Entity

Time filter

Source Type


Zhao L.,State Key Laboratory of Integrated Service Networks | Ge J.,State Key Laboratory of Integrated Service Networks
Proceedings 2010 IEEE International Conference on Information Theory and Information Security, ICITIS 2010 | Year: 2010

Multi-Channel adaptive decision feedback equalizer (MC-ADFE) is a main method to overcome multi-path distortion and eliminate Inter-Symbol Interference (ISI) for underwater acoustic channel. LDPC codes are capacity achieving codes, which can provide considerable coding gain. In this paper, the turbo theory is applied to form iterative multi-channel adaptive decision feedback equalizer (MC-ADFE) and LDPC decoder. In iterative equalization and decoding algorithm (IED), the MC-ADFE and LDPC decoder exchange soft information in an iterative manner, which can enhance the equalizer performance using significant decoding gain. The simulation results show that proposed system scheme can improve the performance of equalizer and get considerable coding gain. So, the performance of underwater acoustic communication system can be improved greatly. © 2010 IEEE. Source


Deng L.,Fudan University | Deng L.,State Key Laboratory of Integrated Service Networks | Zhao J.,Fudan University | Zhao J.,State Key Laboratory of Integrated Service Networks | And 2 more authors.
Jisuanji Yanjiu yu Fazhan/Computer Research and Development | Year: 2010

Network coding is a kind of novel network transmission technology, which can achieve the network multicast capacity. In this paper, an overhead optimization problem (OOS) is presented to minimize the sum cost of network coding and network link in networks with network coding. This optimization takes not only the link transmission overhead but also the coding overhead incurred by network coding into consideration. By using two different network information flow models, which are the receivers' data flow mixture model and the data element flow model, the general forms of both kinds of overhead are presented separately. As this problem is NP-hard, only the approximately optimized solution can be got through heuristic algorithms. In the experiments, different genetic algorithms' performance are evaluated under two different information flow models and different random generated topologies. Because it makes the optimization problem even more complex that takes the network coding overhead into consideration, the computed approximate solution may be poorer than only considering the network link cost. Simulation results show that only under the condition of the overhead coefficient γ>1000, the joint optimization can get a better solution. This result may have considerable importance when network coding comes into practice in the future. Source


Zhu H.,State Key Laboratory of Integrated Service Networks | Zhu H.,Xidian University | Chen S.,State Key Laboratory of Integrated Service Networks | Zhu L.,State Key Laboratory of Integrated Service Networks | And 2 more authors.
Proceedings - 5th International Conference on Intelligent Networking and Collaborative Systems, INCoS 2013 | Year: 2013

In order to conduct fine-grained network management in mobile network, Traffic Classification or Detection is widely used to divide network traffic into different classes, according to their source applications. Many techniques are exploited in Traffic Classification. Among them, machine learning has grown considerably attention because of its accuracy. Feature selection chooses feature combinations for machine learning algorithms, and has significant influence on the accuracy and efficiency. To discover optimal features, all possible combinations need to be evaluated by testing real classifiers. With numerous features, feature selection can cost an abundance of time and computational resources. This paper proposes a feature selection algorithm for C4.5 Decision Tree. This algorithm utilizes structural characteristics of C4.5 algorithm to exclude some of the combinations without actually testing the classifiers. The simulation results demonstrate that the algorithm can reduce the number of tests in seeking the optimal feature combination. © 2013 IEEE. Source


Zhu H.,State Key Laboratory of Integrated Service Networks | Zhu H.,Xidian University | Xue Y.,State Key Laboratory of Integrated Service Networks | Zhang Y.,State Key Laboratory of Integrated Service Networks | And 3 more authors.
Proceedings - 5th International Conference on Intelligent Networking and Collaborative Systems, INCoS 2013 | Year: 2013

As the infrastructure of cloud computing, the virtualization security problem has become more severe since the security requirement of different virtual machines (VMs) may conflict with the others in could computing platform. Based on the mandatory access control mechanism, a new mulitlevel security access control model V-MLR is proposed, which not only provided secure communication mechanism for virtual machine monitor (VMM) and VMs, but also updated the borrowed information in VMM synchronously when it was changed in VMs. Case study and application in Xen showed that V-MLR enhanced the security of virtual machine system without causing significant performance penalty. © 2013 IEEE. Source


Chen Z.,State Key Laboratory of Integrated Service Networks | Gu H.,State Key Laboratory of Integrated Service Networks | Yang Y.,Institute of Microelectronics | Bai L.,State Key Laboratory of Integrated Service Networks | Li H.,State Key Laboratory of Integrated Service Networks
IEEE Computer Architecture Letters | Year: 2014

Optical interconnect is a promising alternative to substitute the electrical interconnect for intra-chip communications. The topology of optical Network-on-Chip (ONoC) has a great impact on the network performance. However, the size of ONoC is limited by the power consumption and crosstalk noise, which are mainly resulted from the waveguide crossings in the topology. In this paper, a diagonal Mesh topology (DMesh) is proposed to relieve the limitation of scalability by reducing the number of waveguide crossing, which is only 20% that of Mesh. In addition, the number of optical routers in DMesh is less than half of that in Mesh-based ONoC. Due to its compact architecture and favorable scalability, DMesh topology is suitable for large-scale ONoC design. © 2002-2011 IEEE. Source

Discover hidden collaborations