1-9hit |
Software-defined networking (SDN) decouples the control and forwarding of network devices, providing benefits such as simplified control. However, due to cost constraints and other factors, SDN is difficult to fully deploy. It has been proposed that SDN devices can be incrementally deployed in a traditional IP network, i.e., hybrid SDN, to provide partial SDN benefits. Studies have shown that better traffic engineering performance can be achieved by modifying the coverage and placement of SDN devices in hybrid SDN, because they can influence the behavior of legacy switches through certain strategies. However, it is difficult to develop and execute a traffic engineering strategy in hybrid SDN. This article proposes a routing algorithm to achieve approximate load balancing, which minimizes the maximum link utilization by using the optimal solution of linear programming and merging the minimum split traffic flows. A multipath forwarding mechanism under the same problem is designed to optimize transmission time. Experiments show that our algorithm has certain advantages in link utilization and transmission time compared to traditional distributed routing algorithms like OSPF and some hybrid SDN routing mechanisms. Furthermore, our algorithm can approximate the control effect of full SDN when the deployment rate of SDN devices is 40%.
Chang-chun ZHANG Long MIAO Kui-ying YIN Yu-feng GUO Lei-lei LIU
A fully-integrated double-channel 5-Gb/s/ch 2:1 serializer array is designed and fabricated in a standard 0.18-$mu $m CMOS technology, which can be easily expanded to any even-number-channel array, e.g. 12 channels, by means of arrangement in a parallel manner. Besides two conventional half-rate 2:1 serializers, both phase-locked loop and delay-locked loop techniques are employed locally to deal with the involved clocking-related issues, which make the serializer array self-contained, compact and automatic. The system architecture, circuit and layout designs are discussed and analyzed in detail. The chip occupies a die area of 673,$mu $m$, imes ,$667,$mu $m with a core width of only 450,$mu $m. Measurement results show that it works properly without a need for additional clock channels, reference clocks, off-chip tuning, external components, and so on. From a single supply of 1.8,V, a power of 200,mW is consumed and a single-ended swing of above 300,mV for each channel is achieved.
Constructing accurate similarity graph is an important process in graph-based clustering. However, traditional methods have three drawbacks, such as the inaccuracy of the similarity graph, the vulnerability to noise and outliers, and the need for additional discretization process. In order to eliminate these limitations, an entropy regularized unsupervised clustering based on maximum correntropy criterion and adaptive neighbors (ERMCC) is proposed. 1) Combining information entropy and adaptive neighbors to solve the trivial similarity distributions. And we introduce l0-norm and spectral embedding to construct similarity graph with sparsity and strong segmentation ability. 2) Reducing the negative impact of non-Gaussian noise by reconstructing the error using correntropy. 3) The prediction label vector is directly obtained by calculating the sparse strongly connected components of the similarity graph Z, which avoids additional discretization process. Experiments are conducted on six typical datasets and the results showed the effectiveness of the method.
Peng SONG Shifeng OU Zhenbin DU Yanyan GUO Wenming MA Jinglei LIU Wenming ZHENG
As a hot topic of speech signal processing, speech emotion recognition methods have been developed rapidly in recent years. Some satisfactory results have been achieved. However, it should be noted that most of these methods are trained and evaluated on the same corpus. In reality, the training data and testing data are often collected from different corpora, and the feature distributions of different datasets often follow different distributions. These discrepancies will greatly affect the recognition performance. To tackle this problem, a novel corpus-invariant discriminant feature representation algorithm, called transfer discriminant analysis (TDA), is presented for speech emotion recognition. The basic idea of TDA is to integrate the kernel LDA algorithm and the similarity measurement of distributions into one objective function. Experimental results under the cross-corpus conditions show that our proposed method can significantly improve the recognition rates.
Peng SONG Shifeng OU Xinran ZHANG Yun JIN Wenming ZHENG Jinglei LIU Yanwei YU
In practice, emotional speech utterances are often collected from different devices or conditions, which will lead to discrepancy between the training and testing data, resulting in sharp decrease of recognition rates. To solve this problem, in this letter, a novel transfer semi-supervised non-negative matrix factorization (TSNMF) method is presented. A semi-supervised negative matrix factorization algorithm, utilizing both labeled source and unlabeled target data, is adopted to learn common feature representations. Meanwhile, the maximum mean discrepancy (MMD) as a similarity measurement is employed to reduce the distance between the feature distributions of two databases. Finally, the TSNMF algorithm, which optimizes the SNMF and MMD functions together, is proposed to obtain robust feature representations across databases. Extensive experiments demonstrate that in comparison to the state-of-the-art approaches, our proposed method can significantly improve the cross-corpus recognition rates.
With the emergence of a large quantity of data in science and industry, it is urgent to improve the prediction accuracy and reduce the high complexity of Gaussian process regression (GPR). However, the traditional global approximation and local approximation have corresponding shortcomings, such as global approximation tends to ignore local features, and local approximation has the problem of over-fitting. In order to solve these problems, a large-scale Gaussian process regression algorithm (RFFLT) combining random Fourier features (RFF) and local approximation is proposed. 1) In order to speed up the training time, we use the random Fourier feature map input data mapped to the random low-dimensional feature space for processing. The main innovation of the algorithm is to design features by using existing fast linear processing methods, so that the inner product of the transformed data is approximately equal to the inner product in the feature space of the shift invariant kernel specified by the user. 2) The generalized robust Bayesian committee machine (GRBCM) based on Tsallis mutual information method is used in local approximation, which enhances the flexibility of the model and generates a sparse representation of the expert weight distribution compared with previous work. The algorithm RFFLT was tested on six real data sets, which greatly shortened the time of regression prediction and improved the prediction accuracy.
Xue ZHANG Anhong WANG Bing ZENG Lei LIU Zhuo LIU
Numerous examples in image processing have demonstrated that human visual perception can be exploited to improve processing performance. This paper presents another showcase in which some visual information is employed to guide adaptive block-wise compressive sensing (ABCS) for image data, i.e., a varying CS-sampling rate is applied on different blocks according to the visual contents in each block. To this end, we propose a visual analysis based on the discrete cosine transform (DCT) coefficients of each block reconstructed at the decoder side. The analysis result is sent back to the CS encoder, stage-by-stage via a feedback channel, so that we can decide which blocks should be further CS-sampled and what is the extra sampling rate. In this way, we can perform multiple passes of reconstruction to improve the quality progressively. Simulation results show that our scheme leads to a significant improvement over the existing ones with a fixed sampling rate.
Lei LIU Takehiro TSURITANI Ramon CASELLAS Ricardo MARTÍNEZ Raül MUÑOZ Munefumi TSURUSAWA Itsuro MORITA
A translucent wavelength switched optical network (WSON) is a cost-efficient infrastructure between opaque networks and transparent optical networks, which aims at seeking a graceful balance between network cost and service provisioning performance. In this paper, we experimentally present a resilient translucent WSON with the control of an enhanced path computation element (PCE) and extended generalized multi-protocol label switching (GMPLS) controllers. An adaptive routing and wavelength assignment scheme with the consideration of accumulated physical impairments, wavelength availabilities and regenerator allocation is experimentally demonstrated and evaluated for dynamic provisioning of lightpaths. By using two different network scenarios, we experimentally verify the feasibility of the proposed solutions in support of translucent WSON, and quantitatively evaluate the path computation latency, network blocking probability and service disruption time during end-to-end lightpath restoration. We also deeply analyze the experimental results and discuss the synchronization between the PCE and the network status. To the best of our knowledge, the most significant progress and contribution of this paper is that, for the first time, all the proposed methodologies in support of PCE/GMPLS controlled translucent WSON, including protocol extensions and related algorithms, are implemented in a network testbed and experimentally evaluated in detail, which allows verifying their feasibility and effectiveness when being potentially deployed into real translucent WSON.
Xiaolei LIU Xiaosong ZHANG Yiqi JIANG Qingxin ZHU
Optimizating the deployment of wireless sensor networks, which is one of the key issues in wireless sensor networks research, helps improve the coverage of the networks and the system reliability. In this paper, we propose an evolutionary algorithm based on modified t-distribution for the wireless sensor by introducing a deployment optimization operator and an intelligent allocation operator. A directed perturbation operator is applied to the algorithm to guide the evolution of the node deployment and to speed up the convergence. In addition, with a new geometric sensor detection model instead of the old probability model, the computing speed is increased by 20 times. The simulation results show that when this algorithm is utilized in the actual scene, it can get the minimum number of nodes and the optimal deployment quickly and effectively.Compared with the existing mainstream swarm intelligence algorithms, this method has satisfied the need for convergence speed and better coverage, which is closer to the theoretical coverage value.