This paper presents the architecture of the iTrust system together with algorithms for maintaining censorship resistance. In iTrust, metadata describing documents, and requests containing keywords, are distributed to randomly chosen nodes in the iTrust network. If a node receives a request containing keywords that match metadata it holds, it sends the URL of the matching document to the requesting node, which then retrieves the document from the source node. A novel detection algorithm estimates the proportion of operational nodes in the iTrust network, by comparing the empirical probabilities of the number of responses received for a node’s request with the analytical probabilities for a match, for various proportions of operational nodes. A novel defensive adaptation algorithm increases the number of nodes to which the requests are distributed, in order to maintain the same high probability of a match when some of the nodes are non-operational or malicious as when all of the nodes are operational. Extensive experimental evaluations demonstrate the effectiveness of the architecture and the algorithms for maintaining censorship resistance in the iTrust network.
- Information network
- Peer-to-peer network
- Performance evaluation
- Publication, search and retrieval
- Search and retrieval