Wendong Wang

A group based reputation system for p2p networks

In large scale P2P networks, it is less likely that repeat interactions will occur between same peers for the asymmetric interests between them. So it is difficult to establish the direct trust relationship between peers and the network is vulnerable to malicious peers. A group based reputation system GroupRep for P2P networks is proposed in this paper to solve this problem. In GroupRep, the trust relationship is classified into three tiers: the trust relationship between groups, between groups and peers, and between peers. A peer evaluates the credibility of a given peer by its local trust information or the reference from the group it belonging to. A filtering cluster algorithm is proposed to filter unfair ratings provided by malicious peers. As a result, good peers are distinguished from malicious ones. Choosing the download source based on the trust value of responders can make good peers happy with the high ratio of the success query and the high satisfaction level under malicious collusive attacks with front peers.

An adaptive group-based reputation system in peer-to-peer networks

As more and more P2P applications being popular in Internet, one of important problem to be solved is inspiring users to cooperate each other actively and honestly, the reputation mechanism which is a hot spot for P2P research has been proposed to conquer it. Because of the characters of virtuality and anonymous in the network, it is very easy for users with bad reputations to reenter the system with new identities to regain new reputations in the reputation systems. In order to get rid of the impact of whitewashers and improve the system performance and efficiency, we propose a new probability-based adaptive initial reputation mechanism. In this new mechanism, newcomers will be trusted based on system’s trust-probability which can be adjusted according to the actions of the newcomers. To avoid the system fluctuating for actions of a few whitewashers, we realize the new reputation mechanism in system with group-based architecture, which can localize the impact of whitewashers in their own groups. Both performance analysis and simulation show that this new adaptive reputation mechanism is more effective.

MDFM: multi-domain fault management for internet services

New requirements of service-oriented fault management are analyzed and a framework MDFM (Multi-Domain Fault Manager) is proposed in this paper to solve the service fault localization problem in multi-domain context. Different from current solutions, our approach decomposes SLS (Service Level Specification) based on network capability, and monitor service performance in each domain along the end-to-end path. As a result, MDFM can localize the approximate domain rapidly on which the root cause resides, therefore causative region is narrowed down and computation cost for fault analysis is reduced. Faults on both server and client sides are considered in MDFM. A prototype has been implemented to prove the feasibility and efficiency of our service fault management framework.

Internet service fault management using active probing in uncertain and noisy environment

The great challenges of Internet service fault management are uncertainty and noise. To address these challenges, we model the service scenario through a multi-layer management model, and propose an approach using active probing to detect and diagnose faults. This approach uses bipartite Bayesian network as dependency model, binary symmetric channel as noise model, and is composed of two phases: fault detection and fault diagnosis. In first phase, we propose a greedy approximation probe selection algorithm (GAPSA), which selects a minimal set of probes while remaining a high probability of fault detection. In second phase, we propose a fault diagnosis probe selection algorithm (FDPSA), which selects probes to obtain more system information based on symptoms observed in previous phase. Simulation results prove the validity and efficiency of our approach.

Fast probabilistic fault diagnosis for large scale system

The challenges of fast fault diagnosis for large scale service systems are analyzed in this paper. A multi-layer management model is proposed to model the service scenario, which builds bipartite Bayesian network to denote dependence relationships. An incremental fault belief assessment method is proposed to analyze symptoms and compute posterior fault probabilities in an event-driven manner. Based on the method, we propose a greedy fault diagnosis algorithm to produce a sub-optimal explanation. To reduce the complexity of fault selection, we transform the fault diagnosis problem of finding MPE into finding most likely assignment of each fault, and propose corresponding algorithm. Simulation results prove the validity and efficiency of our algorithms.

On studying P2P topology based on modified fuzzy adaptive resonance theory

Considering vast and miscellaneous contents in P2P system, intelligent P2P network topology is required to route queries to a relevant subset of peers. Based on the incremental clustering capability of Fuzzy Adaptive Resonance Theory (Fuzzy ART), this paper made use of the modified fuzzy ART to provide small-world P2P construction mechanism, which was not only to categorize peers so that all the peers in a cluster were semantically similar, but, more important, to construct the P2P topology into small-world network structure. In detail, the modified fuzzy ART net was used to cluster peer into one or more appropriate categories according to its data interest, and the reverse selection mechanism in modified fuzzy ART was provided to construct semantic long-range edges among clusters. Simulations demonstrated that P2P small-world network emerged, i. e., highly clustered networks with small diameter, and the information retrieval performance was significantly higher than random topology.

An distributed multi-domain oriented fault diagnosis algorithm

Todays distributed systems often involve multiple administrative domains, each with a different organization and management policy. The centralized fault diagnosis techniques are not applicable to these systems. In this paper, we propose a distributed dependency model for building the dependencies in these systems. Based on the dependency model, two algorithms are proposed, DDA (distributed diagnosis algorithm) and ADDA (advanced distributed diagnosis algorithm). ADDA improves DDA from three aspects: less communication cost, more accurate effect evaluation function and spurious symptom probability. Simulation results show that ADDA is efficient in multi-domain environment.

A reciprocal capacity based adaptive topology protocol for p2p networks

A Reciprocal Capacity based Adaptive Topology Protocol (RC-ATP) for P2P networks is proposed in this paper It is based on the rational belief that a peer is only willing to maintain connections with those which will benefit it in future Reciprocal capacity is defined based on peers capacity of providing services and of recommending service providers As a result, reciprocal peers connect each other adequately Therefore the resulting topologies are more efficient and resilient compared with Adaptive Peer-to-Peer Topologies (APT).

On studying partial coverage and spatial clustering based on jensen-shannon divergence in sensor networks

The idea of partial coverage is provided in this paper, which means that the distance among data trends gathered by neighbor sensors is so small that, in some period, we can cluster those sensors, and replace the cluster with certain sensor in this cluster to form the virtual sensor network topology. But adopting this approach, we need to solve two problems: 1) how to characterize the distance among data trends (rather than raw data) of different sensors; 2) based on the distance, how to form the cluster and use the virtual network to represent the whole sensor network within certain error range. For the first problem, the Jensen-Shannon Divergence (JSD) is used to characterize the distance among different distributions which represent the data trend of sensors. Then, based on JSD, a hierarchical clustering algorithm is provided to form the virtual sensor network topology. Finally, the performance of our approach is evaluated through simulation.

On modeling internet qos provisioning from economic models

The modeling of Internet quality of service (QoS) provisioning is a multidisciplinary research subject. From the viewpoint of game theory, we propose a model that combines QoS index with price factor. We use the MultiNomial Logit (MNL) to model the choice behaviors of users. Each service class is considered an independent competing entity, which aims at maximizing its own utility. Based on noncooperative game, we demonstrate the existence and uniqueness of equilibria between QoS levels and prices among various service classes, and demonstrate the properties of equilibria. We also verified these results via numerical analysis.