Thesis proposals

ANTLab has several thesis proposals for students of the study programs involved in our courses (including of course Telecommunication Engineering and Computer Science and Engineering). Here is a list of areas and topics available (note that the list may not be updated very frequently, so you better ask to ANTLab members for more detailed information).

6G mobile networks

Description:

While the fifth generation is undergoing it’s deployment phase, the scientific community starts to discuss what will come next, as new technologies and methodologies are being considered with the aim to empower the radio mobile network capabilities beyond what 5G promises. 

This context opens a wide spectrum of research activities, mostly applied to technologies that are considered to become relevant in the next generations of mobile wireless networks:

  • Millimiter-wave communication 
  • Integrated access and backhaul (in collaboration with Huawei) 
  • Full duplex mobile radio systems 
  • Satellite networks and space-earth integration
  • Wireless networks virtualization
  • Resource allocation strategies for network slicing 

Research methodologies:

Mathematical programming and optimization (cplex, heuristic algorithms, etc.), Game theory (analysis, algorithmic solution with Matlab), Discrete event simulation (based on Matlab)

Reference professors: 

Prof. Ilario Filippini, Prof. Antonio Capone

PhD students involved:

Eugenio Moro 

Network Functions and Edge Applications Decomposition and Offloading

Description:

Common approaches for “Network Functions” and “Serverless Computing” (i.e. script/function code execution in virtual environment triggered by events) usually work in centralized deployments (e.g., Amazon AWS Lambda, Apache OpenWhisk).

In contrast, Distributed Serverless Computing (DSC) is based on the principle that function execution happens in distributed datacenters, where code can be instantiated in machines located in different physical places (adding a network aspect).

The project aims at achieving:

  • An orchestration software to decide the placement of micro-functions into heterogeneous hardware platforms and the routing of related dataflows in order to support distributed serverless computing considering user mobility, traffic dynamics, and application requirements

  • An application for the ONOS (Open Network Operating System) controller that combines, consolidates and deploys multiple micro-NFs in the network and enforces data flow routing, thus enabling DSC

Research methodologies:

Network emulation (mininet, containernet, etc.), Prototyping (P4 language), Mathematical programming and optimization (cplex, heuristic algorithms, etc.)

Reference professors: 

Prof. Giacomo Verticale, Prof. Antonio Capone

PhD students involved:

Daniele Moro

Pub-Sub Internet of Things Applications

Description:

MQTT and COaP have become the de-facto standards for Internet of Things applications. Nowadays, they often rely on a single broker (or server) hosted on the cloud. This architecture has several drawbacks such as a single point of failure, scalability issues, high latencies, etc.. Moreover, it does not fit the up-coming 5G architecture where Multi-Acces Edge computing (MEC) units will enable (serverless) distributed computing at the edge of the network.

The project aims at:

  • Design and develop routing algorithms for distributed MQTT brokers, where brokers communicate with each other exploiting the edge computing.

  • Move CoAP from a client/server architecture to a publish-subscribe fashion for a distributed deployment.

  • Analyze IoT traffic from heterogeneous sensors predicting anomalies and network bottlenecks.

These activities can be seen as a practical matter (developing real proof of concepts) or theoretically (e.g. mathematical models). Proof of concepts can be also deployed on real hardware in the 5G network.

Research methodologies:

Network/OS emulation (docker, containernet), coding (Python, C, Java, etc.), IoT prototyping (TinyOS, Cooja, Arduino, etc.)

Reference professors: 

Prof. Alessandro Redondi, Prof. Matteo Cesana

PhD students involved:

Edoardo Longo

Artificial Intelligence applied to Traffic analysis

Description:

Network operators are interested in continuously monitoring the satisfaction (QoE) of their customers to minimise the churn rate, which is the percentage of customers that, due to an unsatisfactory service, stop their subscriptions and move to a different operator. In order to monitor the level of satisfaction of their customers, network operators often rely on surveys and questionnaires, whose collection process is often costly and cumbersome. Differently, it is way easier to collect objective measurements regarding network performance either on an aggregate level (i.e., at base station side) or on a user-level (in-device measurements).

The research project aims at:

  • Developing machine learning algorithms able to predict users satisfaction leveraging only network measurements and commercial data regarding users profile 

  • Understanding how an operator can leverage the information regarding the distribution of customers satisfaction to recognize (and detect) anomalies in the network unsatisfaction among users.

Activities could regard on the one hand the improvement of the machine learning model and on the other hand the definition and analysis of realistic users mobility and activity scenarios to investigate the correlation between users satisfaction and the presence of problems in the network.

Research methodologies:

Network KPI Analysis (Python, Matlab, etc.), Machine Learning (Python, Keras, Tensorflow, etc.), Mathematical programming and Optimization (AMPL, MATLAB, etc)

Reference professors: 

Prof. Alessandro Redondi, Prof. Matteo Cesana

PhD students involved:

Andrea Pimpinella

Artificial Intelligence in the Air

Description:

Allocation of hard network slices over the 5G and the future 6G wireless interfaces is a complex task, which requires to find a trade-off among multiple objectives with limited information and limited time to find a solution. New approaches based on Artificial Intelligence are emerging to address this problem.

The candidate will study the available options and design a suitable algorithm.

Research methodologies:

Mathematical modeling and simulation; machine learning

Reference professors: 

Prof. Giacomo Verticale

PhD students involved:

Marco Zambianco

Secure Virtual Network Functions

Description:

Enterprises that want to run in-network virtual appliances (a.k.a. enterprise NFV) have to trust that the operator network is not compromised. This may not be an option for enterprises that operate critical infrastructures or want to protect their industrial secrets.

The research goals are:

  • the development of a virtual switch that can interact with functions implemented in a Trusted Execution Environment (TEE) 

  • the development of a tool for automatic checking that a virtual switch configuration satisfies the security policies

Research methodologies:

Prototyping (P4, C++), performance modeling and measurement

Reference professors: 

Prof. Giacomo Verticale

Stratum on NetFPGA with P4->NetFPGA ToolChain

Description:

Stratum is an open-source silicon independent thin switch OS. It’s developed for white-box switches and exposes a set of interfaces including P4Runtime and OpenConfig to enable interchangeability of forwarding devices and programmability of forwarding behaviors. Currently, Stratum supports Barefoot Tofino, Broadcom Tomahawk, and bmv2 software switch. 

NetFPGA SUME is an FPGA-based network interface card with capabilities for 10 and 100 Gbps operation. NetFPGA can be programmed using P4 (a high-level language targeting networking devices) with the P4->NetFPGA toolchain, opening the possibility to control the NetFPGA card via P4Runtime.

The project aims at adding Stratum support for the NetFPGA platform enabling easy programmability of NetFPGA card and integration with SDN controllers like ONOS.

Research methodologies:

Network emulation (mininet, containernet, etc.), Prototyping (P4 language), Software Development (C, C++), RPC systems (gRPC, P4Runtime).

Reference professors: 

Prof. Giacomo Verticale, Prof. Antonio Capone

PhD students involved:

Daniele Moro