Reputation Aware Obfuscation for Mobile Opportunistic Networks

Current anonymity techniques for mobile opportunistic networks typically use obfuscation algorithms to hide node´s identity behind other nodes. These algorithms are not well suited to sparse and disconnection prone networks with large number of malicious nodes and new opportunistic, adaptive. So, new, opportunistic, adaptive fully localized mechanisms are needed for improving user anonymity. This paper proposes reputation aware localized adaptive obfuscation for mobile opportunistic networks that comprises of two complementary techniques: opportunistic collaborative testing of nodes´ obfuscation behaviour (OCOT) and multidimensional adaptive anonymisation (AA). OCOT-AA is driven by both explicit and implicit reputation building, complex graph connectivity analytics and obfuscation history analyses. We show that OCOT-AA is very efficient in terms of achieving high levels of node identity obfuscation and managing low delays for answering queries between sources and destinations while enabling fast detection and avoidance of malicious nodes typically within the fraction of time within the experiment duration. We perform extensive experiments to compare OCOT-AA with several other competitive and benchmark protocols and show that it outperforms them across a range of metrics over a one month real-life GPS trace. To demonstrate our proposal more clearly, we propose new metrics that include best effort biggest length and diversity of the obfuscation paths, the actual percentage of truly anonymised sources´ IDs at the destinations and communication quality of service between source and destination.