【摘要】：隨著網(wǎng)絡(luò)的發(fā)展,互聯(lián)網(wǎng)服務(wù)成為了人們生活不可缺少的一部分,隨之而來的就是更加嚴(yán)峻的安全問題。目前除了傳統(tǒng)的硬安全手段之外,信任機制作為一種重要的軟安全手段,得到了廣泛的應(yīng)用。信任關(guān)系已經(jīng)成為互聯(lián)網(wǎng)用戶重要的決策依據(jù)。因此,如何建立可靠的信任關(guān)系,建立完善的信任機制是現(xiàn)階段熱門研究課題。本文對現(xiàn)在的信任機制研究進行了分析,并且選取了當(dāng)前熱門新興的特殊應(yīng)用--社交網(wǎng)絡(luò)進行融合。社交網(wǎng)絡(luò)也屬于P2P網(wǎng)絡(luò)拓?fù)浣Y(jié)構(gòu),所以本文在結(jié)合P2P網(wǎng)絡(luò)和社交網(wǎng)絡(luò)屬性的特點,將優(yōu)選信任模型的構(gòu)建,特別是提高信任評估的準(zhǔn)確性、抗攻擊能力和算法效率作為研究的重點,提出了一系列信任模型改進方案。本文的主要研究貢獻如下:首先,針對信任值評估計算,為克服了傳統(tǒng)信任機制中影響因素粒度過粗的問題,在信任特征選取上,綜合考慮上下文的多影響因素,采用灰度關(guān)聯(lián)方法,對特征間權(quán)重進行動態(tài)調(diào)節(jié)。結(jié)合基于相似性的間接信任計算,以解決節(jié)點數(shù)據(jù)貧乏的情況。最后,仿真實驗結(jié)果表明本文提出的信任模型不僅具有良好的準(zhǔn)確性,而且能夠有效地抵御惡意節(jié)點的攻擊。其次,研究了信任機制中的信任反饋問題。為更實時動態(tài)的調(diào)節(jié)信任值,本文采用了基于馬爾科夫鏈預(yù)測的動態(tài)反饋算法,通過馬爾科夫模型的狀態(tài)確定和無后驗性改進,完成基于信任懲罰激勵要素的狀態(tài)建立和基于時間衰減要素的狀態(tài)轉(zhuǎn)移,使用退火算法對調(diào)節(jié)因子給予優(yōu)化,模擬出信任的預(yù)測值。最后,仿真實驗表明本章算法在信任預(yù)測上具有良好的準(zhǔn)確度,并且增強了算法的防御力和健壯性。最后,針對信任機制中的算法效率過低的問題,提出了基于節(jié)點聚類優(yōu)化的算法。分析量化節(jié)點的客觀屬性和交互屬性,構(gòu)造節(jié)點的邏輯坐標(biāo),完成三維邏輯映射。根據(jù)映射的節(jié)點坐標(biāo),采用KMeans方法完成聚類。通過這種方法完成節(jié)點推薦選取。最后,仿真實驗結(jié)果證明通過本章信任模型的節(jié)點選取,既保證了信任評估的準(zhǔn)確性,又減小了算法的時間復(fù)雜度,提高了信任機制的效率。
[Abstract]:With the development of network, Internet service has become an indispensable part of people's life, followed by more serious security problems. In addition to the traditional hard security, trust mechanism, as an important soft security means, has been widely used. Trust relationship has become an important decision-making basis for Internet users. Therefore, how to establish a reliable trust relationship and establish a perfect trust mechanism is a hot research topic at this stage. In this paper, the trust mechanism research is analyzed, and the social network, a popular and emerging special application, is selected to merge. Social network also belongs to the topology of P2P network, so this paper combines the characteristics of P2P network and social network attributes, will optimize the construction of trust model, especially to improve the accuracy of trust assessment. Anti-attack ability and algorithm efficiency are the focus of the research, and a series of trust model improvement schemes are proposed. The main contributions of this paper are as follows: firstly, in order to overcome the problem of coarse influence factors in the traditional trust mechanism, the multi-influencing factors of context are comprehensively considered in the selection of trust characteristics, in order to overcome the problem of coarse influence factors in the traditional trust mechanism, aiming at the evaluation and calculation of trust values. The gray-scale correlation method is used to dynamically adjust the weights between features. The indirect trust calculation based on similarity is used to solve the problem of poor node data. Finally, the simulation results show that the trust model proposed in this paper not only has good accuracy, but also can effectively resist the attack of malicious nodes. Secondly, the problem of trust feedback in trust mechanism is studied. In order to adjust trust value in real-time and dynamically, the dynamic feedback algorithm based on Markov chain prediction is adopted in this paper, and the state determination of Markov model and the improvement of non-experientiality are adopted. The state establishment based on trust penalty incentive element and the state transition based on time attenuation factor are completed. The annealing algorithm is used to optimize the adjustment factor and to simulate the predicted value of trust. Finally, the simulation results show that this algorithm has good accuracy in the prediction of trust, and enhances the robustness and robustness of the algorithm. Finally, an algorithm based on node clustering optimization is proposed to solve the problem of low efficiency of the algorithm in trust mechanism. The objective and interactive attributes of quantized nodes are analyzed, the logical coordinates of nodes are constructed, and the three-dimensional logical mapping is completed. According to the mapped node coordinates, KMeans method is used to complete clustering. This method is used to complete the node recommendation selection. Finally, the simulation results show that the node selection of the trust model in this chapter not only ensures the accuracy of trust evaluation, but also reduces the time complexity of the algorithm, and improves the efficiency of trust mechanism.
【學(xué)位授予單位】：南京航空航天大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP393.08

【參考文獻】

相關(guān)期刊論文 前3條

1 李道豐;張小萍;鐘誠;黃汝維;;基于灰度多屬性關(guān)聯(lián)的動態(tài)信任評價模型研究[J];小型微型計算機系統(tǒng);2016年06期

2 田春岐;鄒仕洪;田慧蓉;王文東;程時端;;一種基于信譽和風(fēng)險評價的分布式P2P信任模型[J];電子與信息學(xué)報;2007年07期

3 �？�;王懷民;尹剛;;DyTrust:一種P2P系統(tǒng)中基于時間幀的動態(tài)信任模型[J];計算機學(xué)報;2006年08期

，

本文編號：2463266