logo SBA


Digital archive of theses discussed at the University of Pisa


Thesis etd-02242020-101601

Thesis type
Tesi di dottorato di ricerca
Thesis title
Deployment and Management of Fog Applications
Academic discipline
Course of study
tutor Prof. Brogi, Antonio
  • application deployment
  • application management
  • cloud computing
  • fog computing
  • Internet of Things
Graduation session start date
Fog computing will support new applications based on the Internet of Things (IoT) by enabling computation all through the IoT to Cloud continuum. Particularly, the Fog will make it possible to support applications with stringent (hardware, end-to-end latency, bandwidth, security, uptime) requirements by deploying application services wherever they are best-placed to properly fulfil all such requirements. Being Fog infrastructures highly dynamic and heterogeneous, they will likely be subject to failures, server workload variations, and changing network conditions, what makes way challenging to optimally deploy and manage application services on top of them.

The first objective of this thesis is to propose and prototype suitable (declarative) models and probabilistic predictive methodologies to support the QoS-, context-, cost- and security-aware deployment of multi-service IoT applications and VNF chains to Fog infrastructures.
The second objective of this thesis is to design and prototype simulation environments to support the design and assessment of correct and effective management policies for Fog applications, by predicting their performance with respect to application uptime, alerting, energy consumption, convergence speed, and robustness against failures and workload variations.
Finally, some preliminary efforts towards a lightweight monitoring tool for Fog infrastructures are also discussed. The proposed prototype monitoring tool collects all data needed by the predictive methodologies we propose for both deployment and management of Fog applications.

The main contributions of this thesis reside in the fact that the proposed models and predictive methodologies are all capable of capturing the intrinsic dynamicity of Fog infrastructures (e.g., failures, workload, churn) as well as the interactions between deployed services and IoT devices. Additionally, the work presented in this thesis is among the first to consider security and trust aspects to decide on the deployment of application services to Fog infrastructures, and to focus on the modelling and simulation of an industrial application management platform, viz. CISCO FogDirector.