Fluid modeling of pollution proliferation in P2P networks

摘要: P2P systems are highly vulnerable to pollution attacks in which attackers inject multiple versions of corrupted content into the system, is then further proliferated by unsuspecting users. However, our knowledge, there no closed-form solutions that describe this phenomenon, nor models how injection impacts a clients' ability receive valid copy. In paper we develop suite fluid model proliferation systems. These lead non-linear differential equations. We obtain for equations; remaining models, efficiently solve equations numerically. The capture variety user behaviors, including propensity popular versions, abandonment after repeated failure good version, freeloading, and local version blacklisting. Our analysis reveals intelligent strategies as well clients seeking recover non-polluted within large-scale networks.