软件学报
軟件學報
연건학보
JOURNAL OF SOFTWARE
2013年
3期
526-539
,共14页
对等网络%广义随机图%进化博弈%资源配置%泛洪
對等網絡%廣義隨機圖%進化博弈%資源配置%汎洪
대등망락%엄의수궤도%진화박혁%자원배치%범홍
P2P network%generalized random graph%evolutionary game%resource deployment%flooding
合理的资源配置能够有效地改进非结构化 P2P 网络的查询性能,提高资源副本的可获得性.当前,资源配置研究多集中在各种类型资源副本的定量分析和分布式配置策略上,节点独立地选择资源副本进行配置,并未考虑节点间配置行为的交互作用.P2P 网络中节点只维护若干与邻居节点的连接,掌握局部信息,因而在交互过程中可将节点视为有限理性节点.在分析查询性能与节点资源配置行为之间关系的基础上,构造查询性能相关的节点收益函数,将资源配置问题模型化为一种进化博弈,通过对进化过程的描述能够有效分析节点在资源配置过程中的交互关系以及可获得的查询性能.仿真实验结果表明,资源配置进化模型可获得更高的查询成功率和近似最优的平均查询跳数,且保持相对较低的冗余度.
閤理的資源配置能夠有效地改進非結構化 P2P 網絡的查詢性能,提高資源副本的可穫得性.噹前,資源配置研究多集中在各種類型資源副本的定量分析和分佈式配置策略上,節點獨立地選擇資源副本進行配置,併未攷慮節點間配置行為的交互作用.P2P 網絡中節點隻維護若榦與鄰居節點的連接,掌握跼部信息,因而在交互過程中可將節點視為有限理性節點.在分析查詢性能與節點資源配置行為之間關繫的基礎上,構造查詢性能相關的節點收益函數,將資源配置問題模型化為一種進化博弈,通過對進化過程的描述能夠有效分析節點在資源配置過程中的交互關繫以及可穫得的查詢性能.倣真實驗結果錶明,資源配置進化模型可穫得更高的查詢成功率和近似最優的平均查詢跳數,且保持相對較低的冗餘度.
합리적자원배치능구유효지개진비결구화 P2P 망락적사순성능,제고자원부본적가획득성.당전,자원배치연구다집중재각충류형자원부본적정량분석화분포식배치책략상,절점독입지선택자원부본진행배치,병미고필절점간배치행위적교호작용.P2P 망락중절점지유호약간여린거절점적련접,장악국부신식,인이재교호과정중가장절점시위유한이성절점.재분석사순성능여절점자원배치행위지간관계적기출상,구조사순성능상관적절점수익함수,장자원배치문제모형화위일충진화박혁,통과대진화과정적묘술능구유효분석절점재자원배치과정중적교호관계이급가획득적사순성능.방진실험결과표명,자원배치진화모형가획득경고적사순성공솔화근사최우적평균사순도수,차보지상대교저적용여도.
@@@@Resource deployment is an effective means to improve search performance and can also be used to enhance the availability of resource replicas in unstructured P2P networks. Most of the current studies focus on the quantitative analysis of various types of resource replicas and distributed deployment strategies. During the resource deployment process each node selects resource replica exclusively for deployment;however, the process lacks a consideration for deployment behavior interactions among participating nodes. In a P2P network, each node keeps in touch with several other neighbors and are aware of local information, so each node can be assumed to be bounded rational. This paper designs the performance-related payoff function through analyzing the relation between search performance and resource deployment behaviors of nodes, and then models the resource deployment as an evolutionary game. In terms of the description of game evolution, the study can effectively analyze the interactions among nodes and the expected search performance. The simulation results indicate that the proposed resource deployment evolutionary model achieves higher success rate and approximate optimal average hop counts while maintaining a relatively low redundancy.