Performance analysis of P2P networks with malicious nodes

Peer-to-Peer (P2P) networks is a distributed structure, which is widely used in file sharing, distributed computing, deep search engines. Each node plays the dual role of client and server, which makes the P2P network further exacerbates the insecurity of P2P systems. For the behavior of nodes in P2P systems, requesting nodes are abstracted into customers and service nodes are abstracted into servers. An M/M/ c queueing model is developed with dynamically changing servers in almost unobservable cases, strategies are introduced, including start-up period, working sleep of service nodes and negative customers, etc. Using the matrix-geometric solution and Gauss-Seidel iteration method, average waiting length, average sojourn time, total energy consumption and throughput are derived. Through numerical experiments, the process of P2P file sharing is simulated, the optimal value of social utility is obtained using Nash equilibrium and social optimal strategy, and a theoretical basis is provided for node scheduling in P2P systems.