• Journal of Convergence for Information Technology
• /
• v.8 no.2
• /
• pp.77-82
• /
• 2018

Recently, as cloud services become popular with general users, users' information is freely transmitted and received among the information used in the cloud environment, so security problems related to user information disclosure are occurring. we propose a method to secure personal information of multiple users by making personal information stored in the cloud server and a key for accessing the shared information so that the privacy information of the multi users using the cloud service can be prevented in advance do. The first key used in the proposed scheme is a key for accessing the user 's personal information, and is used to operate the information related to the personal information in the form of a multi - layer. The second key is the key to accessing information that is open to other users than to personal information, and is necessary to associate with other users of the cloud. The proposed scheme is constructed to anonymize personal information with multiple hash chains to process multiple kinds of information used in the cloud environment. As a result of the performance evaluation, the proposed method works by allowing third parties to safely access and process the personal information of multiple users processed by the multi - type structure, resulting in a reduction of the personal information management cost by 13.4%. The efficiency of the proposed method is 19.5% higher than that of the existing method.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.6
• /
• pp.1031-1041
• /
• 2020

Cloud computing technology plays an important role in the deep learning industry as deep learning services are deployed frequently on top of cloud infrastructures. In such cloud environment, virtualization technology provides logically independent and isolated computing space for each tenant. However, recent studies demonstrate that by leveraging vulnerabilities of virtualization techniques and shared processor architectures in the cloud system, various side-channels can be established between cloud tenants. In this paper, we propose a novel attack scenario that can steal internal information of deep learning models by exploiting the Meltdown vulnerability in a multi-tenant system environment. On the basis of our experiment, the proposed attack method could extract internal information of a TensorFlow deep-learning service with 92.875% accuracy and 1.325kB/s extraction speed.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.1
• /
• pp.58-73
• /
• 2021

Cloud Service Providers (CSPs) enable their users to access Cloud computing and storage services from anywhere in quick and flexible manners through the Internet. With the basis of 'pay-as-you-go' model, it makes the interactions between CSPs and the users play a vital role in shaping the Cloud computing market. A pool of virtualized and dynamically scalable Cloud services that delivered on demand to the users is associated with guaranteed performance and cost-provisioning. It needed a costing scheme for determining suitable charges in order to secure lease pricing of the Cloud services. However, it is hard to meet the satisfied prices for both CSPs and users due to their conflicting needs. Furthermore, there is lack of Service Level Agreements (SLAs) that allowing the users to take part into price negotiating process. The users may lose their interest to use Cloud services while reducing CSPs profit. Therefore, this paper proposes a generic costing scheme for Cloud services using General Equilibrium Theory (GET). GET helps to formulate the price function for various services' factors to match with various demands from the users. It is initially determined by identifying the market circumstances that a general equilibrium will be hold and reached. Specifically, there are two procedures of agreement made in response to (i) established equilibrium supply and demand, and (ii) service price formed and constructed in a price range. The SLAs in our costing scheme is integrated to satisfy both CSPs and users' needs while minimizing their conflicts. The price ranging strategy is deliberated to provide prices' options to the users with respect their budget limit. Meanwhile, the CSPs can adaptively charge based on users' preferences without losing their profit. The costing scheme is testable and analyzed in multi-tenant computing environments. The results from our simulation experiments demonstrate that the proposed costing scheme provides better users' satisfaction while fostering fairness pricing in the Cloud market.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.7
• /
• pp.3794-3820
• /
• 2019

In recent years, a cloud environment with the ability to detect illegal behaviours along with a secured data storage capability is much needed. This study presents a cloud storage framework, wherein a 128-bit encryption key has been generated by combining deoxyribonucleic acid (DNA) cryptography and the Hill Cipher algorithm to make the framework unbreakable and ensure a better and secured distributed cloud storage environment. Moreover, the study proposes a DNA-based encryption technique, followed by a 256-bit secure socket layer (SSL) to secure data storage. The 256-bit SSL provides secured connections during data transmission. The data herein are classified based on different qualitative security parameters obtained using a specialized fuzzy-based classification technique. The model also has an additional advantage of being able to decide on selecting suitable storage servers from an existing pool of storage servers. A fuzzy-based technique for order of preference by similarity to ideal solution (TOPSIS) multi-criteria decision-making (MCDM) model has been employed for this, which can decide on the set of suitable storage servers on which the data must be stored and results in a reduction in execution time by keeping up the level of security to an improved grade.

• International journal of advanced smart convergence
• /
• v.4 no.2
• /
• pp.61-68
• /
• 2015

For a recent most companies to make efficient business management, we are using a groupware service of integrated information system of the entire group. Groupware service integrates the cooperation in excellent synergy and duplicates has been business functions of the business, through the improvement of multi-purpose business processing capacity, there is an advantage of reduced operating costs. However, if the parent company and subsidiary, or to handle common tasks such information agency, which may cause differences in the format of the data passed in the case of a need to provide a document. Therefore, in this paper, in order to solve the heterogeneity problem in data between groups, the data system of the hub base of the cloud is provided. The proposed system is intended to improve the groupware environment including the interoperability of integrated standardized environmental data sharing service.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.3
• /
• pp.911-931
• /
• 2021

Recently, most cloud services use Docker container environment to provide their services. However, there are no researches to evaluate the performance of communication libraries for multi-GPU based distributed deep learning in a Docker container environment. In this paper, we propose an efficient communication architecture for multi-GPU based deep learning in a Docker container environment by evaluating the performances of various communication libraries. We compare the performances of the parameter server architecture and the All-reduce architecture, which are typical distributed deep learning architectures. Further, we analyze the performances of two separate multi-GPU resource allocation policies - allocating a single GPU to each Docker container and allocating multiple GPUs to each Docker container. We also experiment with the scalability of collective communication by increasing the number of GPUs from one to four. Through experiments, we compare OpenMPI and MPICH, which are representative open source MPI libraries, and NCCL, which is NVIDIA's collective communication library for the multi-GPU setting. In the parameter server architecture, we show that using CUDA-aware OpenMPI with multi-GPU per Docker container environment reduces communication latency by up to 75%. Also, we show that using NCCL in All-reduce architecture reduces communication latency by up to 93% compared to other libraries.

• Journal of the Korea Convergence Society
• /
• v.8 no.11
• /
• pp.7-13
• /
• 2017

In recent years, cloud computing technology has been socially emerged that provides services remotely as various devices are used. However, there are increasing attempts by some users to provide cloud computing services with malicious intent. In this paper, we propose a property - based multi - level hierarchical approach to facilitate authentication access for users accessing servers in cloud environment. The proposed method improves the security efficiency as well as the server efficiency by hierarchically distributing a set of attribute values by replacing the order of the user 's attribute values in the form of bits according to a certain rule. In the performance evaluation, the proposed method shows that the accuracy of authentication according to the number of attributes is higher than that of the existing method by an average of 15.8% or more, and the authentication delay time of the server is decreased by 10.7% on average. As the number of attributes increases, the average overhead change is 8.5% lower than that of the conventional method.

• Journal of Digital Convergence
• /
• v.10 no.10
• /
• pp.465-473
• /
• 2012

The mobile devices and cloud computing technology introduced new content services such as N-Screen service. The DRM techniques have been developed to support interoperability and multi-platform for new environment of content service. Nevertheless, it is still inconvenient for the consumers to purchase a new DRM-supported device or to migrate some purchased contents into new device due to the change of the subscription of service. Therefore, in this paper, cloud-based DRM model which is referred as DRMaaS (DRM-as-a-Service) model, is proposed to allow the consumer to freely use and move some DRM-protected contents in various smart devices regardless of subscription of service.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.2
• /
• pp.670-686
• /
• 2017

MapReduce (MRV1), a popular programming model, proposed by Google, has been well used to process large datasets in Hadoop, an open source cloud platform. Its new version MapReduce 2.0 (MRV2) developed along with the emerging of Yarn has achieved obvious improvement over MRV1. However, MRV2 suffers from long finishing time on certain types of jobs. Speculative Execution (SE) has been presented as an approach to the problem above by backing up those delayed jobs from low-performance machines to higher ones. In this paper, an adaptive SE strategy (ASE) is presented in Hadoop-2.6.0. Experiment results have depicted that the ASE duplicates tasks according to real-time resources usage among work nodes in a cloud. In addition, the performance of MRV2 is largely improved using the ASE strategy on job execution time and resource consumption, whether in a multi-job environment.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.10
• /
• pp.429-436
• /
• 2017

There is a growing tendency to outsource sensitive or important data in cloud computing recently. However, it is very important to protect the privacy of outsourced data. So far, a variety of secure and efficient multi-keyword search schemes have been proposed in cloud computing environment composed of a single data owner and multiple data users. Zhang et. al recently proposed a search protocol based on multi-keyword in cloud computing composed of multiple data owners and data users but their protocol has two problems. One is that the cloud server can illegally infer the relevance between data files by going through the keyword index and user's trapdoor, and the other is that the response for the user's request is delayed because the cloud server has to execute complicated operations as many times as the size of the keyword index. In this paper, we propose an improved multi-keyword based search protocol which protects the privacy of outsourced data under the assumption that the cloud server is completely unreliable. And our experiments show that the proposed protocol is more secure in terms of relevance inference between the data files and has higher efficiency in terms of processing time than Zhang's one.