Cybersecurity And Viral Marketing

Spreading dynamics on networks aims at the understanding and exploitation of the propagating laws of different things in networks.Many real-world complex systems,from macro systems such as communication system and social system,to micro systems like neural system and cellular system,can be represented by networks,which provide people the direct or indirect approach to the studies of these systems.As human society becomes more and more networked,cyberspace comes into being.While cyberspace provides a platform for people to quickly acquire information and communicate with each other,it also provides a shortcut for cyber malefactors to carry out a variety of cyber attacks,leading to severe consequences.Hence,ensuring the security of cyberspace is of great practical significance.Among different cyber attacks,there are two types of cyber security threats: computer viruses and advanced persistent threat(APT for short).Computer viruses are malware that can propagate through the Internet and can cause huge economic losses,while APT as a new type of cyber attack,has posed a severe threat to modern society.In order to mitigate the negative impact of computer viruses and APT,the defender must develop some effective defense strategies.Due to the spreading nature of computer viruses and the lateral movement of APT,the first three issues of this thesis are on how to effectively defend against these two types of cyber attacks by means of spreading dynamics on networks.Inspired by the first three works,spreading dynamics on networks was further applied to the research of viral marketing.Viral marketing is a business strategy that makes use of online social networks(OSNs)to spread marketing messages in a viral fashion so that the profit of the marketing campaign can be increased.Different OSNs such as Twitter and Facebook provide shortcuts for marketers to quickly propagate marketing messages,greatly amplifying the potential of viral marketing.Due to the spreading nature,spreading dynamics on networks is especially suited to studying the underlying mechanism of viral marketing,and then the marketers can propose some effective viral marketing strategies accordingly.The last two issues of this thesis are on how to effectively promote products by means of viral marketing strategies in the framework of spreading dynamics on networks.The contributions of this thesis are sketched as follows.(1)Study on the energy-efficient patching(EEP)problem of wireless sensor networksWireless sensor networks(WSNs for short)are vulnerable to computer viruses,leading to their malfunctions.To protect a WSN from virus attack,the virus library associated with each sensor node must be updated in a timely way,i.e.,new viruses patches must be constantly injected into the network and then forwarded to cover the whole network.As the patch injection and forwarding require energy consumption,the defender has to weigh the losses caused by the virus against the energy for implementing the patching strategy,because the energy of the WSN is usually limited.We refer to this problem as the energy-efficient patching(EEP)problem.By using node-level modeling technique and optimal control theory,we show the overall framework for solving this problem.(2)Study on the APT defense problemAn organization,especially the one with high profile and reputation,is vulnerable to APT.To mitigate the negative impact of APT on the organization,the defender needs to protect the secure hosts from being compromised as well as recover the compromised hosts.As APT attacks are usually time-continuous,the APT defense strategy must be time-varying.Moreover,as the defense resources of the defender are limited,the defender has to weigh the losses caused by APT attacks against the cost for implementing the APT defense strategy.We refer to this problem as the APT defense problem.By employing the node-level modeling technique and optimal control theory,we give the overall framework for solving this problem.(3)Study on the dynamic cloud storage recovery(DCSR)problemAdvanced persistent threat(APT)for data theft has posed a severe threat to cloud storage systems(CSSs).To protect a CSS from APT,the cloud defender has to dynamically allocate the limited security resources to recover the compromised storage servers,aiming at mitigating his total loss.However,the security resources of the cloud defender are limited.As a result,the cloud defender has to weigh the losses resulted from the leakage of data against the cost for recovering the compromised storage servers.In addition,the attacker trys to maximize the net benefit which takes the difference between the expected benefit coming from the exfiltration of valuable data and expected cost coming from the implementation of the attack strategy.In this context,the interactions between the attacker and the defender can be captured by game model.We refer to this problem as the dynamic cloud storage recovery(DCSR)problem.By means of dynamical system theory and differential game theory,we display the overall framework for solving this problem.(4)Study on the word-of-mouth marketing model with positive and negative commentsThere must be positive and negative comments(word-of-mouth)on a product and these comments can spread over different OSNs.In real world,there are many factors that influence the spreading laws of these comments.In order to understand the influence of different factors on the spreading of these comments,we propose a compartment-level dynamical system characterizing the spreading processes of the two types of comments.On this basis,we derive the equilibrium of the system and quantify the marketer's expected overall profit.Through theoretical analysis and computer experiments we obtain some interesting results,deepening our understanding of the spreading laws of word-of-mouth.(5)Study on the dynamic discount pricing problem of word-of-mouth marketingIn an OSN,different nodes have different influence.Giving discounts according to their influence in the OSN is an effective way for marketing.Moreover,due to the flaws of static discount pricing such as high cost,dynamic discount pricing is a better choice.When developing the dynamic discount pricing strategy,the marketer has to weigh the profit against the cost.We refer to this problem as the dynamic discount pricing problem.By means of node-level modeling technique and optimal control theory,we exhibit the overall framework for solving this problem.