本文選題：霧計算　切入點(diǎn)：系統(tǒng)魯棒性　出處：《北京科技大學(xué)》2017年博士論文

【摘要】：伴隨著移動智能終端、無線網(wǎng)絡(luò)技術(shù)的日益推廣,越來越多的數(shù)據(jù)產(chǎn)生并分布于網(wǎng)絡(luò)環(huán)境中。數(shù)據(jù)的增長給計算服務(wù)帶來壓力,傳統(tǒng)的云計算網(wǎng)絡(luò)架構(gòu)無法滿足異構(gòu)、低延時、密集化的網(wǎng)絡(luò)接入和服務(wù)需求。霧計算將云計算擴(kuò)展到了網(wǎng)絡(luò)的邊緣,解決了云計算移動性差、地理信息感知弱、時延高等問題。然而,霧計算興起的同時也給用戶的數(shù)據(jù)安全帶來了新的挑戰(zhàn)。一方面,霧計算位于網(wǎng)絡(luò)的邊緣,更靠近用戶,復(fù)雜的應(yīng)用環(huán)境和多樣的業(yè)務(wù)服務(wù)使得數(shù)據(jù)可靠性得不到保證,這就需要采用更好的保護(hù)技術(shù)加強(qiáng)霧計算系統(tǒng)的穩(wěn)定性、可靠性。另一方面,霧計算環(huán)境中業(yè)務(wù)數(shù)據(jù)面臨著各式各樣的安全威脅,傳統(tǒng)的數(shù)據(jù)安全機(jī)制很難抵御惡意入侵和破壞,需要新的數(shù)據(jù)安全技術(shù)來應(yīng)對。本論文采用復(fù)雜網(wǎng)絡(luò)理論、微分博弈理論、超圖理論等作為數(shù)學(xué)工具,圍繞霧計算的結(jié)構(gòu)模型、系統(tǒng)安全魯棒性、入侵檢測技術(shù)、惡意程序傳播模型、密鑰管理技術(shù)等方面展開深入研究,其主要工作和創(chuàng)新性成果概括如下:(1)提出一種基于BA無標(biāo)度網(wǎng)絡(luò)和ER隨機(jī)網(wǎng)絡(luò)混合的霧計算演化模型。針對霧計算演化過程,采用BA無標(biāo)度網(wǎng)絡(luò)和ER隨機(jī)網(wǎng)絡(luò)混合模型分析網(wǎng)絡(luò)特性。研究霧計算系統(tǒng)模型在遭受隨機(jī)攻擊、蓄意攻擊后,其結(jié)構(gòu)和功能的變化,并給出相應(yīng)計算公式。所提模型為分析霧計算系統(tǒng)魯棒性提供了一個通用框架,對保障霧計算數(shù)據(jù)安全,構(gòu)建抗隨機(jī)攻擊、蓄意攻擊的霧計算環(huán)境有著指導(dǎo)意義。(2)提出基于微分博弈的霧計算入侵檢測算法模型。在考慮入侵檢測系統(tǒng)檢測率、誤報率基礎(chǔ)上,提出一種基于非合作微分博弈的霧計算入侵檢測最優(yōu)決策方法,在保證數(shù)據(jù)安全的前提下達(dá)到盡可能減少能量消耗的目的。算法利用非合作微分博弈理論為霧計算入侵檢測安全策略建模,根據(jù)業(yè)務(wù)負(fù)載動態(tài)調(diào)整防御資源量。通過求解反饋納什均衡解,分析霧計算入侵檢測系統(tǒng)中攻擊者和防御者之間的相互作用以及各自的最優(yōu)策略。仿真實(shí)驗(yàn)對基于微分博弈的霧計算入侵檢測算法模型的最優(yōu)策略進(jìn)行模擬,分析了入侵檢測系統(tǒng)在靜態(tài)最優(yōu)策略以及本文提出的動態(tài)最優(yōu)策略下的系統(tǒng)收益。實(shí)驗(yàn)顯示,與靜態(tài)最優(yōu)策略相比,提出的動態(tài)最優(yōu)策略的能量消耗要明顯低于靜態(tài)最優(yōu)策略。(3)提出一種基于微分博弈的霧計算惡意程序傳播防御模型�；趥魅静恿W(xué)模型,在考慮霧計算節(jié)點(diǎn)特征的基礎(chǔ)上,將霧計算節(jié)點(diǎn)劃分為易感染(Susceptible)、感染(Infectious)、康復(fù)(Recovered)、死亡(Dead)四種狀態(tài),進(jìn)而分析霧計算惡意程序傳播過程�；谖⒎植┺睦碚�,建立了一個霧計算環(huán)境下的惡意程序動態(tài)傳播模型,分析了霧計算和惡意程序在安全補(bǔ)丁安裝過程中的相互作用與利益沖突。通過求解博弈模型的鞍點(diǎn)策略,分析了霧計算系統(tǒng)和惡意程序各自的動態(tài)最優(yōu)策略,并與靜態(tài)最優(yōu)策略進(jìn)行了對比。實(shí)驗(yàn)顯示,所提出的動態(tài)最優(yōu)策略的能量消耗要低于靜態(tài)最優(yōu)策略,并明顯抑制了惡意程序在霧計算環(huán)境中的傳播。(4)提出一種基于超圖結(jié)構(gòu)的霧計算密鑰管理方案。針對霧計算網(wǎng)絡(luò)節(jié)點(diǎn)計算能力、存儲能力相對較弱的問題,采取超圖理論來匹配霧計算網(wǎng)絡(luò)拓?fù)鋸亩档兔荑�更新的開銷�；谟脩粢混F—云三層服務(wù)結(jié)構(gòu),將霧計算網(wǎng)絡(luò)劃分為兩個子網(wǎng)絡(luò),并對每個子網(wǎng)的密鑰更新方案進(jìn)行設(shè)計。通過系統(tǒng)性能對比,所設(shè)計的密鑰管理方案能夠在保證系統(tǒng)前向、后向安全的前提下,有效降低了系統(tǒng)的密鑰開銷。
[Abstract]:With the growing popularity of mobile intelligent terminal, wireless network technology, more and more data is produced and distributed in the network environment. The growth of data bring pressure to the traditional computing services, cloud computing network architecture can not meet the heterogeneous, low delay, intensive network access and service demand. The fog calculation will be extended to cloud computing the edge of the network, to solve the cloud poor mobility, geographic information awareness is weak, the problem of higher delay. However, the rise of the fog calculation but also to the user's data security has brought new challenges. On the one hand, the network is located in the edge of the fog calculation, closer to the user, the complex application environment and a variety of business services such as the reliability of the data can not be guaranteed, it needs to strengthen the stability of fog computing systems using protection technology better reliability. On the other hand, fog computing environment facing every kind of business data. Data security threats, traditional security mechanism is difficult to resist the malicious attack and destruction, need to deal with the new technology of data security. This paper uses complex network theory, differential game theory, such as hypergraph theory as a mathematical tool, the structure model of fog computing, system security robustness, intrusion detection, malware propagation model and in-depth research on key management techniques and other aspects, the main work and innovative achievements are summarized as follows: (1) based on the BA scale-free network and ER network hybrid evolutionary computation model. Aiming at the fog fog computing evolution process, using BA network and ER network characteristics of stochastic network mixed model standard. Research on fog computing system model subjected to random attack, attack, change its structure and function, and the corresponding calculation formula. The proposed model provides for the analysis of fog computing system robustness 浜嗕竴涓，

本文編號：1677046