Real-time flow transmission is widely used in various online applications, such as video transmission, social networks, etc. Most of these tasks which require strict deadlines need to be handled by distributed systems, so the deadline guarantee of the whole system can be divided into each node. However, due to insufficient computing power or software errors, some nodes are prone to delay data processing, leading to the loss of computing deadline, resulting in the loss of the cutoff event of the whole system, and causing the data processing delay of the next link. In view of the performance loss caused by this situation, this paper proposes a cache-based processing framework to speed up the forwarding of real-time data flows. By estimating the data processing ability of nodes online, when the processing ability of nodes is out of order, some computing tasks can be mapped to other preparatory nodes online, so as to achieve fast task. Migration ensures the deadline of computing tasks on nodes. By adjusting reasonable system parameters, the processing capacity of nodes in distributed system can be accelerated, so that the processing capacity of the system can be increased dynamically and the data flow forwarding capability of the whole system can be accelerated. The results show that this method can increase the number of data forwarding by 23% and reduce the data loss rate by 31%.

Automatic expression recognition of human faces has been an active research area for decades. In this work, to improve the facial expression recognition effect, a new method based on SSVM algorithm and multi-source texture feature fusion using KECA is proposed. Multi-source texture features are introduced to describe the facial expression, containing GMCL, Gabor feature, and HOG feature. The results indicate that multi-source texture features are conducive to improve the recognition effect and make up for the deficiency of single texture feature in facial expression description. In addition, KECA and SSVM algorithms show better performance than traditional methods in feature extraction and classification. To further verify the effectiveness of the proposed method, three sets of comparative experiments are carried out: PCA+SVM (based on Gabor feature), PCA+SVM (based on GMCL+Gabor+HOG feature), KECA+SVM (based on GMCL+Gabor+HOG feature). The results of JAFFE database indicate that the accuracy of proposed method, equal to 93.04%, is at least 2.61% higher than conventional method. The results of JAFFE database demonstrate the validity of proposed method.

Data centers provide services for various real-time applications, such as social networks, instance message, which produce a large number of bursty and urgent mice flow. However, the traditional flow queuing model cannot schedule these flow effectively, which leading the performance loss for some important applications. In this paper, a preemptive scheduling in data center for bursty flow (PSBF) was proposed. The scheduling leverage the preemptive scheduling for more critical flow and set the right queue for other data flow to keep the high aggregated throughput. The experiment and analysis show that our scheduling can handle bursty flow better than traditional flow scheduling for data center.

Modular data center networks leverage recursive topology to speed up several traffic patterns and increase the network capacity for various data shuffle applications. However, the applications of online social networks and instant messaging may be deployed in the same data center for high cost-effective investment. The large amount of mice flows which produced from these applications may cause partial path congestion, and these congestion in turn can prevent the mice flow access that reduces message loss or delay for online social networks. In this stydy, we define the cost function for different parallel paths and model the path selection process, and propose an adaptive flow scheduling for modular data center networks (AFMD) to search more suitable path for both large flows and mice flows with the greedy algorithm. The experiment shows that AFMD can improve large flow throughput by up to 25.9 % and the mice flow access ratio by up to 16.5 % and AFMD exhibits effective complementarity and compatibility to BCube networks.

Online social networks (OSN) have become a prevalent communication tool with a user scale of billions. Users send messages and share photos and videos to their friends every day. As a consequence, a large amount of flows are produced by file storage and backup in the back end. Most flows are deadline intensive, and the scale of flow account is large; these characteristics create new challenges for flow scheduling in data center networks. In this study, we conclude the flow distribution of OSN based on previous studies. OSN flow can be classified into a large amount of bursty flow and a limited amount of large flow. The bursty mice flows are mostly critical real-time requirements, which support instant messages and routing information; the large flows seldom require real-time transmission, because they mainly support file transferring and updating. A mixed flow causes bottlenecks in rack switches easily, thereby resulting in deadline loss for real-time applications and lower throughput. The study divides flow based on different metrics and provides priority-based flow scheduling for OSN datacenter (PFO). PFO allocates rate for different flows on the basis of flow size and deadline information. Our objective is to decrease the average completion time for bursty flows and ensure a high throughput. The analysis and evaluation shows that the proposed scheduling increases the goodput more than DCTCP and D2TCP, and decreases the fraction of missed deadlines.

Data centers employ virtualization to organize thousands of servers. Applications such as MapReduce and Hadoop are deployed in virtual machines (VMs). However, with the development of client requirements, the traffic among VMs, which host different types of applications, becomes irregular and unpredictable. This traffic may cause congestion links and performance violations in applications. In this study, we first analyze the performance impact of changes in flow distribution and then attempt to migrate VMs with large traffic flows to a suitable destination server to decrease the number of congestion links. We design a multi-level bloom filter to detect large flows and propose a flow scheduling algorithm based on VM migration (FSVMM) for data centers to optimize network performance. The FSVMM is evaluated with real trace data. Results show that the FSVMM can reduce the number of congestion links and improve user experience.

The data center networks (DCN) leverage redundant physical links to overcome the bottleneck of traditional tree topology and achieve high bisection bandwidth and goodput for cloud computing platform, so how to use these redundancy links efficiently becomes the key factor for routing designer. This paper analyzes the large flow collision in modular data center network and propose simple experiments to validate the performance decreasing in the aggregated throughput. And then, a recursive congestion-aware routing algorithm (called CaRA) is proposed for decreasing flow collision. We simulate the CaRA with NS-2 simulator, and the results show that CaRA can improve one-to-one packet forward throughput and all-to-all throughput most time. The peak throughput values in these two situations are 50% and 20% respectively.

The self-organizing nature of its architecture of wireless sensor networks (WSNs) introduce unique challenges for privacy preservation of data. This paper analyzes the cluster-based private data aggregation protocol (CPDA), and proposes a privacy protection protocol based on hierarchical cluster. We firstly reorganize nodes in cluster according to the logic structure of binary-tree in which each node transmits slice data instead of the full one. And then, we establish the hierarchical privacy tree to manage the group of keys, which purpose is privacy protection in case of data aggregation.

It is urgent for China's IT service management about how to carry out IT operations and maintenance efficiently through a standard way.To improve the stability of the operation of the system,the demand of maintenance service for the information system is analyzed,and the development and design of the intelligent fault diagnosis and processing system based on ITSS is introduced in detail.

Many colleges and universities have set up Linux course to meet the social requirement for talents .However , there still exist many problems in the setting of the curriculum and the practice ability of students can not be improved ef-fectively .In this paper ,we analyze the course system of Linux and the demand of instructional techniques combined with the project driven teaching model .Based on the above ,we present the method of how to apply the project driven teaching mode for this curriculum to achieve the students'autonomous learning and enhance the interest of Linux study .

In order to ensure the data privacy of the sensor nodes and improve the network life cycle,this paper proposes a low energy-consuming private data aggregation method based on unequal clustering(UCDA).Firstly,the tentative cluster heads construct clusters of unequal sizes by using uneven competition ranges according to the distance to Base station.Then,the fragment recombination technology is adopted.To reduce the communication overhead and obtain the data privacy.the node in the overlap area of different clusters slice its private data into pieces,and then it send to the cluster heads for mixing in the next process.Simulation results show that UCDA can preserve data privacy,get accurate data aggregation results and cost less communication and computation overhead when compared with the SMART algorithm.

During the process of the data fusion in WSNs, the redundant highly correlated data and the overlapping region may lead to low precision of monitoring information and energy depletion. In this paper, we proposes the use of the correlation function in fuzzy theory to calculate the mutual support degree between the nodes, and compress the data from the nodes with higher mutual support degree. In order to reduce the energy consumption, we also consider that the redundant nodes, which should be changed into sleeping mode, could be chosen in terms of the fusion result and the expectation of QoS. The results demonstrate that the presented method can obtain much higher precision of collecting data and more reliable, also it can prolong the network lifetime effectively.

In wireless sensor networks, the nodes usually need satisfy the QoS requirement of forwarding the packets with different priorities. This way traffic flows are served in a differentiated manner, with higher priority traffic flows being allocated more bandwidth on the average than the lower priority traffic flows. This paper proposed an adaptive congestion control strategy for multi-hop WSNs. Two types of buffers in the relay nodes for the differentiated classes of flow are defined as the Premium flow and the Ordinary flow. The node periodically samples its queue occupancy and the packet rates of its upstream neighbors, and infers whether congestion will occur in the next interval. Upon receiving the congestion feedback, these upstream nodes decrease their rate according to the queue occupancies. The simulation results show that the proposed strategy can avoid the network congestion and jitter effectively, and obtain high utilization of the nodes' capacity and buffer size.