Fanyuan Ma

PWSD: A Scalable Web Service Discovery Architecture Based on Peer-to-Peer Overlay Network

The Web services are distributed across the Internet, but the existing Web service discovery is processed in a centralized approach such as UDDI, which has the limitations of single point failure and performance bottleneck. We propose PWSD, a scalable Web service discovery architecture based on peer-to-peer overlay network to overcome these limitations. In PWSD, the service descriptions are managed in a completely decentralized way. Moreover, since the basic peer-to-peer routing algorithm cannot be applied directly in the service discovery process, we extend the basic peer-to-peer routing algorithm with XML, which enables PWSD to support XML-based complicate query. Experimental results show that PWSD has good robustness and scalability.

iSurfer: A Focused Web Crawler Based on Incremental Learning from Positive Samples

This paper presents a focused Web crawling system iSurfer for information retrieval from the Web. Different from other focused crawlers, iSurfer uses an incremental method to learn a page classification model and a link prediction model. It employs an online sample detector to incrementally distill new samples from crawled Web pages for online updating of the model learned. Other focused crawling systems use classifiers that are built from initial positive and negative samples and can not learn incrementally. The performances of these classifiers depend on the topical coverage of the initial positive and negative samples. However, the initial samples, particularly the negative ones, with a good coverage of target topics are difficult to find. Therefore, the iSurfer’s incremental learning strategy has an advantage. It starts from a few positive samples and gains more integrated knowledge about the target topics over time. Our experiments on various topics have demonstrated that the incremental learning method can improve the harvest rate with a few initial samples.

A grid resource management approach based on P2P technology

This paper presents a resource management approach in grid computing environment. Using P2P technology, this novel approach can manage dynamic grid computing resources efficiently. Different kinds of grid computing resources are organized into a P2P overlay network and available information is published in type of grid service. Task requests for computational resources are also presented as grid services. Problem of resources scheduling, is solved by services discovery. During the course of allocating computing resources to an application, task requests are forward in the P2P overlay network following the Chord routing protocol. With this approach, the computational resources of a grid system can be scheduled dynamically according to the real-time workload on each peer. Furthermore, the application of this approach is introduced into DDGRID, a grid system for drug discovery and design, to evaluate the performance. Solid experimental results show that computational resources of a grid system can be managed efficiently, and the system can hold a perfect load balance state and robustness

Building efficient super-peer overlay network for DHT systems

DHT technique has been widely applied in P2P systems because it provides reliable services. However, large overheads are inevitable for maintaining the topology of DHT overlay, which limits its application especially in highly dynamic network environments. This paper proposes a DHT based super-peer overlay network called SPChord in which peer nodes are clustered according to the session time and physical network locality. With an evolving process, SPChord gradually makes DHT overlay stable and reliable. Therefore the high maintaining overheads for DHT overlay are effectively controlled. Experimental results show that the DHT overlay maintaining overheads are reduced dramatically while the overlay stability and the data availability are also greatly improved.

Distributed Information Retrieval Based on Hierarchical Semantic Overlay Network

One fundamental problem that confronts information retrieval is to efficiently support query with higher accuracy and less logic hops. This paper presents HSPIR (Hierarchical Semantic P2P-based Information Retrieval) that distributes document indices through the P2P network hierarchically based on documents semantics generated by Latent Semantic Indexing (LSI) [1]. HSPIR uses CAN [2] and Range Addressable network organize [3] nodes into a hierarchical overlay network. Comparing with other P2P search techniques [4, 5] those are based on simple keyword matching, HSPIR has better accuracy for it considers the advanced relevance among documents. We use Agglomerative Information Bottleneck (AIB) [6] to cluster documents and train Directed Acyclic Graph Support Vector Machines (DAGSVM) based on these clustered documents. Owning to the hierarchical overlay network, the average number of logical hops per query is smaller than other flat architectures.

Querying XML Data over DHT System Using XPeer

DHT systems like Chord, Pastry, CAN and Tapstry can only handle semantics-free, large-granularity requests for objects by identifier (typically a name). How to implement the content-based query in DHT system is a challenge. Query the XML data using XPath language can provide expressiveness for the DHT system. In this paper, we propose an XML-based content query system, termed as XPeer, built on DHT systems like Chord, Pastry or Tapstry. Besides the inherent properties provided by DHT, XPeer has several unique features. First, XPeer can utilize XML to implement content-based query using XPath as the query language; Second, XML data in XPeer can be totally heterogeneous, without conforming to the same XML schema or DTD. Third, XPeer can support range query over DHT. Finally, the XPeer can be easily extended to real P2P application like Napster, Gnutella and Freenet.

A Novel Agent-Based Load Balancing Algorithm for Grid Computing

This paper presents an agent-based resource scheduling algorithm for Grid computing systems. With this scheduling algorithm, the computational resources of the Grid can be dynamically scheduled according to the real-time working load on each node. Thus a Grid system can hold an excellent load balance state. Furthermore, the application of this algorithm is introduced into the practical protein molecules docking applications, which run at the DDG, a Grid computing system for drug discovery and design. Solid experimental results show the load balance and robustness of the proposed algorithm.

Targeted local immunization in scale-free peer-to-peer networks

The power-law node degree distributions of peer-to-peer overlay networks make them extremely robust to random failures whereas highly vulnerable under intentional targeted attacks. To enhance attack survivability of these networks, DeepCure, a novel heuristic immunization strategy, is proposed to conduct decentralized but targeted immunization. Different from existing strategies, DeepCure identifies immunization targets as not only the highly-connected nodes but also the nodes with high availability and/or high link load, with the aim of injecting immunization information into just right targets to cure. To better trade off the cost and the efficiency, DeepCure deliberately select these targets from 2-local neighborhood, as well as topologically-remote but semantically-close friends if needed. To remedy the weakness of existing strategies in case of sudden epidemic outbreak, DeepCure is also coupled with a local-hub oriented rate throttling mechanism to enforce proactive rate control. Extensive simulation results show that DeepCure outperforms its competitors, producing an arresting increase of the network attack tolerance, at a lower price of eliminating viruses or malicious attacks.

PBiz: An E-business Model Based on Peer-to-Peer Network

The resource of e-Business is distributed, but the existing e-Business resource management is processed in a centralized approach, which has the limitations of single point failure and performance bottleneck. We propose PBiz, an e-Business model based on peer-to-peer to overcome these limitations. In PBiz, resource is managed in a totally decentralized approach. Moreover, we extend the routing algorithm Chord with XML to enable PBiz could support SQL-like query. Experimental results show that PBiz has good robustness and extensibility.

DDGrid: Harness the Full Power of Supercomputing Systems

This paper discusses the design and implementation of drug discovery grid (DDGrid) in details. The purpose of the drug discovery grid is to set up a grid environment with aggregated compute and data resources to provide the drug virtual screening services and the pharmaceutical chemistry information services. Then, emphasizes the analysis about the architecture and key technology of implementation of the DDGrid, and gives a particular description of the components of the DDGrid. Finally, the system evaluation of the DDGrid is conducted, and the job scheduling policy is analyzed