Things
Wireless Sensor Networks and IoT: analysis, design and implementation of wireless sensor networks for monitoring applications. We consider a variety of low-power wireless technologies, both short-range (Zigbee, Bluetooth, Wi-Fi) and long-range (LoRa, Sigfox) and focus on all the layers of the networking stack: access to the wireless medium, intelligent routing solutions and transport protocols, user applications. The core applications we focus on are
- Indoor localization and tracking of objects and people
- Body sensor networks and monitoring of biomedical signals
- Visual surveillance with wireless cameras (object and face recognition, parking lot monitoring, etc…)
- Indoor and outdoor quality of air monitoring
Communications technologies and Protocols for Smart Metering (Electricity/Gas/Water) and Smart Grid. We work closely with Electricity Distribution System Operators and with the Italian Electricity, Gas, and Water Market Regulation Authority in assessing the communications technologies for wide scale smart metering (PLC, WMBUS, LPWANs, cellular-based technologies) and for implementing Smart Grid Services over the Medium Voltage network (automated remote tripping, selective tripping).
Intercell Interference Coordination in 4G and 5G networks We design and simulate of ABSF-like coordination algorithms to schedule base station activities under SINR constraints, considering both best-effort and guaranteed user traffic demands. We develop both optimal centralized solutions and quasi-optimal distributed algorithms with time convergence guarantees.
Mm-wave Access in 5G networks We address the cell discovery problem in mm-wave access networks by leveraging context information (like user location, location accuracy, past access attempts, etc.) in order to mitigate the effects of the directional transmission of synchronization signals. We face the case of both single and multiple mm-wave base stations involved in the search with techniques to limit the number of the beamforming attempts before the channel acquisition.
Internet of Energy/Smart Grid
Base station power profile and energy management strategies We profiled the energy consumption of several types of base station relying on network operator’s counters database. In addition, we have developed traffic analysis techniques to predict network performance. Relying on these energy profiles and performance predictions, we design dynamic energy saving algorithms that allows to switch off redundant network components with probabilistic bounds on the performance decrease. The algorithms allow to estimate energy cost savings as well.
Algorithms for Automated Demand Response and Demand-Side Management of electricity. We design protocols for automatic optimal scheduling of electric loads such as appliances, recharge of electric vehicles, HVACs, both at the house level and at the campus level, considering time-varying tariffs, demands from the DSO, and production forecast from renewable sources. We schedule charge and discharge of batteries to maximize self-consumption of production by renewable sources.
Traffic Data Analysis
Network Data Analysis: analysis of data collected from edge network devices (access points and routers) or captured through wireless sniffer. Using machine-learning inspired tools and methodologies, the main goal of this research line is to extract valuable information from the captured traffic traces. Example of such information are:
- User behavior characterization
- Density estimation and localization (how many people are present in an area and what are the most preferred places)
- Device characterization
- Traffic characterization and prediction for network optimization purposes
Evaluation of privacy-by-design data collection strategies for consumption measurement data. We use quantitative techniques to formally evaluate the privacy level provided by aggregation and anonymization techniques used for collecting and redistributing consumption measurement data to third-party entities.
Software Defined Networks
Evolution of the data plane: We work on the design and testing of new stateful abstractions of the data plane to support programming paradigm based on the offloading of packet processing functions in the network switches. We contribute to open source projects for soft-switches and controllers for defining stateful architecture.
Traffic Engineering applications: We develop applications for software defined networks targeting specific traffic engineering problems like failure detection and recovery, load balancing, monitoring, optimized and robust routing, traffic prediction, etc. We use optimization methodologies based on mathematical programming and heuristic algorithms and we test our application through prototype implementation.