Tsinghua National Laboratory for Information Sciences and Technology

Beijing, China

Tsinghua National Laboratory for Information Sciences and Technology

Beijing, China
SEARCH FILTERS
Time filter
Source Type

Dai L.,Tsinghua National Laboratory for Information Sciences and Technology
IEEE Transactions on Wireless Communications | Year: 2017

Millimeter-wave massive MIMO with lens antenna array can considerably reduce the number of required radiofrequency (RF) chains by beam selection. However, beam selection requires the base station to acquire the accurate information of beamspace channel. This is a challenging task, as the size of beamspace channel is large while the number of RF chains is limited. In this paper, we investigate the beamspace channel estimation problem in mmWave massive MIMO systems with lens antenna array. Specifically, we first design an adaptive selecting network for mmWave massive MIMO systems with lens antenna array, and based on this network, we further formulate the beamspace channel estimation problem as a sparse signal recovery problem. Then, by fully utilizing the structural characteristics of mmWave beamspace channel, we propose a support detection (SD)-based channel estimation scheme with reliable performance and low pilot overhead. Finally, the performance and complexity analyses are provided to prove that the proposed SD-based channel estimation scheme can estimate the support of sparse beamspace channel with comparable or higher accuracy than conventional schemes. Simulation results verify that the proposed SD-based channel estimation scheme outperforms conventional schemes and enjoys satisfying accuracy, even in the low SNR region as the structural characteristics of beamspace channel can be exploited. IEEE


Lu W.,Tsinghua National Laboratory for Information Sciences and Technology
Geophysics | Year: 2013

I have developed an accelerated sparse time-invariant Radon transform (RT) in the mixed frequency-time domain based on iterative 2D model shrinkage in the time domain. I denote it as SRTIS. In the traditional sparse time-invariant RT in the mixed frequency-time domain, the sparse RT is modeled as a sparse inverse problem that is solved by the iteratively reweighted least-squares (IRLS) algorithm in the time domain, and the forward and inverse RTs are implemented in the frequency domain. In this method, IRLS is replaced by iterative 2D model shrinkage, i.e., the sparsity of the Radon model is promoted by some simple 2D model shrinkage operations in the time domain. Synthetic and real data demultiple examples using the parabolic RTs are given to demonstrate the better performance of the SRTIS when compared with the least-squares-based RT, the frequency domain sparse RT, and the traditional time-domain sparse RT in the mixed frequency-time domain. © 2013 Society of Exploration Geophysicists.


Zhou L.,Tsinghua National Laboratory for Information Sciences and Technology
IEEE Sensors Journal | Year: 2014

In this paper, we present a new minimal solution for the extrinsic calibration of a 2D light detection and ranging (LIDAR) sensor and a perspective camera. This problem is formulated as registering three planes and the corresponding coplanar lines. All existing algorithms solve this problem by its geometric structure in the original or the dual 3D space. In contrast, our algorithm directly exploits the algebraic structure of the polynomial system to resolve this problem. This new algorithm is more abstract, however, and provides a more broadly applicable method to other problems that need to handle the similar polynomial system. The rotation matrix is estimated first. Then, the translation vector can be calculated by solving a system of three linear equations. Although the new approach is conceptually simple, it has 720 different versions caused by different permutations of variables. This results in different computational orders and affects the numerical behavior of the algorithm. A simple heuristic scheme is proposed to select the permutation of variables that yields numerically stable computational order with respect to the given input. Simulation and experimental results show that the proposed algorithm outperforms the existing state-of-the-art algorithms in terms of accuracy and numerical stability. © 2013 IEEE.


Wang Y.-X.,Tsinghua National Laboratory for Information Sciences and Technology | Zhang Y.-J.,Tsinghua National Laboratory for Information Sciences and Technology
IEEE Transactions on Knowledge and Data Engineering | Year: 2013

Nonnegative Matrix Factorization (NMF), a relatively novel paradigm for dimensionality reduction, has been in the ascendant since its inception. It incorporates the nonnegativity constraint and thus obtains the parts-based representation as well as enhancing the interpretability of the issue correspondingly. This survey paper mainly focuses on the theoretical research into NMF over the last 5 years, where the principles, basic models, properties, and algorithms of NMF along with its various modifications, extensions, and generalizations are summarized systematically. The existing NMF algorithms are divided into four categories: Basic NMF (BNMF), Constrained NMF (CNMF), Structured NMF (SNMF), and Generalized NMF (GNMF), upon which the design principles, characteristics, problems, relationships, and evolution of these algorithms are presented and analyzed comprehensively. Some related work not on NMF that NMF should learn from or has connections with is involved too. Moreover, some open issues remained to be solved are discussed. Several relevant application areas of NMF are also briefly described. This survey aims to construct an integrated, state-of-the-art framework for NMF concept, from which the follow-up research may benefit.©2013 IEEE.


Wang L.,Tsinghua National Laboratory for Information Sciences and Technology | Li L.-P.,Tsinghua National Laboratory for Information Sciences and Technology
International Journal of Electrical Power and Energy Systems | Year: 2013

Solving the non-convex economic load dispatch (ELD) problem with evolutionary algorithms has gained increasing research in recent years. In this paper, a differential harmony search (DHS) algorithm is proposed by combining the mechanisms of both differential evolution and harmony search. In the DHS, the pitch adjustment operation is cooperated with the different mutation operation to enhance the exploitation ability of harmony search, and both the memory consideration and the pitch adjustment are used to enhance the exploration ability of evolution search. In addition, a repair procedure and three simple selection rules are proposed for constraint handling. Numerical simulations are carried out based on different kinds of testing problems with various constraints including valve point effects, multi-fuels, ramp rate limit and prohibited operation zones. Simulation results and comparisons with the some existing algorithms demonstrate the effectiveness, efficiency and robustness of the proposed DHS algorithm. Finally, the effect of parameter setting on the DHS is investigated as well. © 2012 Elsevier Ltd. All rights reserved.


Zheng B.,Tsinghua National Laboratory for Information Sciences and Technology | Zhong Y.,Tsinghua National Laboratory for Information Sciences and Technology
IEEE Transactions on Industrial Electronics | Year: 2011

This paper presents the design and experimental results of the robust attitude regulation for a 3-DOF Quanser helicopter benchmark. The control objective is to asymptotically track the desired elevation and pitch reference signals generated by an exogenous system, while keeping all the closed-loop states bounded, even if there exist unknown uncertainties. Based on the output-regulation theory and signal compensation technique, a new design method of state feedback robust controller is proposed. This method permits the existence of unknown uncertainties and external disturbances in both the helicopter and the exosystem. Only knowledge about the bounds of these uncertainties is required. The experimental results demonstrate the desired performance of the proposed controller implemented on the Quanser helicopter. © 2011 IEEE.


Fang C.,Tsinghua National Laboratory for Information Sciences and Technology | Wang L.,Tsinghua National Laboratory for Information Sciences and Technology
Computers and Operations Research | Year: 2012

In this paper, we propose an effective heuristic based on the framework of the shuffled frog-leaping algorithm (SFLA) for solving the resource-constrained project scheduling problem (RCPSP). We encode the virtual frog as the extended activity list (EAL) and decode it by the SFLA-specific serial schedule generation scheme (SSSGS). The initial population is generated by the regret-based sampling method and the priority rule. Then, virtual frogs are partitioned into several memeplexes, and each memeplex evolves by adopting the effective resource-based crossover (RBCO). To enhance the exploitation ability, a combined local search including permutation-based local search (PBLS) and forwardbackward improvement (FBI) is performed in each memeplex. To maintain the diversity of each memeplex, virtual frogs are periodically shuffled and reorganized into new memeplexes. Basing on some theoretical analysis, speed-up evaluation methods are proposed to improve the efficiency of the SFLA, which are also suitable for other heuristics designed for RCPSP. In addition, we make use of a design-of-experiment method to determine the set of suitable parameters for the SFLA. Computational results and comparisons with some typical existing algorithms demonstrate the effectiveness of the proposed SFLA. © 2011 Elsevier Ltd. All rights reserved.


Wang L.,Tsinghua National Laboratory for Information Sciences and Technology | Fang C.,Tsinghua National Laboratory for Information Sciences and Technology
Computers and Operations Research | Year: 2012

In this paper, an estimation of distribution algorithm (EDA) is proposed to solve the multi-mode resource-constrained project scheduling problem (MRCPSP). In the EDA, the individuals are encoded based on the activity-mode list (AML) and decoded by the multi-mode serial schedule generation scheme (MSSGS), and a novel probability model and an updating mechanism are proposed for well sampling the promising searching region. To further improve the searching quality, a multi-mode forward backward iteration (MFBI) and a multi-mode permutation based local search method (MPBLS) are proposed and incorporated into the EDA based search framework to enhance the exploitation ability. Based on the design-of-experiment (DOE) test, suitable parameter combinations are determined and some guidelines are provided to set the parameters. Simulation results based on a set of benchmarks and comparisons with some existing algorithms demonstrate the effectiveness of the proposed EDA. © 2011 Elsevier Ltd.


Dai L.,Tsinghua National Laboratory for Information Sciences and Technology | Wang Z.,Tsinghua National Laboratory for Information Sciences and Technology | Yang Z.,Tsinghua National Laboratory for Information Sciences and Technology
IEEE Journal on Selected Areas in Communications | Year: 2013

Large-scale orthogonal frequency division multiplexing (OFDM) multiple-input multiple-output (MIMO) is a promising candidate to achieve the spectral efficiency up to several tens of bps/Hz for future wireless communications. One key challenge to realize practical large-scale OFDM MIMO systems is high-dimensional channel estimation in mobile multipath channels. In this paper, we propose the time-frequency training OFDM (TFT-OFDM) transmission scheme for large-scale MIMO systems, where each TFT-OFDM symbol without cyclic prefix adopts the time-domain training sequence (TS) and the frequency-domain orthogonal grouped pilots as the time-frequency training information. At the receiver, the corresponding time-frequency joint channel estimation method is proposed to accurately track the channel variation, whereby the received time-domain TS is used for path delays estimation without interference cancellation, while the path gains are acquired by the frequency-domain pilots. The channel property that path delays vary much slower than path gains is further exploited to improve the estimation performance, and the sparse nature of wireless channel is utilized to acquire the path gains by very few pilots. We also derive the theoretical Cramér-Rao lower bound (CRLB) of the proposed channel estimator. Compared with conventional large-scale OFDM MIMO systems, the proposed TFT-OFDM MIMO scheme achieves higher spectral efficiency as well as the coded bit error rate performance close to the ergodic channel capacity in mobile environments. © 2012 IEEE.


Liu Y.-J.,Tsinghua National Laboratory for Information Sciences and Technology
CAD Computer Aided Design | Year: 2013

A natural metric in 2-manifold surfaces is to use geodesic distance. If a 2-manifold surface is represented by a triangle mesh T, the geodesic metric on T can be computed exactly using computational geometry methods. Previous work for establishing the geodesic metric on T only supports using half-edge data structures; i.e., each edge e in T is split into two halves (h e1,he2) and each half-edge corresponds to one of two faces incident to e. In this paper, we prove that the exact-geodesic structures on two half-edges of e can be merged into one structure associated with e. Four merits are achieved based on the properties which are studied in this paper: (1) Existing CAD systems that use edge-based data structures can directly add the geodesic distance function without changing the kernel to a half-edge data structure; (2) To find the geodesic path from inquiry points to the source, the MMP algorithm can be run in an on-the-fly fashion such that the inquiry points are covered by correct wedges; (3) The MMP algorithm is sped up by pruning unnecessary wedges during the wedge propagation process; (4) The storage of the MMP algorithm is reduced since fewer wedges need to be stored in an edge-based data structure. Experimental results show that when compared to the classic half-edge data structure, the edge-based implementation of the MMP algorithm reduces running time by 44% and storage by 29% on average. © 2012 Elsevier Ltd. All rights reserved.

Loading Tsinghua National Laboratory for Information Sciences and Technology collaborators
Loading Tsinghua National Laboratory for Information Sciences and Technology collaborators