A SCALABLE APPROACH TO JOINT CYBER INSURANCEAND SECURITY-AS-A-SERVICE PROVISIONING IN CLOUDCOMPUTING

 

ABSTRACT

As computing services are increasingly cloud-based, corporationsare investing in cloud-based security measures. The Security-asa-Service(SECaaS) paradigm allows customers to outsource securityto the cloud, through the payment of a subscription fee. However, nosecurity system is bulletproof, and even one successful attack can resultin the loss of data and revenue worth millions of dollars. To guardagainst this eventuality, customers may also purchase cyber insuranceto receive recompense in the case of loss. To achieve cost effectiveness,it is necessary to balance provisioning of security and insurance, evenwhen future costs and risks are uncertain. To this end, we introduce astochastic optimization model to optimally provision security and insuranceservices in the cloud. Since the model we design is a mixed integerproblem, we also introduce a partial Lagrange multiplier algorithm thattakes advantage of the total unimodularity property to find the solutionin polynomial time. We also apply sensitivity analysis to find the exacttolerance of decision variables to parameter changes. We show theeffectiveness of these techniques using numerical results based on realattack data to demonstrate a realistic testing environment, and find thatsecurity and insurance are interdependent.

EXISTING SYSTEM:

There are two aspects to the system model which we proposein this paper. The first is the problem of security serviceallocation, and the second is cyber insurance provisioning.Research in this area primarily addresses the problem ofsecurity allocation, the setting of cyber insurance parameters,and whether there is a symbiotic relationship betweenInternet security and cyber insurance.The notion of Security-as-a-Service (SECaaS) was introducedin, where it was proposed as a way of securingcloud-based data through encryption and distribution ofdata. addresses the problem of selecting cloud serviceproviders (CSP) with security considerations as a priority.The authors propose a framework to manage risk througha combination of technology, processes, and people. considers allocation of resources in a parallel computingcontext with the security overhead considered for bothheterogeneous and homogeneous systems. similarly dividessecurity services by priority to optimize processing requirementsin a mobile cloud context. Considering real-timesystems that are security-critical, provisions securityservices to optimize performance, where the scheduling ofjobs is combined with the allocation of security services. takes an approach that is both security-aware and budgetaware.introduces the idea of firewall-style SECaaSproviders, similarly introduces an API called FlowTapto provide a security policy enforcement and monitoringinfrastructure for network traffic. This is shown to be importantas, whilst users can install security software in a virtualenvironment, they have no control over the network trafficin the cloud. We consider this approach for the SECaaSproviders in this paper, which focuses on network trafficanalysis.

PROPOSED SYSTEM:

Ourcontributions are summarized as follows:_ We devise a stochastic optimization for a customerto jointly provision security services and buy cyberinsurance premiums across multiple time periods.We account for uncertainty in traffic quantities andattack frequency, as well as future uncertainty ofsecurity service prices and insurance premiums._ Due to the tractability problems of integer programming,we introduce a partial Lagrange multiplieralgorithm to find the optimal solution in, at worst,polynomial time. We provide proofs of convergenceand scalability._ We perform a sensitivity analysis, which providesprecise values for solution tolerance to parameterchange. We then demonstrate the effectiveness of ourmethods through evaluation of an example scenario,based on analysis of real attack data to providerealistic parameter settings.

 

 

CONCLUSION

In this paper we have presented a combined approach tosecurity and cyber insurance provisioning in the cloud. Usinga stochastic optimization, we have presented a methodof optimally provisioning both services in the face of uncertaintyregarding future pricing, incoming traffic and cyberattacks. Since our optimization involves solving an integerprogramming problem, we present the partial Lagrangemultiplier method, which exploits the total unimodularityproperty to guarantee integer solutions, while relaxing theproblem to a linear programming problem. This problemis solved iteratively using a subgradient method, which weprove converges to the optimal solution in at worst polynomialtime. Using the solution produced by the algorithm, weapply an analytical sensitivity analysis approach that givesprecise sensitivity values for individual parameters. Finallywe provide an experimental evaluation of our contributionsusing realistic traffic and attack data derived by runningreal traffic data through an Intrusion Detection System. Themain challenge of cyber insurance is the number of assumptionsthat must be made, for example, the ability to detectcyber attacks, establish accurate damages, and successfullymake insurance claims. Future extensions could consider theinteraction of applications, where the security performanceof one part of the system can impact the security of otherparts. We have introduced real honeypot data, but futureextensions could consider more extensive data to producemore accurate options. Accuracy of data could further beextended through the implementation of systems to updateparameters on a daily or weekly basis, to improve futuredecisions.

REFERENCES

[1] McAfee, “Net Losses: Estimating the Global Cost of Cybercrime,”Center for Strategic and International Studies, Economic Impact ofCybercrime II, Jun. 2014.

[2] (2016) Identity theft resource center data breach reports. [Online].Available: http://www.idtheftcenter.org/2016databreaches.html