Search

The main purpose of this paper is to improve the resilience of data center cooling system and reduce the energy consumption in data center by using the Tilted Active Tile (TATs). We deploy multiple TATs in cold aisle environment then analyze the thermal performance of racks. TATs also help to improve the reliability in case of cooling failure by increased ride through time. We observed an extended ride‐through time for data centers deployed with Tilted Active Tiles (TATs) when the CRAC blower failed. Proposed approach can maintain cooling system temperature for up to 10 minutes. We also introduced ANN based thermal models to predict the thermal performance of TATs, To train the ANN models, we collect the rack inlet and CRAC outlet temperature from Data Center. We also fixed blank panel in empty U part of racks and then analyzed that the thermal performance of racks are better and temperature distribution at bottom, middle and top of the rack were same under 30°angle. The airflow of TATs and passive tile influence on surrounding racks. That’s why we analyzed the disturbance of TATs on surrounding rack and observed that the TATs at 30°angle has positiveimpact on the surrounding racks.

With the fast development of network technology, great amount of data have been accumulated. Plenty of them are organized by using heterogeneous information networks (HIN). So mining heterogeneous information networks efficiently is very important. Node clustering is an essential part of this task. And several clustering algorithms have been proposed. As all these algorithms contain complicated optimization procedure and matrix calculation procedure, complexity of these algorithms is very high. To overcome the shortage described above, in this paper, a new clustering algorithm is proposed. In this algorithm, several parameters should be inputted. These parameters include a heterogeneous information network, meta-paths that are used and the names of target types. During the clustering procedure, a homogeneous network will be built by the proposed algorithm firstly. All the target objects of HIN are treated as nodes of this network. The instances of meta-paths are edges. After the homogeneous network is constructed, label propagation procedure can be performed. Then the clustering result will be obtained. Obviously, by using the proposed algorithm to perform clustering, the complex optimization procedure and matrix calculation procedure are eliminated. As the convergence rate of label propagation procedure is fast, the proposed algorithm is very efficient. Besides, we can find that label propagation procedure can be executed in parallel. Thus, the proposed algorithm is easy to be parallelized. In this situation, it is fit for processing large scale HIN based on server cluster. From experimental results, we can find that the proposed algorithm running faster than all the other algorithms for comparison.

Sequence alignment is a key link of bioinformatics analysis. The basic local alignment search tool (BLAST) is a popular sequence alignment algorithm with high accuracy. However, the BLAST is inefficient in comparing and analyzing a massive amount of gene sequencing data. To solve the problem, this paper designs a distributed parallel BLAST method called SparkBLAST, based on the big data technique Spark. Under the in-memory computing framework Spark, SparkBLAST identifies the task of sequence alignment, divides the sequence dataset, and compares the sequence data. The Apache Hadoop YARN was adopted to task scheduling and resource allocation. Finally, the SparkBLAST was compared with standalone BLAST through experiments. The results show that SparkBLAST realized the speedup ratio of 3.95, without sacrificing the accuracy. In other words, SparkBLAST greatly outshines the standalone BLAST in calculation efficiency. The research findings provide bioinformatics researchers a highly efficient tool for sequence alignment.

A framework for traffic congestion prediction and visualization based on machine learning and Fuzzy Comprehensive Evaluation named MF-TCPV is proposed in this paper. The framework uses DataX and DataV to implement the integration of multi-source heterogeneous traffic data and the visualization of congestion prediction results. A deep prediction model named LSTM-SPRVM based on deep learning algorithms, machine learning algorithms, and Spark parallelization technology for the prediction of traffic congestion features in the future is proposed. In MF-TCPV, traffic congestion is divided into six levels based on Fuzzy Comprehensive Evaluation and traffic congestion features such as average speed, road occupancy rate, and traffic flow density. MF-TCPV is validated based on the real data of Whitemud Drive in Canada. The experimental results demonstrate that MF-TCPV is capable of predicting the traffic congestion accurately and displaying prediction results visually. LSTM-SPRVM is better than other existing deep learning models in terms of prediction accuracy, and MF-TCPV can intuitively visualize the prediction results of traffic congestion.

With the fast development of Internet, many fields have accumulated great amount of data. And plenty of them are organized in heterogeneous information network (HIN), so analyzing HIN efficiently is very necessary. Several HIN clustering algorithms have been proposed in recent years. Most of them are based on meta-path. As instances of meta-paths can connect all the target objects directly, while clustering, all these algorithms only focus on the relationship of two target objects that are connected directly by an instance of meta-paths. In this situation, information contained in the relationship of two target objects that are not directly connected by an instance of meta-paths is neglected. These target object pairs may be very helpful for obtaining better clustering result. In order to take these target object pairs into consideration, a structural neighbor searching method is applied into the proposed algorithm. By using this method, all the indirectly connected neighbors of target objects are considered while performing clustering. Moreover, as more than one meta-paths can be found in a heterogeneous information network and each meta-path can impact the clustering result in different degree, a weight value should be assigned to each meta-path. These weight values are used to represent the relative importance of meta-paths. To obtain these weight values, a calculation method is proposed and the algorithm described in this paper tries to calculate weight values by using this method. With weight values of meta-paths and structural neighbors of target objects, all the vectors of target objects can be calculated. Then the hierarchical clustering procedure will be performed based on these vectors.

This paper proposes the design and the implementation of a Spark parallelization plan for improving the Smith-Waterman (SW) algorithm, named the Spark-OSW algorithm. Then, the Spark-OSW was verified through accuracy, performance and acceleration tests. The results show that the proposed algorithm achieved 100% accuracy, ran much faster than the SW, and performed well in cluster environment. The research findings shed important new light on the database search for gene sequences.

A virtual machine placement strategy based on sine and cosine perturbation and reverse learning particle swarm optimization is proposed to solve the problem of insufficient optimization of internal resource consumption in data center. First of all, the integer encoding method is used to solve the shortcoming of the tedious operation of binary encoding on the virtual machine placement problem. Secondly, the quality of the initial solution is improved by the inverse learning strategy to initialize the population, the method of sine and cosine perturbation is used to avoid the particle swarm optimization algorithm falling into the locally optimal solution, and the ability of exploration and development is explored by the open downward parabola adaptive adjustment. Then, with minimizing resource consumption as the optimization goal, a constrained optimization model for virtual machine placement in the data center is established. Finally, the relevant experiments prove that this strategy can effectively reduce resource consumption and ensure service quality, and it has a good application prospect.

Accurate prediction of high-speed traffic flow is of great significance for alleviating traffic pressure, reducing congestion, and reducing environmental pollution. Aiming at the low efficiency of the KNN prediction algorithm in the training and testing of traffic big data on a single machine, this paper proposes a distance-weighted KNN algorithm based on Spark. At the same time, cross-validation is used to tune the model K value, and the feature vector of the data is selected according to the inherent spatio-temporal correlation of the data. Experiments have proved that compared with traditional KNN, the weighted KNN model based on spatio-temporal feature relationships under Spark greatly reduces the time to find K nearest neighbors in the KNN algorithm, and the prediction efficiency is significantly improved while ensuring the prediction accuracy. This method can provide reliable support for highway management in the future.

All-to-All Comparison (ATAC) is a key computing mode to improve computing performance and is applied in many disciplines. Through the research on the data distribution strategy of the ATAC calculation, a novel solution framework of the ATAC data distribution strategy based on tabu search is proposed for the first time. An ATAC data distribution model is constructed, and the related algorithms of the model are designed based on the tabu search, and it is implemented on MATLAB. The framework consists of three parts: 1) the driver module responsible for interacting with users; 2) the load balancing module that solves the data distribution strategy that meets the load balancing situation; 3) the storage optimization module that solves the data distribution strategy that meets the load balancing and optimized storage. Experiments prove that the framework for solving the ATAC data distribution strategy based on tabu search has the following advantages: A) The data distribution strategy obtained by the framework enables the nodes of the distributed system participating in the ATAC calculation to achieve load balancing and complete data localization. B) The data distribution scheme can reduce 40%-50% of the storage space in the distributed system. C) The framework is more advanced than the existing ATAC data distribution algorithm in calculation speed.

Cloud computing has become more and more vital in digital world as a basic technique of many services. As the foundation of cloud computing, the cloud platform hosts various cloud services running on it and providing interfaces to cloud users. Thus, it becomes a high-value target in attacker's eyes. This paper proposes a novel attack pattern named DECH for intruding and controlling a cloud platform standing on the side of attackers. From the defenders' perspective, this paper points out the key measures to enforce cloud service security are to restrict the Trust Computing Base (TCB) size and strengthen the isolation between cloud components. Then we provide countermeasures against “Detect Escape Control and Harvest” (DECH) based on the combination of Software Guard Extensions (SGX) and Trust Domain Extensions (TDE). This research provisions references for building secure cloud services.

The problem of urban road congestion is the key to be solved urgently in China's urban traffic. To effectively predict it, this paper proposes a method for predicting urban road congestion based on the Spark platform parallel Gradient Boosting Decision Tree algorithm. First, the basic principle of GBDT algorithm is briefly introduced. Secondly, the GBDT algorithm based on the parallel design of the Spark big data platform is used to predict urban road congestion. Finally, through accuracy experiments and scalability experiments, the effectiveness of the algorithm and the performance of the algorithm under different numbers of nodes are verified in the Spark cluster. Experiments prove that the method proposed in this paper can effectively predict urban road congestion, reduce the running time, improve the prediction efficiency, and provide effective help for urban road management.

The output power prediction of wind farm is the key to effective utilization of wind energy and reduction of wind curtailment. However, the prediction of output power has long been a difficulty faced by both academia and the wind power industry, due to the high stochasticity of wind energy. This paper attempts to improve the ultra-short-term prediction accuracy of output power in wind farm. For this purpose, an output power prediction model was constructed for wind farm based on the time sliding window (TSW) and long short-term memory (LSTM) network. Firstly, the wind power data from multiple sources were fused, and cleaned through operations like dimension reduction and standardization. Then, the cyclic features of the actual output powers were extracted, and used to construct the input dataset by the TSW algorithm. On this basis, the TSW-LSTM prediction model was established to predict the output power of wind farm in ultra-short-term. Next, two regression evaluation metrics were designed to evaluate the prediction accuracy. Finally, the proposed TSW-LSTM model was compared with four other models through experiments on the dataset from an actual wind farm. Our model achieved a super-high prediction accuracy 92.7% as measured by d_MAE, an evidence of its effectiveness. To sum up, this research simplifies the complex prediction features, unifies the evaluation metrics, and provides an accurate prediction model for output power of wind farm with strong generalization ability.

HDFS has been the core storage service of Hadoop big data platform for many years, with the continuous development of deep learning, artificial intelligence and other technologies, the amount of data continues to grow, the initial batch processing offline scenario did not fit the user’s needs, the data’s access heat appears big difference. Therefore, the HDFS storage system needs to be optimized according to the degree of heat and cold of the data. The heterogeneous storage and erasure-correction code technology of HDFS can improve the storage efficiency to some extent, but users need to specify the behavior of the storage for that particular data. Therefore, it will cause a waste of cluster resources. SSM systems can follow a predetermined set of rules, according to the degree of heat and cold, it uses different storage strategies to optimize the data, so as to improve the efficiency of the whole storage system.

Aiming at improving the low efficiency and inaccuracy of the traditional thunderstorm prediction model, a paralleled and improved naive Bayesian thunderstorm prediction model SPNBC is proposed. The native Bayesian classifier (NBC) model assumes that the attribute data set is independent, which leads to inaccurate prediction. The principal component analysis algorithm is used to optimize the Bayesian algorithm to build PNBC. First, a new attribute data set is constructed by the principal component analysis (PCA) algorithm to eliminate the dependency between attributes, and a naive Bayesian classifier is constructed on the new attribute data set. Secondly, SPNBC is obtained by parallel design of PNBC based on spark framework to improve the time efficiency. Finally, taking the thunderstorm prediction in Hohhot as the application background, the accuracy rate, false alarm rate, speedup ratio and scalability ratio were used to analyze the experiment. The experimental results show that the SPNBC of this paper is better than the traditional naive Bayes algorithm and neural network algorithm, and the prediction accuracy rate and empty alarm rate are larger when dealing with massive data Performance advantages.

Traffic flow prediction with high accuracy is definitely considered as one of most important parts in the Intelligent Transportation Systems. As interfering by some external factors, the raw traffic flow data containing noise may cause decline of prediction performance. This study proposes a prediction method by combining denoising schemes and support vector machine model to improve prediction accuracy. This study comprehensively evaluated the multi-step prediction performance of models with different denoising algorithms using the traffic volume data collected from three loop detectors located on highway in city of Minneapolis. In the prediction performance comparison, five denoising methods including EMD (Empirical Mode Decomposition), EEMD (Ensemble Empirical Mode Decomposition), MA (Moving Average), BW filter (Butterworth) and WL (Wavelet) are considered as candidates, specially, four wavelet types, coif (coiflet), db (daubechies), haar and sym (symlet), are further compared based on accuracy evaluation indicators. The prediction results show that the prediction results of the model combined with denoising algorithm are better that of the model without denoising strategy. Furthermore, the improvement of the EEMD on prediction performance is higher than other denoising algorithms, and WL method with db type achieves higher accuracy than other three types. Through comparing prediction accuracy of different denoising models, this study provides valuable suggestions for selecting the appropriate denoising approach for traffic flow prediction.

This paper concentrates on the issue of weather data mining. We propose a ARIMA algorithm based on Hadoop framework, and implement an effective weather data analyzing and forecasting system. We present the procedure to parallelize the ARIMA algorithm in the Hadoop environment, and construct a scalable, easy-to maintain, and effective weather forecasting system. Several experiments are conducted and results show that the proposed system is highly effective in terms of data storage, management, as well as query.

This paper analyzes the Struts framework, the Hibernate framework, and the combined framework of Struts and Hibernate. The paper also discusses how to use the combined framework to realize the application and design of the management system for scientific research projects. This system can efficiently solve the problem of the poor expansibility and poor maintainability in the application of traditional Java Web.

Through intensive study of web page automatic generation technology, this paper proposes a web page code automatic generation system on the basis of XSLT technology and describes a practical implementation of the system. Web Page code automatic generate system can achieve auto-generation of underlying system and repetitive codes, reduce boring duplication of coding in software development, improve the efficiency of developer as well as enhance system maintainability.