Super-proximity routing in structured peer-to-peer overlay networks

Peer to Peer systems are emerging as one of the most popular Internet applications. Structured Peer to Peer overlay networks use identifier based routing algorithms to allow robustness, load balancing, and distributed lookup needed in this environment. However, identifier based routing that is independent of Internet topology tends to be of low efficiency. Aimed at improving the routing efficiency, the super proximity routing algorithms presented in this paper combine Internet topology and overlay routing table in choosing the next hop. Experimental results showed that the algorithms greatly improve the efficiency of Peer to Peer routing.

Accelerated transmission in peer-to-peer network

The design and evaluation of accelerated transmission （AT） systems in peer-to-peer networks for data transmission are introduced. Based on transfer control protocol （TCP） and peer-to-peer （P2P） substrate networks, AT can select peers of high performance quality, monitor the transfer status of each peer, dynamically adjust the transmission velocity and react to connection degradation with high accuracy and low overhead. The system performance is evaluated by simulations, and the interrelationship between network flow, bandwidth utilities and network throughput is analyzed. Owing to the collaborative operation of neighboring peers, AT accelerates the process of data transmission and the collective network performance is much more satisfactory.

Research on the trust evaluation model of the P2P system based on the probability statistics method

The trust evaluation model of the existing P2P system is facing the attack behavior of two kinds of strategy malicious nodes： the strategic deception and the dishonest recommendation, which seriously affect the accuracy and the effectiveness of the trust evaluation of the model calculation nodes. In view of the disadvantages of the existing models, the author puts forward a trust evaluation model based on the probabilistic method. The model uses the concepts of the subjective trust relationship in the human society, and on the basis of the direct experience and the feedback information, calculate the direct trust and the recommendation trust of the nodes by using the probability statistical method. And through the identification of the importance of the direct experience, distinguish the reliability of the feedback information and recommenders, and improve the effectiveness of the trust evaluation model. The simulation experiment shows that compared with the existing trust evaluation model, this model can effectively restrain the threat of the strategic deception and the dishonesty recommendation, especially the attack of the complex cooperative cheating on the system.

DEFEAT:A decentralized federated learning against gradient attacks

As one of the most promising machine learning frameworks emerging in recent years,Federated learning(FL)has received lots of attention.The main idea of centralized FL is to train a global model by aggregating local model parameters and maintain the private data of users locally.However,recent studies have shown that traditional centralized federated learning is vulnerable to various attacks,such as gradient attacks,where a malicious server collects local model gradients and uses them to recover the private data stored on the client.In this paper,we propose a decentralized federated learning against aTtacks(DEFEAT)framework and use it to defend the gradient attack.The decentralized structure adopted by this paper uses a peer-to-peer network to transmit,aggregate,and update local models.In DEFEAT,the participating clients only need to communicate with their single-hop neighbors to learn the global model,in which the model accuracy and communication cost during the training process of DEFEAT are well balanced.Through a series of experiments and detailed case studies on real datasets,we evaluate the excellent model performance of DEFEAT and the privacy preservation capability against gradient attacks.