Privacy Preserving Efficient Data Dissemination in Decentralized Online Social Networks

A growing concern for users of Online Social Networks (OSN) is the privacy and control of user data due to the client-server architecture of the popular ecosystems. In this dissertation, we discuss how to develop a decentralized OSN platform so that the user has direct control of his/her information. When designing a new OSN, it is critical to understand how users interact over the OSN platform and provide the users with similar functionality. To this end, we analyzed online activities of Facebook volunteers to obtain a better understanding of the interactions in the user's friend circle. We performed a feasibility study to assess whether storage clouds can provide speeds comparable to popular OSNs, and observed that storage clouds have better performance than OSNs with delivering larger data while has comparable performance with smaller objects. We then, propose a privacy preserving decentralized Personal Online Social Network (POSN). POSN removes the central authorities from the OSN as users share their content with intended peers through smart devices and cloud backend. POSN takes advantage of free storage clouds to efficiently distribute encrypted user data and provide timely access to the content. The proposed scheme ensures that data is accessible only to the desired friends and gives the user full control over their data access. Access control is directly managed by the user via the cloud API and removes centralized identity management that exists even in some of the decentralized architectures. Data processing is handled by smartphones, since they not only provide continuous Internet access but also possess computational power reaching similar levels of traditional computers.We developed access control mechanisms where user's identity is ensured through their email address or phone number and fine-grained access provided through a hierarchy of access tokens. We also analyzed content dissemination in decentralized OSNs and developed mechanisms that rely on friendship communities to aggregate information from and reduce overhead on users. Finally, we developed a low latency data dissemination mechanism through online users and friendship communities so that smaller objects such as comments are disseminated in a timely manner.

Level thesis:degree_level

Doctorate Degree

Year dc:date.issued

2017

Author dc:creator

• Erdin, Esra

Advisor dc:contributor.advisor

• Gunes, Mehmet H.

Committee members dc:contributor.committeemember

• Yuksel, Murat
• Nicolescu, Monica
• Varol, Yaakov
• Li, Ming
• Fadali, Sami

Statement dc:rights

• In Copyright(All Rights Reserved)

Handle dc:identifier.uri

http://hdl.handle.net/11714/2073

OAI identifier oai:identifier

oai:scholarwolf.unr.edu:11714/2073