Collaborative location privacy-aware forwarding for opportunistic mobile networks

• Main Author: Zakhary, Sameh Rasmy
• Published: University of Nottingham 2016
• Subjects: 004.6
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.718849

The worldwide deployment of mobile devices incorporated into our everyday activi­ties made available an abundance of private location information. This information has sparked many innovative applications, but raised challenging users’ location-privacy problems. This thesis is concerned with the problem of offering source fc-anonymity location-privacy when contacting a Location-based Service (LBS) using Opportunistic Networks (OppNets). We propose a novel, fully distributed, self organized and collab­orative fc-anonymity protocol (Location-Privacy-Aware Forwarding (LPAF) protocol) to protect users’ location information and offer better privacy while communicating with an untrusted LBS over OppNet. LPAF enables users to collaborate in building a social-based untraceable obfuscation path to communicate with the LBS. We utilize a lightweight multi-hop Markov-based stochastic model for location prediction to guide queries towards the LBS’s location as well as to reduce required resources in terms of re-transmission overheads. We develop a formal analytical model and present theoretical analysis and simulation of the proposed protocol performance. We perform extensive simulation over pseudo-realistic city-map using map-based mobility models, and using real-world data traces to compare LPAF to existing state-of-the-art and benchmark location-privacy protocols. We show that LPAF manages to perform better across three performance dimensions (he. quality of service -success ratio-, quality of anonymiza­tion -number of obfuscation hops- and energy efficiency -obfuscation re-transmission overhead-). LPAF achieves higher privacy levels and better success ratio and delay com­pared to other protocols while maintaining lower overheads. Simulation results show that LPAF outperforms other distributed protocols in terms of success ratio for pseudo realistic scenarios. We have conducted a more realistic evaluation over OppNets using two real-world data traces. Results show that LPAF can offer better location-privacy and higher success ratio compared to other protocols in scenarios with moderate social network size, but with a slight increase in delay.