1-8hit |
Chyi-Ren DOW Jyh-Horng LIN Shiow-Fen HWANG Yi-Wen WANG
In ad-hoc mobile radio networks, nodes are organized into non-overlapping clusters. These clusters are independently controlled and dynamically reconfigured when the topology changes. This work presents a Distributed Label clustering scheme (DL) that partitions nodes into clusters using a weight-based criterion. The DL scheme allows the border nodes to determine their roles first to avoid selecting unsuitable clusterheads. In order to resolve the clusterhead change problem, the DL scheme restricts the number of clusterhead changes. The DL scheme also restricts the size of the virtual backbone by reducing the number of clusters. This scheme is distributed and can be executed at each node with only the knowledge of one-hop neighbors. The simulation results demonstrate that our scheme outperforms other clustering schemes in terms of the number of clusters, stability of the clusters and control overhead when the topology changes.
Chi-Min LI Shen-Wen WANG Pao-Jen WANG
Given the rapid development of current wireless communication systems has led to two major challenges: energy conservation and interference avoidance. Addressing these challenges is critical for sustaining modern green communications. This paper proposes two energy-efficient schemes for a heterogeneous network environment. The schemes include a cell switching strategy and a power control technique. The proposed schemes can save energy while maintaining the service quality for users. Simulation results showed that compared with conventional schemes, the proposed schemes reduced energy consumption by up to 18% more and further enhanced the system energy efficiency by up to 22% without using any switch-off procedure.
Xiaoting WANG Yiwen WANG Shichao LI Ping LI
The crossbar-based switch fabric is widely used in today's high performance switches, due to its internally nonblocking and simply implementation properties. Usually there are two main switching architectures for crossbar-based switch fabric: internally bufferless crossbar switch and crosspoint buffered crossbar switch. As internally bufferless crossbar switch requires a complex centralized scheduler which limits its scalability to high speeds, crosspoint buffered crossbar switch has gained more attention because of its simpler distributed scheduling algorithm and better switching performance. However, almost all the scheduling algorithms proposed previously for crosspoint buffered crossbar switch either have unsatisfactory scheduling performance under non-uniform traffic patterns or show poor service fairness between input traffic flows. In order to overcome the disadvantages of existing algorithms, in this paper we propose two novel high performance scheduling algorithms named MCQF_RR and IMCQF_RR for crosspoint buffered crossbar switches. Both algorithms have a time complexity of O(log N), where N is the number of input/output ports of the switch. MCQF_RR takes advantage of the combined weight information about queue length and service waiting time of input queues to perform scheduling. In order to further reduce the scheduling complexity and make it feasible for high speed switches, IMCQF_RR uses the compressed queue length information instead of original queue length information to schedule cells in input VOQs. Simulation results show that our novel scheduling algorithms MCQF_RR and IMCQF_RR can demonstrate excellent delay performance comparable to existing high performance scheduling algorithms under both uniform and non-uniform traffic patterns, while maintain good service fairness performance under severe non-uniform traffic patterns.
Zi-wen WANG Guo-rui FENG Ling-yan FAN Jin-wei WANG
The sparse representation models have been widely applied in image super-resolution. The certain optimization problem is supposed and can be solved by the iterative shrinkage algorithm. During iteration, the update of dictionaries and similar patches is necessary to obtain prior knowledge to better solve such ill-conditioned problem as image super-resolution. However, both the processes of iteration and update often spend a lot of time, which will be a bottleneck in practice. To solve it, in this paper, we present the concept of image quality difference based on generalized Gaussian distribution feature which has the same trend with the variation of Peak Signal to Noise Ratio (PSNR), and we update dictionaries or similar patches from the termination strategy according to the adaptive threshold of the image quality difference. Based on this point, we present two sparse representation algorithms for image super-resolution, one achieves the further improvement in image quality and the other decreases running time on the basis of image quality assurance. Experimental results also show that our quantitative results on several test datasets are in line with exceptions.
Fashen LI Jianrong SUN Xuewen WANG Jianbo WANG
Mn1-xZnxFe2O4 thin films with various Zn contents, 300 nm in thickness, were synthesized on glass substrates directly by electroless plating in aqueous solution at 90 without a heat treatment. With XRD, SEM, VSM, the crystallographic structure, morphology of the films and the macroscopic magnetic properties were characterized. The Mn-Zn ferrite films have a single phase spinel structure and well-crystallized columnar grains grow perpendicularly to the substrate. The change of the coercivity is not consistent with that of the bulk materials. As the Zn content in the films increases, the value of Hc decreases firstly, and then increases. At x=0.5, the minimum value of Hc is 3.7 kA/m and the value of Ms is 419.6 kA/m. The hyperfine magnetic fields, cation occupations and the distribution of the magnetic moments in film plane were studied by the conversion electron Mossbauer spectroscopy (CEMS).
Maoxi LI Qingyu XIANG Zhiming CHEN Mingwen WANG
The-state-of-the-art neural quality estimation (QE) of machine translation model consists of two sub-networks that are tuned separately, a bidirectional recurrent neural network (RNN) encoder-decoder trained for neural machine translation, called the predictor, and an RNN trained for sentence-level QE tasks, called the estimator. We propose to combine the two sub-networks into a whole neural network, called the unified neural network. When training, the bidirectional RNN encoder-decoder are initialized and pre-trained with the bilingual parallel corpus, and then, the networks are trained jointly to minimize the mean absolute error over the QE training samples. Compared with the predictor and estimator approach, the use of a unified neural network helps to train the parameters of the neural networks that are more suitable for the QE task. Experimental results on the benchmark data set of the WMT17 sentence-level QE shared task show that the proposed unified neural network approach consistently outperforms the predictor and estimator approach and significantly outperforms the other baseline QE approaches.
Mengbo ZHANG Lunwen WANG Yanqing FENG Haibo YIN
Spectrum sensing is the first task performed by cognitive radio (CR) networks. In this paper we propose a spectrum sensing algorithm for orthogonal frequency division multiplex (OFDM) signal based on deep learning and covariance matrix graph. The advantage of deep learning in image processing is applied to the spectrum sensing of OFDM signals. We start by building the spectrum sensing model of OFDM signal, and then analyze structural characteristics of covariance matrix (CM). Once CM has been normalized and transformed into a gray level representation, the gray scale map of covariance matrix (GSM-CM) is established. Then, the convolutional neural network (CNN) is designed based on the LeNet-5 network, which is used to learn the training data to obtain more abstract features hierarchically. Finally, the test data is input into the trained spectrum sensing network model, based on which spectrum sensing of OFDM signals is completed. Simulation results show that this method can complete the spectrum sensing task by taking advantage of the GSM-CM model, which has better spectrum sensing performance for OFDM signals under low SNR than existing methods.
Bo LIU Yi-Jun MAN Zhi-Min YUAN Lei ZHU Ji-Wen WANG
Future high density magnetic recording requires a nanometer spaced head-slider interface, high track seeking velocity and high spindle speed. Such a combination greatly increases the likelihood of slider-disk and slider-particle-disk impact. Furthermore, the impact can generate high flash temperature and leads to data reliability problems, such as partial or full data erasure. This work report a method to conduct controlled experimental investigations into the possibility of such a data erasure even when the temperature is far below the Curie temperature. Results indicate that the high density magnetic transitions are of high likelihood of being affected by the flash temperature. Investigations also extended to micromagnetic modeling of the flash temperature effect. Results suggest that thermally induced local stress can play important roll in the data erasure process. Modeling results also exhibit that smaller grain size and higher recording density are also of higher likelihood of getting the transitions being affected by the flash temperature.