面向版权分析的P2P协同行为研究

发布时间：2019-05-23 00:39

【摘要】：作为网络资源共享的典型代表，P2P（Peer-to-Peer）技术成为资源共享的最主要方式，用户通过P2P技术共享音乐、电影、文档和游戏等。P2P网络中的盗版泛滥严重损害了数字媒体公司的利益。其中，“串通盗版”（即花钱购买资源的P2P客户端免费分享资源给其他用户）背后的协同行为成为P2P版权的主要威胁源，学术界和工业界对这种威胁给予了关注。研究P2P的版权问题可以归结为如下两种方式：第一种方式利用协议级签名的方法来自动识别和有害预防；第二种方式属于应用级监管，需找出非法内容和IP之间的关系进行非实时的针对用户的警告或处罚。值得注意的是，第一种方式是嵌入特定个人授权协议（例如PAP）到标准的P2P协议中，，而第二种方式在旨在不影响或不改变公认开放共享特性的标准P2P协议。事实上，在工业化实施过程中，第二种方式更实用也更容易被各方接受。本论文研究工作属于第二种方式的范畴，基于CBF(Count BloomFilter)对协同行为分析开展了针对性研究，其创新与特色主要包括以下内容： 1.提出了一种MCBFC(Multi-Count-BloomFilter-Cycle)结构存储P2P用户行为following关系，其中MCBFC是由时间标签和N个CBF组成的环状结构，每个CBF中的计数器Count都连接有一个链表，用户行为following关系(U seri, R k, User j)经哈希函数散列到相应的链表中存储； 2.提出了一种针对协同行为的基于内容反馈的following分析方法。论文提出了PFCF(Probabilistic Following with Content Feedback)模型，并分析了PFCF模型部署位置，对PFCF模型中的MCBFC存储结构，提出了一种用户行为关联挖掘算法，算法通过检测到的资源和用户，挖掘用户的相关行为，同时根据挖掘出来的用户关联行为数据进行following概率分析以及基于内容反馈的following概率分析，通过用户行为following分析预防阻止P2P版权； 3.基于CBF和MCBFC两种存储结构做了充分实验。主要从碰撞率、空间分配、时间复杂度三个方面进行对比，验证了MCBFC结构不仅比CBF结构好的存储效率，通过牺牲可容忍计算代价有效支撑历史行为关系分析，而且MCBFC环存储结构可以保存一定时间的用户行为历史数据用于分析。
[Abstract]:As a typical representative of network resource sharing, P2P (Peer-to-Peer) technology has become the most important way of resource sharing. Users share music and movies through P2P technology. Documents and games. The proliferation of piracy in P2P networks has seriously damaged the interests of digital media companies. Among them, the collaborative behavior behind "collusion piracy" (that is, P2P clients who spend money to share resources with other users for free) has become the main source of P2P copyright threat, which has been paid attention to by academia and industry. The copyright problem of P2P can be summed up in the following two ways: the first way is to use the protocol level signature method to automatically identify and prevent harm; The second method belongs to application-level supervision, and it is necessary to find out the relationship between illegal content and IP to warn or punish users in non-real time. It is worth noting that the first way is to embed specific personal authorization protocols (such as PAP) into standard P2P protocols, while the second is in standard P2P protocols that do not affect or change recognized open sharing characteristics. In fact, in the process of industrialization, the second way is more practical and easier to accept. The research work of this paper belongs to the category of the second way. Based on CBF (Count BloomFilter), the research on collaborative behavior analysis is carried out. Its innovation and characteristics mainly include the following contents: 1. In this paper, a MCBFC (Multi-Count-BloomFilter-Cycle) structure is proposed to store P2P user behavior following relationship, in which MCBFC is a ring structure composed of time label and N CBF, and the counter Count in each CBF is connected with a linked list. The user behavior following relation (U seri, R k, User j) is hashed into the corresponding linked list by hash function. 2. A content feedback based following analysis method for collaborative behavior is proposed. In this paper, the PFCF (Probabilistic Following with Content Feedback) model is proposed, and the deployment location of PFCF model is analyzed. For the MCBFC storage structure in PFCF model, a user behavior association mining algorithm is proposed. Mining the related behavior of users, at the same time, according to the excavated user association behavior data, following probability analysis and following probability analysis based on content feedback are carried out, and the P2P copyright is prevented by user behavior following analysis. 3. Full experiments are carried out based on CBF and MCBFC storage structures. This paper mainly compares the collision rate, space allocation and time complexity, and verifies that the MCBFC structure is not only better than the CBF structure in storage efficiency, but also effectively supports the historical behavior relationship analysis by sacrificing the tolerable calculation cost. Moreover, the MCBFC ring storage structure can hold the historical data of user behavior for a certain period of time for analysis.
【学位授予单位】：北京工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【相似文献】