ISSN 1000-1239 CN 11-1777/TP

Table of Content

01 September 2016, Volume 53 Issue 9
A Survey on the Approaches of Building Solid State Disk Arrays
Li Xiangnan, Zhang Guangyan, Li Qiang, Zheng Weimin
2016, 53(9):  1893-1905.  doi:10.7544/issn1000-1239.2016.20150910
Asbtract ( 995 )   HTML ( 3)   PDF (4585KB) ( 724 )  
Related Articles | Metrics
Flash-based solid state disks (SSDs) use flash memory chips as their storage media, which have the features of non-volatility, small size, light weight, high shock resistance, high performance, and low power consumption. Single SSDs have the drawbacks of poor random write performance and limited erase endurance. Organizing multiple SSDs with the RAID technology is promising in delivering high reliability, large capacity and high performance. However, researchers have demonstrated that applying RAID algorithms to SSDs directly does not work well and have proposed some SSD-aware RAID algorithms. In this paper, we first analyze the drawbacks of flash-based solid state disks and the RAID technology, and use performance, reliability and price as the evaluation criteria of the approaches to building SSD arrays, and choose random writes, small writes, garbage collection, load balance, erase/program cycles, wear leveling and redundancy levels as the analysis metrics. Then, we analyze and compare the advantages and disadvantages of two types of array building approaches on the disk level and the flash-chip level respectively. Finally, we summarize those different approaches and point out prospective research directions in the future.
A Hybrid Coding Scheme Based on Intra- and Inter-Device Redundancy
Bian Jianchao, Zha Yaxing, Luo Shoushan, Li Wei
2016, 53(9):  1906-1917.  doi:10.7544/issn1000-1239.2016.20150558
Asbtract ( 838 )   HTML ( 0)   PDF (3731KB) ( 430 )  
Related Articles | Metrics
The development and application of cloud computing set higher requirement for the fault-tolerant capability of the storage systems. Erasure code has been widely used to generate device-level redundancy to protect against device failures, while has less space efficiency when resisting the sector failures. Current optimization schemes for the sector failures only resist the failures of small amounts of the sectors or specific sectors. In this paper, we propose a hybrid coding scheme (intra- and inter-device redundancy, IIDR) combining inter-device redundancy with intra-device redundancy based on the homomorphism property of MDS (maximum distance separable) codes, which employs global parity sector against sector failures in the data disks when adding parity device against device failures, and optimizes the ability to process single-sector errors taking advantage of intra-device coding to generate local parity sectors. In the end, the correctness proof and performance analysis are shown in this paper, and the results indicate that our scheme can protect against device failures and sector failures of any distribution, and the computing cost of recovering single-sector errors is much lower, and the update performance is better. Compared with traditional intra-device coding schemes, our scheme comes with less space usage.
Concurrent Node Reconstruction for Erasure-Coded Storage Clusters
Huang Jianzhong, Cao Qiang, Huang Siti, Xie Changsheng
2016, 53(9):  1918-1929.  doi:10.7544/issn1000-1239.2016.20150075
Asbtract ( 757 )   HTML ( 0)   PDF (4132KB) ( 485 )  
Related Articles | Metrics
A key design goal of erasure-coded storage clusters is to minimize network traffic incurred by reconstruction I/Os, because reducing network traffic helps to shorten reconstruction time, which in turn leads to high system reliability. An interleaved reconstruction scheme (IRS) is proposed to address the issue of concurrently recovering two and more failed nodes. With analyzing the I/O flows of centralized reconstruction scheme (CRec) and decentralized reconstruction scheme (DRec), it is revealed that reconstruction performance bottleneck lies in the manager node for CRec and replacement nodes for DRec. IRS improves CRec and DRec from two aspects: 1) acting as rebuilding nodes, replacement nodes deal with reconstruction I/Os in a parallel manner, thereby bypassing the storage manager in CRec; 2) all replacement nodes collaboratively rebuild all failed blocks, exploiting structural properties of erasure codes to transfer each surviving block only once during the reconstruction process, and achieving high reconstruction I/O parallelism. The three reconstruction schemes (i.e., CRec, DRec, and IRS) are implemented under (k+r, k) Reed-Solomon-coded storage clusters where real-world I/O traces are replayed. Experimental results show that, under an erasure-coded storage cluster with parameters k=9 and r=3, IRS outperforms both CRec and DRec schemes in terms of reconstruction time by a factor of at least 1.63 and 2.14 for double-node and triple-node on-line reconstructions, respectively.
A Survey on Security and Privacy of Emerging Non-Volatile Memory
Xu Yuanchao, Yan Junfeng, Wan Hu, Sun Fengyun, Zhang Weigong, Li Tao
2016, 53(9):  1930-1942.  doi:10.7544/issn1000-1239.2016.20150581
Asbtract ( 1012 )   HTML ( 3)   PDF (3606KB) ( 582 )  
Related Articles | Metrics
In recent years, emerging non-volatile memory (NVM) technologies, such as phase change memory (PCM), spin-transfer torque RAM (STT-RAM), and memristor have gained great attention of researchers. NVM has both byte-addressable and non-volatile features, thereby making it possible to replace both traditional main memory and persistent storage. Also, NVM can be used in hybrid memory and storage architecture. Due to the advantages of low latency, high density, and low power, NVM has become the promising memory technology because of the effect of alleviating memory wall problem. However, applications can access NVM directly through ordinary load/store interface, and more important, data resided in the NVM still retains after power loss, thus it imposes new challenges of security and privacy. This paper surveys several security problems about NVM and existing solutions including persistent memory leak, stray writes, metadata security, malicious wearout attacks, and non-volatile pointer. Then, privacy issues and existing studies about NVM, such as data protection and information leaks, are discussed. Finally, we explore other potential security and privacy issues related to NVM and propose several possible solutions, such as convergence of permission and protection, security of non-volatile cache, volatile NVM, and program security.
An Approach for Identifying SDC-Causing Instructions by Fault Propagation Analysis
Ma Junchi, Wang Yun, Cai Zhenbo, Zhang Qingxiang, Wang Ying, Hu Cheng
2016, 53(9):  1943-1952.  doi:10.7544/issn1000/1239.2016.20148367
Asbtract ( 689 )   HTML ( 1)   PDF (2637KB) ( 353 )  
Related Articles | Metrics
Single event upset (SEU) is caused by external radiation in outer space and it has a great influence on computing reliability of space devices. As process technology scales, space devices become more susceptible to SEU. SEU could result in silent data corruption (SDC), which means wrong outcomes of a program without any crash detected. SDC may lead to serious failure and hence cannot be ignored. As SDC-causing fault always propagates silently, it is very difficult to detect SDC. To develop SDC detectors, SDC-causing instructions of a program should be identified as the first step. However, this step usually needs a huge number of fault injections, which is extremely time-consuming and not achievable for most applications. In this paper, we build data dependence graph (DDG) to capture the dependencies among the values of instructions. Then the inter-function and intra-function propagation that leads to SDC is analyzed and the sufficient condition of SDC-causing instructions is demonstrated. Further, we propose a novel method of identifying SDC-causing instructions. Taking advantage of the trace files of injection, our method can detect underlying SDC-causing instructions without any expensive computations. Validation efforts show that our method yields high accuracy and coverage rate with a great reduction of injection cost.
A Cache Management Algorithm for the Heterogeneous Storage Systems
Li Yong, Wang Ran, Feng Dan, Shi Zhan
2016, 53(9):  1953-1963.  doi:10.7544/issn1000-1239.2016.20150157
Asbtract ( 984 )   HTML ( 0)   PDF (3269KB) ( 732 )  
Related Articles | Metrics
The scale of storage system is becoming larger with the rapid increase of the amount of produced data. Along with the development of computer technologies, such as cloud computing, cloud storage and big data, higher requirements are put forward to storage systems: higher capacity, higher performance and higher reliability. In order to satisfy the increasing requirement of capacity and performance, modern data center widely adopts multi technologies to implement the dynamic increasing of storage and performance, such as virtualization, hybrid storage and so on, which makes the storage systems trend more and more heterogeneous. The heterogeneous storage system introduces multiple new problems, of which one key problem is the degradation of performance as load unbalance. Thats because the difference of capacity and performance between heterogeneous storage devices make the parallelism technologies hardly to obtain high performance, such as RAID, Erasure code. For this problem, we propose a caching algorithm based on performance prediction and identification of workload characteristic, named Caper (access-pattern aware and performance prediction-based cache allocation algorithm). The main idea of Caper algorithm is to allocate the load according to the capacity of the storage devices, which aims to alleviate the load unbalance or eliminate the performance bottleneck in the heterogeneous storage systems. The Caper algorithm is composed of three parts: prediction of performance for I/O request, analysis of caching requirement for storage device, and caching replacement policy. The algorithm also classifies the application workload into three types: random access, sequential access, and looping access. In order to ensure high caching utility, the algorithm adjusts the size of logic cache partition based on the analysis of the caching requirement. Besides, in order to adapt to the heterogeneous storage system, the Caper algorithm improves the Clock cache replacement algorithm. The experimental results also show that the Caper algorithm can significantly improve the performance about 26.1% compared with Charkraborty algorithm under mixed workloads, 28.1% compared with Forney algorithm, and 30.3% compared with Clock algorithm. Even adding prefetching operation, Caper algorithm also improves the performance about 7.7% and 17.4% compared with Charkraborty algorithm and Forney algorithm respectively.
Cost-Sensitive Large Margin Distribution Machine
Zhou Yuhang, Zhou Zhihua
2016, 53(9):  1964-1970.  doi:10.7544/issn1000-1239.2016.20150436
Asbtract ( 1283 )   HTML ( 2)   PDF (1285KB) ( 825 )  
Related Articles | Metrics
In many real world applications, different types of misclassification often suffer from different losses, which can be described by costs. Cost-sensitive learning tries to minimize the total cost rather than minimize the error rate. During the past few years, many efforts have been devoted to cost-sensitive learning. The basic strategy for cost-sensitive learning is rescaling, which tries to rebalance the classes so that the influence of different classes is proportional to their cost, and it has been realized in different ways such as assigning different weights to training examples, resampling the training examples, moving the decision thresholds, etc. Moreover, researchers integrated cost-sensitivity into some specific methods, and proposed alternative embedded approaches such as CS-SVM. In this paper, we propose the CS-LDM (cost-sensitive large margin distribution machine) approach to tackle cost-sensitive learning problems. Rather than maximize the minimum margin like traditional support vector machines, CS-LDM tries to optimize the margin distribution and efficiently solve the optimization objective by the dual coordinate descent method, to achieve better generalization performance. Experiments on a series of data sets and cost settings exhibit the impressive performance of CS-LDM; in particular, CS-LDM is able to reduce 24% more average total cost than CS-SVM.
Model Selection for Gaussian Kernel Support Vector Machines in Random Fourier Feature Space
Feng Chang, Liao Shizhong
2016, 53(9):  1971-1978.  doi:10.7544/issn1000-1239.2016.20150489
Asbtract ( 1260 )   HTML ( 4)   PDF (1355KB) ( 812 )  
Related Articles | Metrics
Model selection is very critical to support vector machines (SVMs). Standard SVMs typically suffer from cubic time complexity in data size since they solve the convex quadratic programming problems. However, it usually needs to train hundreds/thousands of SVMs for model selection, which is prohibitively time-consuming for medium-scale datasets and very difficult to scale up to large-scale problems. In this paper, by using random Fourier features to approximate Gaussian kernel, a novel and efficient approach to model selection of kernel SVMs is proposed. Firstly, the random Fourier feature mapping is used to embed the infinite-dimensional implicit feature space into an explicit random feature space. An error bound between the accurate model obtained by training kernel SVM and the approximate one returned by the linear SVM in the random feature space is derived. Then, in the random feature space, a model selection approach to kernel SVM is presented. Under the guarantee of the model error upper bound, by applying the linear SVMs in the random feature space to approximate the corresponding kernel SVMs, the approximate model selection criterion can be efficiently calculated and used to assess the relative goodness of the corresponding kernel SVMs. Finally, comparison experiments on benchmark datasets for cross validation model selection show the proposed approach can significantly improve the efficiency of model selection for kernel SVMs while guaranteeing test accuracy.
A Clustering Ensemble Algorithm for Incomplete Mixed Data
Shi Qianyu, Liang Jiye, Zhao Xingwang
2016, 53(9):  1979-1989.  doi:10.7544/issn1000-1239.2016.20150592
Asbtract ( 989 )   HTML ( 0)   PDF (4754KB) ( 596 )  
Related Articles | Metrics
Cluster ensembles have recently emerged a powerful clustering analysis technology and caught high attention of researchers due to their good generalization ability. From the existing work, these techniques held great promise, most of which generate the final results for complete data sets with numerical attributes. However, real life data sets are usually incomplete mixed data described by numerical and categorical attributes at the same time. And these existing algorithms are not very effective for an incomplete mixed data set. To overcome this deficiency, this paper proposes a new clustering ensemble algorithm which can be used to ensemble final clustering results for mixed numerical and categorical incomplete data. Firstly, the algorithm conducts completion of incomplete mixed data using three different missing value filling methods. Then, a set of clustering solutions are produced by executing K-Prototypes clustering algorithm on three different kinds of complete data sets multiple times, respectively. Next, a similarity matrix is constructed by considering all the clustering solutions. After that, the final clustering result is obtained by hierarchical clustering algorithms based on the similarity matrix. The effectiveness of the proposed algorithm is empirically demonstrated over some UCI real data sets and three benchmark evaluation measures. The experimental results show that the proposed algorithm is able to generate higher clustering quality in comparison with several traditional clustering algorithms.
A Novel Computation Method for Adaptive Inertia Weight of Task Scheduling Algorithm
Li Xuejun, Xu Jia, Zhu Erzhou, Zhang Yiwen
2016, 53(9):  1990-1999.  doi:10.7544/issn1000-1239.2016.20151175
Asbtract ( 1180 )   HTML ( 5)   PDF (3596KB) ( 785 )  
Related Articles | Metrics
Particle swarm optimization (PSO) is a primary intelligence algorithm to solve task scheduling problem for workflow systems in cloud computing. The inertia weight is one of the most important parameters to achieve a balance between the global and local search in PSO algorithm. However, traditional adaptive inertia weight-based PSO task scheduling algorithms usually get local optimal and cause longer execution time and higher cost for scheduling plan. The traditional adaptive inertia weight does not comprehensively represent the information of particle position, and then cannot make a suitable balance between global and local search. Hence, a novel computation method for adaptive inertia weight is proposed to improve the computation method of success value of each particle. This method shows the position state of each particle more accurately and then improves the adaptability of inertia weight by comparing the fitness of each particle with the global best particle. Then a new inertia weight-based PSO algorithm is presented to solve task scheduling problem for cloud workflow systems. The novel weight can adjust the particle velocity more correctly so that the algorithm avoids premature convergence by a proper balance between local and global search. Comparing our new adaptive inertia weight with other five traditional inertia weights (viz. constant, index decreasing, linear decreasing, random, and adaptive inertia weight), the results show that our new adaptive inertia weight-based scheduling algorithm can always achieve stable convergence speed, the optimal fitness and execution cost of the scheduling plan (i.e. roughly 18% reduction of execution cost).
Data Gathering Using Dynamic Clustering Based on WSNs Compressive Sensing Algorithm
Zhang Ce, Zhang Xia, Li Ou, Wang Chong, Zhang Dalong
2016, 53(9):  2000-2008.  doi:10.7544/issn1000-1239.2016.20150459
Asbtract ( 753 )   HTML ( 0)   PDF (3014KB) ( 392 )  
Related Articles | Metrics
One of the major challenges of designing wireless sensor networks data gathering algorithm is to reduce energy consumption, achieve better balanced consumption and prolong the lifetime of sensor network. For the current clustering algorithms of data gathering in wireless sensor networks neglecting the impact of event sources on data spatial correlation, a CS-based dynamic clustering algorithm centred on event source (CS-DCES) is proposed in this paper. Utilizing the model of Euclidean distance spatial correlation and joint sparsity model-1, the nodes that affected by the same event source are clustered together and reconstruct those nodes readings within cluster in order to increase the spatial correlation of data and reduce the number of each cluster measurement. The algorithm exploits the compressive data to calculate the location of event source and clusters dynamically. Moreover, according to simulation, we analyze the three factors that influence the algorithm performance, and they are attenuation coefficient of event sources, distance between event sources and the number of event sources, and finally the application condition of CS-DCES is given. The simulations show that CS-DCES outperforms the existing data gathering algorithms in decreasing communication cost, saving the network energy consumption and extending the network survival time under the same accuracy.
The Optimal Beacon Interval for Synchronous MAC in Low Duty-Cycle Wireless Sensor Networks
Xing Yulong, Chen Yongrui, Yi Weidong, Duan Chenghua
2016, 53(9):  2009-2015.  doi:10.7544/issn1000-1239.2016.20150463
Asbtract ( 806 )   HTML ( 0)   PDF (2473KB) ( 314 )  
Related Articles | Metrics
Energy efficiency is a fundamental theme in the design of wireless sensor networks protocols, especially for medium access control (MAC) protocols. An energy-efficient MAC protocol can significantly elongate the lifetime of wireless sensor networks by reducing the duty-cycle of sensor nodes to an ultra-low level. Synchronous MAC can be even more efficient in data transfer at the cost of requiring tight time synchronization through periodical beacon dissemination. The length of the beacon interval may greatly affect the energy efficiency of a synchronous MAC. A shorter beacon interval leads to higher synchronization cost due to frequent beacon sending and receiving, while a longer beacon interval will lead to a larger guard time and longer idle listening due to clock drift. Therefore, there is a tradeoff between these two parts of energy consumption. In this paper, we investigate the optimal beacon interval for synchronous MAC in low duty-cycle sensor networks, and then present a strategy that adaptively utilizes the optimal beacon interval in a TDMA-based MAC protocol (called Opt-TDMA). By configuring the beacon interval to its optimal value according to the data packets rate and network size, Opt-TDMA can reduce the overall power consumption of both sending/receiving beacons and data packets. Experimental results demonstrate that Opt-TDMA is more energy-efficient than pure TDMA protocol and SCP-MAC by using optimal beacon interval and contention-free transmission.
The Redundant Tree Algorithm for Uninterrupted Data Delivery
Xia Nu, Li Wei, Lu You, Jiang Jian, Shan Feng, Luo Junzhou
2016, 53(9):  2016-2029.  doi:10.7544/issn1000-1239.2016.20150380
Asbtract ( 759 )   HTML ( 0)   PDF (5913KB) ( 405 )  
Related Articles | Metrics
With the development of the big data applications, the requirement for guaranteeing un-interrupted data delivery is growing gradually. To address the problem of single node/link failure, researchers had conducted lots of work. However, current methods for guaranteeing uninterrupted data delivery face the deficiencies of low performance of the primary/secondary delivery paths and the low resistant ability to the multiple nodes/links failure. In order to solve the above problems, a circle selecting by edge sorting based redundant tree algorithm for uninterrupted data delivery CSES is presented. At first, a minimal delivery tree with the data source as the root node is built in this algorithm, on which the routing hop count of the primary paths is the smallest. Secondly, in order to reduce the hop count of the secondary paths and to enhance the resistant ability of the multiple nodes/edges failure, the edges in network topology but not included in the tree are sorted based on the sum of the nodes on the paths between the root node and the two nodes of the edge. Then, the edges are added to the minimal tree to form the redundant circles and the redundant branches of the redundant circles are added to the minimal tree. Finally, a redundant tree rooted by the data source is built. The experimental results show that, compared with the other redundant tree algorithms, the hop count of the primary/secondary paths on the redundant tree built by CSES algorithm is smaller and the resistant ability of multiple nodes/edges failure of CSES algorithm is better.
A Privacy-Preserving Data Aggregation Scheme in Wireless Sensor Networks
Fu Shuai, Jiang Qi, Ma Jianfeng
2016, 53(9):  2030-2038.  doi:10.7544/issn1000-1239.2016.20150456
Asbtract ( 928 )   HTML ( 2)   PDF (1837KB) ( 463 )  
Related Articles | Metrics
Privacy preservation is one of the most challenging problems on secure data aggregation in wireless sensor networks (WSNs). In WSNs, current data aggregation schemes with privacy preservation cannot meet the requirements of security and energy saving, and have several disadvantages such as complex computation, considerable communication load or low-security. An energy-efficient and data-loss resilient data aggregation scheme with privacy preservation is proposed in this paper. Two different forms of perturbation data are adopted to protect the data privacy of each node from being disclosed to the sink and any other nodes in the network. Firstly, from the point of view of sink intrusion, we describe the design scheme of initial perturbation data. On the basis of it, we present the construction method of second data perturbation and the operation procedures of aggregation validation for intermediate aggregators and the sink. To resist various external attacks efficiently, the technique of message authentication code is introduced. Security and property analysis show that the proposed scheme can ensure the security of nodes on the premise of lower energy power. In addition, it has a strong ability against data-loss, and both its security and energy efficiency perform better than current works.
Intrusion Detection Techniques for Industrial Control Systems
Yang An, Sun Limin, Wang Xiaoshan, Shi Zhiqiang
2016, 53(9):  2039-2054.  doi:10.7544/issn1000-1239.2016.20150465
Asbtract ( 1679 )   HTML ( 18)   PDF (3057KB) ( 1634 )  
Related Articles | Metrics
In recent decades, with the introduction of Ethernet and the more close connection with external network, an increasingly larger number of vulnerabilities have been found in the industrial control system (ICS), exposing its serious security problem. These security issues cannot be handled completely due to the variety of the vulnerability. Therefore, we must construct the defense-in-depth system for ICS. In particular, the intrusion detection system (IDS) is one of the most important parts in the defense-in-depth system of ICS. The IDS is able to discover the potential intrusion by misuse detection and anomaly detection. In this survey, we analyze the architecture and characteristics of ICS and provide the detailed descriptions of the security concept of ICS. Then, according to the characteristics of ICS, we put forward a clear requirement of ICS IDS and elaborate its connotation. Moreover, we categorize the existing IDS methods based on the detection strategy, including traffic detection, protocol detection and equipment state detection. In each category, we analyze the detection technique and discuss the detection algorithm. Finally, for future work, from the perspective of the disadvantages of current solutions and the constraints for ICS applications, we summarize some research trends of ICS IDS from the aspects of performance metric, detection technique and detection architecture.
A Requirements Elicitation Approach Based on Feature Model and Collaborative Filtering
Peng Zhenlian, Wang Jian, He Keqing, Tang Mingdong
2016, 53(9):  2055-2066.  doi:10.7544/issn1000-1239.2016.20150426
Asbtract ( 962 )   HTML ( 1)   PDF (2992KB) ( 575 )  
Related Articles | Metrics
With the rapid development of Internet and Web service related technologies,developing software system on Internet is increasingly popular. Software development is a multi-knowledge-intensive process in which requirements elicitation plays a key role. Software systems deployed on Internet need to meet the needs of various kinds of customers and users who are geographically distributed,which increases the difficulties of software requirements elicitation. Meanwhile,more and more software systems that share similar functions are deployed on Internet,which provides a new way to elicit software requirements. Toward this end,recommender systems have been leveraged in the requirements elicitation to recommend missing features for software products by comparing the requirements descriptions of existing similar software systems. In order to improve the prediction accuracy of the recommended features of the software system,a requirements elicitation approach based on feature model and KNN (K-nearest neighbors) collaborative filtering recommendation system is proposed in this paper. An algorithm named FM_KNN is presented by utilizing constraint relations between features in KNN collaborative filtering recommendation system. Experiments conducted on a real data set and a simulated data set, by comparing the proposed FM_KNN with two existing methods, i.e., KNN and an approach that combines association rule mining with KNN, show that the proposed approach can achieve higher precision.
A Temporal Logic with a Semantics Defined on the Static Structure and Dynamic Behavior of Program
Chen Donghuo, Liu Quan, Jin Haidong, Zhu Fei, Wang Hui
2016, 53(9):  2067-2084.  doi:10.7544/issn1000-1239.2016.20150370
Asbtract ( 722 )   HTML ( 0)   PDF (3145KB) ( 425 )  
Related Articles | Metrics
The paper presents an interval temporal logic—CFITL(control flow interval temporal logic) which is used to specify the temporal properties of abstract model of program, for example control flow graph, generally abbreviated to CFG. The targeted logic differs from the general sense of temporal logics, typically CTL and LTL, whose semantical models are defined in term of the state-based structures. The semantics of CFITL is defined on an ordered CFG structure, which encodes the static structure and dynamic behavior of program. The ordered CFGs are a subset of CFGs, and their topology can be summarized by partially ordered sets, such that the induced natural number intervals can be mapped onto the well-formed structures of program. In other words, the CFITL formulae have the ability of specifying the properties related to not only the dynamic behavior but also static structure of programs. After the syntax and semantics of CFITL are expounded, the problem of model checking over CFITL is detailedly discussed. Furthermore, two types of algorithms are designed: one is based on the computation of reachable state space as well as the another is based on bounded model checking employing the SMT(satisfiability modulo theories) solvers power. Because programs implemented by advanced programming languages inevitably involve complex abstract data types with unbounded domains and operators, and the semantics of CFITL is more complex than the one of CTL, the method of SMT based model checking is more practical than the method of direct search of state space. In the sequel, a prototype tool is implemented, and some case studies are conducted.
Review of the Design of Data Center Network for Cloud Computing
Wang Binfeng, Su Jinshu, Chen Lin
2016, 53(9):  2085-2106.  doi:10.7544/issn1000-1239.2016.20150962
Asbtract ( 1972 )   HTML ( 21)   PDF (7967KB) ( 1223 )  
Related Articles | Metrics
Under the influence of the development of cloud computing, the data center network is going through tremendous changes, which are reflected not only in the improvement of network size, bandwidth, abundant links, scalability, agility and the cutting down of the cost, but also in other aspects, such as the support for the VMotion dynamically, the network virtualization etc. On the basis of the main challenges of data center network for cloud computing, the paper firstly describes the network architecture of data center for cloud computing in two different aspects, which regards switches and servers as the forwarding center respectively. Next, from the perspective of the flexible deployment of network resources, the paper deeply analyzes the related proposals and key technologies concerning the VMotion dynamically in data center for cloud computing. Further, in order to summarize the application of the network virtualization in data center for cloud computing in detail, the paper mainly analyzes a variety of designs for the virtual network architecture, which includes two different types: the scalability type and the performance guaranteed type. Finally, taking the rapid development of advanced technologies into account, the paper puts forward several points of future prediction in terms of the development of data center network for cloud computing.
Energy Consumption Modeling and Optimization Analysis for MapReduce
Liao Bin, Zhang Tao, Yu Jiong, Yin Lutong, Guo Gang, Guo Binglei
2016, 53(9):  2107-2131.  doi:10.7544/issn1000-1239.2016.20148443
Asbtract ( 829 )   HTML ( 3)   PDF (10503KB) ( 843 )  
Related Articles | Metrics
The continuous expansion of the cloud computing centers scale and neglect of energy consumption factors exposed the problem of high energy consumption and low efficiency. To improve the MapReduce framework utilization of energy consumption, we build an energy consumption model for MapReduce framework. First, we propose a task energy consumption model which is based on CPU utilization estimation, energy consumption accumulation of main components and the average energy consumption estimation as well as the job energy consumption model of MapReduce. Specifically, after analyzing the energy optimization under energy consumption model, we come up with three directions to optimize energy consumption of MapReduce: optimize MapReduce energy consumption of job execution, reduce MapReduce energy consumption of task waiting and improve the energy utilization rate of MapReduce cluster. We further propose the data placement policy to decrease energy consumption of task waiting under heterogeneous environment and the minimum resource allocation algorithms to improve energy utilization rate of MapReduce jobs by the deadline constraints. A large number of experiments and data analysis of energy consumption demonstrate the effectiveness of energy consumption model and optimum policy of energy consumption.