Wang X.,Beihang University |
Li Y.,Beihang University |
Zou X.,National Computer Network Emergency Response Technical Team Coordination Center
Proceedings - 18th IEEE International Conference on High Performance Computing and Communications, 14th IEEE International Conference on Smart City and 2nd IEEE International Conference on Data Science and Systems, HPCC/SmartCity/DSS 2016 | Year: 2016
Nowadays, virtual private network has become more and more significant for internet users. As one of the key techniques among VPN architecture, the implementation of data encryption algorithm is playing an important role. In this paper, we implements a Chinese encryption algorithm SM4 towards network flows based on Tilera-Gx multiprocessor, considering some of the features of the processor: SIMD and cache. We test our improving methods on local file encryption and network flows encryption, and evaluate the performances of them, and it turns out that these methods are more efficient on throughput and delay compared with the traditional software implementation. When the size of ciphertext each time encrypted is 1024 byte, implementation with our methods achieves a speedup of about 1.50. © 2016 IEEE.
Wang B.,Tsinghua National Laboratory for Information Sciences and Technology |
Wang K.,National Computer Network Emergency Response Technical Team Coordination Center |
Lu Z.,ZTE Cooperation |
Xie T.,Tsinghua National Laboratory for Information Sciences and Technology |
Quan J.,Tsinghua University
IEEE International Symposium on Broadband Multimedia Systems and Broadcasting, BMSB | Year: 2015
With the development of mobile Internet and Internet of things (IoT), the 5th generation (5G) wireless communications will foresee explosive increase in mobile traffic. To address challenges in 5G such as higher spectral efficiency, massive connectivity, and lower latency, some non-orthogonal multiple access (NOMA) schemes have been recently actively investigated, including power-domain NOMA, multiple access with low-density spreading (LDS), sparse code multiple access (SCMA), multiuser shared access (MUSA), pattern division multiple access (PDMA), etc. Different from conventional orthogonal multiple access (OMA) schemes, NOMA can realize overloading by introducing some controllable interferences at the cost of slightly increased receiver complexity, which can achieve significant gains in spectral efficiency and accommodate much more users. In this paper, we will discuss basic principles and key features of three typical NOMA schemes, i.e., SCMA, MUSA, and PDMA. What's more, their performance in terms of uplink bit error rate (BER) will be compared. Simulation results show that in typical Rayleigh fading channels, SCMA has the best performance, while the BER performance of MUSA and PDMA are very close to each other. In addition, we also analyze the performance of PDMA using the same factor graph as SCMA, which indicates that the performance gain of SCMA over PDMA comes from both the difference of factor graph and the codebook optimization. © 2015 IEEE.
Zhu C.,Beijing University of Posts and Telecommunications |
Liu X.,National Computer Network Emergency Response Technical Team Coordination Center |
Yang Y.,Beijing University of Posts and Telecommunications |
Zhang H.,National Computer Network Emergency Response Technical Team Coordination Center |
Xu Q.,National Computer Network Emergency Response Technical Team Coordination Center
Proceedings - 2013 IEEE 7th International Symposium on Service-Oriented System Engineering, SOSE 2013 | Year: 2013
Scheduling algorithms aim to minimize the completion time and maximize the throughput of multiple tasks. However, traditional scheduling schemes cannot satisfy the requirements of internet-based virtual computing environment which contains massive and heterogeneous computing resources. Furthermore, continuously emerging Internet applications are based on virtual computing, even worsening the situation. In this paper, a scheduling optimization algorithm is proposed to achieve application-oriented resource scheduling. Computing resources are classified by their static properties in advance and ranked according to their performance grade, then tasks are marked according to their requirements, and finally a "resource sliding window" model is employed to perform resource scheduling according to the properties and requirements of tasks, achieving the goal of "proper resources serving proper tasks". Besides, the proposed algorithm suppresses traditional scheduling algorithms in load balancing which is sacrificed by many scheduling algorithms for minimizing completion time. Experimental results demonstrate that the proposed scheduling scheme effectively improve resource utilization and success rate of task execution. © 2012 IEEE.
Zhu C.-G.,Beijing University of Posts and Telecommunications |
Xin-Ran L.,National Computer network Emergency Response technical Team Coordination Center |
Yi-Xian Y.,Beijing University of Posts and Telecommunications |
Hong Z.,National Computer network Emergency Response technical Team Coordination Center |
Qian-Hua X.,National Computer network Emergency Response technical Team Coordination Center
Advances in Information Sciences and Service Sciences | Year: 2012
It is difficult to effectively solve the problem of reliable computing by the traditional fault-tolerant mechanisms due to the openness, autonomy, dynamic features of Internet-based Virtual Computing Environment (IVCE). The features of new applications on Internet which have large amount of coarse-grained concurrent tasks and the widely distributed, huge number of computing resources make the scheduling of task redundancy possible. In this paper, we propose a redundant scheduling mechanism of task based on the trust model and design the related algorithm. The experimental results show that the proposed scheduling algorithm can effectively improve system reliability and resource utilization.
Zhang Z.-Y.,Beijing University of Posts and Telecommunications |
Zhou R.-L.,National Computer Network Emergency Response Technical Team Coordination Center |
Sun B.,Beijing University of Posts and Telecommunications |
Hu Y.,Beijing University of Posts and Telecommunications
Beijing Youdian Daxue Xuebao/Journal of Beijing University of Posts and Telecommunications | Year: 2016
In order to set up an algorithm suitable for data distribution network of extensible messaging and presence protocol (XMPP) server cluster system with a relatively large number of sessions and the number of sessions, the weighted least connections algorithm was improved. This algorithm dynamically acquires the server's occupancy resources, and calculates the current load capacity and load level in real time. The servers' loading status is divided into three stages by the threshold values, and the servers are scheduled with two different algorithms. The algorithm can effectively improve the load performance of the server cluster. © 2016, Editorial Department of Journal of Beijing University of Posts and Telecommunications. All right reserved.
Chen X.,National Computer Network Emergency Response Technical Team Coordination Center |
Wang W.,Harbin Engineering University |
Man D.,Harbin Engineering University |
Xuan S.,Harbin Engineering University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012
Duplicate webpages can affect the user experience of search engine. This paper proposed webpage deletion algorithm based on hierarchical filtering according to the features of duplicate webpage. The webpage feature extraction is divided into three layers, which are paragraphs, sentences and words. The webpage features are formed by layer filtering redundant information. In the sentence layer paragraph sentences are extracted according to the sentence semantics, while in the word layer the sentences are denoised filtering based on statistics of the part of speech in them. This algorithm improves the noise immunity and the original coverage of the feature extraction. The experiments show that the proposed method can accurately filter out duplicate webpage. © Springer-Verlag Berlin Heidelberg 2012.
Sun L.,Tsinghua University |
Sun L.,National Computer Network Emergency Response Technical Team Coordination Center |
Guan X.,Tsinghua University |
Guan X.,Xi'an Jiaotong University
Qinghua Daxue Xuebao/Journal of Tsinghua University | Year: 2015
Online social networks have many topics but people's attention spans are very limited, so the many topics compete for the scarce user attention spans. This competitive relationship affects the propagation of information and the formation of public opinions. Most existing research has focused on the spread of individual topics. This study investigates the spread of multiple topics and methods to quantitatively describe the competitive characteristics. Topic and user level methods are developed to measure resource changes, topic competition intensity, user attention transitions and topic relevance. Metrics are developed for the resource variation rate and the user attention transition rate. Measurements of the actual data collected from Sina Weibo show that the total user attention level is almost stable with most user attention transitions moving from existing topics to newly appearing topics and between similar topics. These measurements provide modeling of the multi-topic propagation processes. © 2015, Press of Tsinghua University. All right reserved.
Guo C.Q.,National Computer network Emergency Response technical Team Coordination Center |
Xu J.F.,National Computer network Emergency Response technical Team Coordination Center
Applied Mechanics and Materials | Year: 2014
With the rapid development of network bandwidth, the matching-performance of regular expression is gradually of crucial importance for networking security. There are many hardware acceleration designs of regular expression matching on the basis of NFA and DFA, of which NFA designs require more logic circuit resources while the DFA designs more memory resources. However, because there are too many states and transition edges in DFA, the performance of DFA is much inferior to the performance of NFA. In this paper we designed a DFA regular expression matching algorithm fully based on FPGA logic circuit. The algorithm exploits the feature of DFA that many transitions for a state may have the same next state pointer and setting a default transition for each state of DFA will result in the reduction of logic circuit and the simplification of the electronic circuit. To evaluate performance, this DFA algorithm was mapped onto the Altera Cyclone FPGA, and got the experimental results based on the L7-filter rule set. The performance of the DFA algorithm acquired an approximate performance compared to the NFA algorithm. Experimental result shows that, compared with the NFA algorithm, in the improved DFA plan, 10% rules got a higher throughput, reaching 60% in the best case; while 62% rules cost less logic resources, saving 87% logic resources in the best case. © (2014) Trans Tech Publications, Switzerland.
Hailong L.,Northwestern Polytechnical University |
Yuyao H.,Northwestern Polytechnical University |
Wen J.,National Computer Network Emergency Response Technical Team Coordination Center |
Haifeng X.,Northwestern Polytechnical University
Proceedings of the World Congress on Intelligent Control and Automation (WCICA) | Year: 2015
The Characteristics of Cascaded Z-Source inverter and traditional Control strategy are analysed, and a maximum boost control strategy is proposed in this paper. Based on SPWM with third harmonic injection, all the zero vector states are replaced by shoot-through states, it can reduce the switching frequency as well as improve the shoot through duty cycle effectively, then increases the boost factor. Theoretical analysis is made, the formulas are deduced. Compared to the traditional Control strategy, under the same times of the voltage, maximum boost control with third harmonic injection improves the modulation index significantly, thus extending the input range of the cascaded Z-source inverter. Simulation and experimental results are given to demonstrate the new features of the improved. © 2014 IEEE.
Guo Y.,CAS Beijing Institute of Acoustics |
Yuan Q.,National Computer network Emergency Response technical Team Coordination Center |
Zhao X.,CAS Beijing Institute of Acoustics |
Liu J.,CAS Beijing Institute of Acoustics |
Yan Y.,CAS Beijing Institute of Acoustics
Journal of Computers (Finland) | Year: 2013
Audio mixing greatly degrades the watermarking security. Consequently, it is of importance to introduce mixing-attack-proof audio watermarking algorithms. This article investigates audio mixing attack. A randomized embedding spread spectrum audio watermarking algorithm is accordingly proposed to improve the watermark immunity to mixing attacks. First, balanced modulation is introduced to get rid of the host signal interference and improve the correlation score stability. With more stable correlation scores, a randomized pseudo-noise embedding strategy is proposed to decrease the interference among different mixed components and bit error rate. To further improve the synchronization performances, a multiple synchronization strategy which simultaneously synchronizes different mixed components is also proposed. The experimental results indicate that the proposed algorithm shows a strong level immunity to mixing attacks and it can resist the mixing attack with up to 5 mixed components. © 2013 ACADEMY PUBLISHER.