The Internet has become along the years a pervasive network interconnecting billions of users and is now playing the role of collector for a multitude of tasks, ranging from professional activities to personal interactions. From a technical standpoint, novel architectures, e.g., cloud-based services and content delivery networks, innovative devices, e.g., smartphones and connected wearables, and security threats, e.g., DDoS attacks, are posing new challenges in understanding network dynamics. In such complex scenario, network measurements play a central role to guide traffic management, improve network design, and evaluate application requirements. In addition, increasing importance is devoted to the quality of experience provided to final users, which requires thorough investigations on both the transport network and the design of Internet services. In this thesis, we stress the importance of users’ centrality by focusing on the traffic they exchange with the network. To do so, we design methodologies complementing passive and active measurements, as well as post-processing techniques belonging to the machine learning and statistics domains. Traffic exchanged by Internet users can be classified in three macro-groups: (i) Outbound, produced by users’ devices and pushed to the network; (ii) unsolicited, part of malicious attacks threatening users’ security; and (iii) inbound, directed to users’ devices and retrieved from remote servers. For each of the above categories, we address specific research topics consisting in the benchmarking of personal cloud storage services, the automatic identification of Internet threats, and the assessment of quality of experience in the Web domain, respectively. Results comprise several contributions in the scope of each research topic. In short, they shed light on (i) the interplay among design choices of cloud storage services, which severely impact the performance provided to end users; (ii) the feasibility of designing a general purpose classifier to detect malicious attacks, without chasing threat specificities; and (iii) the relevance of appropriate means to evaluate the perceived quality of Web pages delivery, strengthening the need of users’ feedbacks for a factual assessment.

Network Traffic Measurements, Applications to Internet Services and Security / Bocchi, Enrico. - (2017). [10.6092/polito/porto/2665103]

Network Traffic Measurements, Applications to Internet Services and Security

BOCCHI, ENRICO
2017

Abstract

The Internet has become along the years a pervasive network interconnecting billions of users and is now playing the role of collector for a multitude of tasks, ranging from professional activities to personal interactions. From a technical standpoint, novel architectures, e.g., cloud-based services and content delivery networks, innovative devices, e.g., smartphones and connected wearables, and security threats, e.g., DDoS attacks, are posing new challenges in understanding network dynamics. In such complex scenario, network measurements play a central role to guide traffic management, improve network design, and evaluate application requirements. In addition, increasing importance is devoted to the quality of experience provided to final users, which requires thorough investigations on both the transport network and the design of Internet services. In this thesis, we stress the importance of users’ centrality by focusing on the traffic they exchange with the network. To do so, we design methodologies complementing passive and active measurements, as well as post-processing techniques belonging to the machine learning and statistics domains. Traffic exchanged by Internet users can be classified in three macro-groups: (i) Outbound, produced by users’ devices and pushed to the network; (ii) unsolicited, part of malicious attacks threatening users’ security; and (iii) inbound, directed to users’ devices and retrieved from remote servers. For each of the above categories, we address specific research topics consisting in the benchmarking of personal cloud storage services, the automatic identification of Internet threats, and the assessment of quality of experience in the Web domain, respectively. Results comprise several contributions in the scope of each research topic. In short, they shed light on (i) the interplay among design choices of cloud storage services, which severely impact the performance provided to end users; (ii) the feasibility of designing a general purpose classifier to detect malicious attacks, without chasing threat specificities; and (iii) the relevance of appropriate means to evaluate the perceived quality of Web pages delivery, strengthening the need of users’ feedbacks for a factual assessment.
2017
File in questo prodotto:
File Dimensione Formato  
thesis.pdf

accesso aperto

Tipologia: Tesi di dottorato
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 12.82 MB
Formato Adobe PDF
12.82 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2665103
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo