Adaptive membership maintenance under the sybil attack coupled with churn in P2P networks

Churn refers the continuous process of node arrival and departure in distributed peer-to-peer network. Enormous nodes joining the system concurrently will results in a flash crowd scenario for special nodes in the system, and enormous nodes crashing in a moment might come out a broken network. Sybil attack refers to the attack against identify in which a malicious user obtains multiple fake identities and pretends to be distinct nodes around honest nodes in the system simultaneous. By controlling a large fraction of the system, or a large fraction of the local scope, the malicious user can take further boring actions like DDoS attacks, False Voting, Invalid DHT routing, identifying worm spam and so on. It is so difficult to predict or defense the sybil attacks with the nature of the open and anonymous in P2P networks. Even though we can identify some sybil nodes or edges, the malicious nodes can leave the network at the same time, result in stronger churn attack and is not forbidden, that case should be seriously noticed. In this paper, we present a novel protocol with positive communication policy among peers and adaptive neighbor maintenance scheme under sybil coupled with the churn, to try to keep the normal nodes around, to keep desirable properties such as a low network diameter and clustering. The protocol is resilient against the sybil and churn, and fit for information dissemination and file sharing in unstructured P2P systems especially.