NEC Laboratory

Beijing, China

NEC Laboratory

Beijing, China
Time filter
Source Type

Wang J.,Nanjing Southeast University | Zhu D.,Nanjing Southeast University | Zhao C.,Nanjing Southeast University | Li J.C.F.,NEC Laboratory | Lei M.,NEC Laboratory
IEEE Communications Letters | Year: 2013

The benefit of device-to-device (D2D) communication hinges on intelligent resource sharing between cellular and D2D users. This letter aims to optimize resource sharing for D2D communication to better utilize uplink resources in a multi-user cellular system with guaranteed quality of normal cellular communications. Despite the nonconvex difficulty, we provide an analytical characterization of the globally optimal resource sharing strategy, and furthermore propose two suboptimal strategies with less complexity. The superiority of the proposed resource sharing strategies is demonstrated through numerical examples. © 1997-2012 IEEE.

Wang J.,Nanjing Southeast University | Zhu D.,Nanjing Southeast University | Zhang H.,Nanjing Southeast University | Zhao C.,Nanjing Southeast University | And 2 more authors.
Signal Processing | Year: 2014

Cellular network assisted device-to-device (D2D) communication can improve spectrum utilization by jointly coordinating cellular and D2D users. This paper studies D2D communication sharing multiple cellular channels and optimizes the overall system performance by maximizing the weighted sum rate (WSR) of the cellular and D2D users. We first provide an analytical characterization of the optimal resource sharing in the single channel case. Then, based on it, we propose a simple but efficient algorithm to maximize the WSR in the general multichannel case. Furthermore, we also propose two alternative algorithms to achieve better performance at a cost of higher complexity. In addition, a simple condition is provided to verify the global optimality of the obtained solution. Numerical results show that the proposed algorithms outperform the existing single-channel design, and interestingly the optimality condition is satisfied in many cases, thus justifying the merit of the proposed algorithms from both theoretical and practical aspects. © 2014 Elsevier B.V.

Huang R.,NEC Laboratory | Cao Z.,CAS Institute of Automation
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

Optical inspection techniques have been widely used in industry as they are non-destructive, efficient to achieve, easy to process, and can provide rich information on product quality. Defect patterns such as rings, semi-circles, scratches, clusters are the most common defects in the semiconductor industry. Most methods cannot identify two scale-variant or shift-variant or rotation-variant defect patterns, which in fact belong to the same failure causes. To address these problems, a new approach has been proposed in this paper to detect these defect patterns in noisy images obtained from printed circuit boards, wafers, and etc. A median filter, background removal, morphological operation, segmentation and labeling are employed in the detection stage of our method. Support vector machine (SVM) is used to identify the defect patterns which are resized. Classification results of both simulated data and real noisy raw data show the effectiveness of our method. © 2013 Springer-Verlag.

Chen B.,Iowa State University | Zhang L.,Iowa State University | He Y.,NEC Laboratory
IEEE Power and Energy Society General Meeting | Year: 2015

Demand side management has attracted a lot of attention as a method to regulate customer behavior and improve system reliability. In this paper, we solve the day-ahead energy pricing problem in the distribution electricity system by taking into account the fact that customers can change their consumption behavior in response to price changes. We propose two pricing models under two different scenarios. In the first scenario, customers' consumption profiles at different prices are available data. We propose an integer programming model to maximize the revenue of the utility company. In the second scenario, we assume only the consumption profiles at the current price, which is static, are available. A game theoretic bilevel optimization model is built to describe the relationship between electricity price and customers' behavior. We then compare and analyze the results of the two models. © 2015 IEEE.

Zhuang H.,Huawei | Zhang J.,Huawei | Yu G.,Zhejiang University | Wang C.,NEC Laboratory | Daneshmand A.,AT and T Laboratory
Proceedings - IEEE Symposium on Computers and Communications | Year: 2010

Relay Stations (RSs) selection and spectrum allocation is a very important problem in cognitive Multi-hop Cellular Networks (MCN). In this paper, we propose a practical joint RS selection and spectrum allocation scheme to maximize the system capacity while fairness is guaranteed. In the proposed scheme, Mobile Stations (MS) are adaptively and optimally selected to be RS based on spatiality of spectrum sets and multi-hop links. Further, we simplify the optimal scheme and propose the RS non-deterministic heuristic algorithms and then compare with conventional RS predetermined algorithm in which RS are predefined before joint RS selection and resource allocation. Simulation results show that the RS nondeterministic algorithm outperforms the conventional RS predetermined algorithm in terms of spectrum efficiency. © 2010 IEEE.

Djordjevic I.B.,University of Arizona | Jovanovic A.Z.,University of Niš | Peric Z.H.,University of Niš | Wang T.,NEC Laboratory
Optics InfoBase Conference Papers | Year: 2014

An optimized-vector-quantization-based signal-constellation-design (OVQ-SCD) suitable for multidimensional-optical-transport is proposed, in which signal-constellation-radiitransformation-function is optimized and near-uniform-distribution of points is achieved. The OVQ-SCDs significantly outperform corresponding counterparts and can be used to enable beyond 1Pb/s serial-optical-transport. © 2014 OSA.

Chen X.,Carnegie Mellon University | Qi Y.,NEC Laboratory | Bai B.,NEC Laboratory | Lin Q.,Carnegie Mellon University | Carbonell J.G.,Carnegie Mellon University
Proceedings of the 11th SIAM International Conference on Data Mining, SDM 2011 | Year: 2011

Latent semantic analysis (LSA), as one of the most popular unsupervised dimension reduction tools, has a wide range of applications in text mining and information retrieval. The key idea of LSA is to learn a projection matrix that maps the high dimensional vector space representations of documents to a lower dimensional latent space, i.e. so called latent topic space. In this paper, we propose a new model called Sparse LSA, which produces a sparse projection matrix via the ℓ1 regularization. Compared to the traditional LSA, Sparse LSA selects only a small number of relevant words for each topic and hence provides a compact representation of topic-word relationships. Moreover, Sparse LSA is computationally very efficient with much less memory usage for storing the projection matrix. Furthermore, we propose two important extensions of Sparse LSA: group structured Sparse LSA and non-negative Sparse LSA. We conduct experiments on several benchmark datasets and compare Sparse LSA and its extensions with several widely used methods, e.g. LSA, Sparse Coding and LDA. Empirical results suggest that Sparse LSA achieves similar performance gains to LSA, but is more efficient in projection computation, storage, and also well explain the topic-word relationships. Copyright © SIAM.

Loading NEC Laboratory collaborators
Loading NEC Laboratory collaborators