Dandan WANG Qingcai CHEN Xiaolong WANG
Text Categorization (TC) is a task of classifying a set of documents into one or more predefined categories. Centroid-based method, a very popular TC method, aims to make classifiers simple and efficient by constructing one prototype vector for each class. It classifies a document into the class that owns the prototype vector nearest to the document. Many studies have been done on constructing prototype vectors. However, the basic philosophies of these methods are quite different from each other. It makes the comparison and selection of centroid-based TC methods very difficult. It also makes the further development of centroid-based TC methods more challenging. In this paper, based on the observation of its general procedure, the centroid-based text classification is treated as a kind of ranking task, and a unified framework for centroid-based TC methods is proposed. The goal of this unified framework is to classify a text via ranking all possible classes by document-class similarities. Prototype vectors are constructed based on various loss functions for ranking classes. Under this framework, three popular centroid-based methods: Rocchio, Hypothesis Margin Centroid and DragPushing are unified and their details are discussed. A novel centroid-based TC method called SLRCM that uses a smoothing ranking loss function is further proposed. Experiments conducted on several standard databases show that the proposed SLRCM method outperforms the compared centroid-based methods and reaches the same performance as the state-of-the-art TC methods.
Kosuke MARUYAMA Hiroshi KAMEDA
A ghost reduction algorithm for multiple angle sensors tracking objects under dual hypotheses is proposed. When multiple sensors and multiple objects exist on the same plane, the conventional method is unable to distinguish the real objects and ghosts from all possible pairs of measurement angle vectors. In order to resolve the issue stated above, the proposed algorithm utilizes tracking process considering dual hypotheses of real objects and ghosts behaviors. The proposed algorithm predicts dynamics of all the intersections of measurement angle vector pairs with the hypotheses of real objects and ghosts. Each hypothesis is evaluated by the residuals between prediction data and intersection. The appropriate hypothesis is extracted trough several data sampling. Representative simulation results demonstrate the effectiveness of the proposed algorithm.
Hongqing ZHU Meiyu DING Daqi GAO
The nth partial sums of a classical Fourier series have large oscillations near the jump discontinuities. This behaviour is the well-known Gibbs phenomenon. Recently, the inverse polynomial reconstruction method (IPRM) has been successfully implemented to reconstruct piecewise smooth functions by reducing the effects of the Gibbs phenomenon for Fourier series. This paper addresses the 2-D fractional Fourier series (FrFS) using the same approach used with the 1-D fractional Fourier series and finds that the Gibbs phenomenon will be observed in 1-D and 2-D fractional Fourier series expansions for functions at a jump discontinuity. The existing IPRM for resolution of the Gibbs phenomenon for 1-D and 2-D FrFS appears to be the same as that used for Fourier series. The proof of convergence provides theoretical basis for both 1-D and 2-D IPRM to remove Gibbs phenomenon. Several numerical examples are investigated. The results indicate that the IPRM method completely eliminates the Gibbs phenomenon and gives exact reconstruction results.
In this paper, one new class of quaternary generalized cyclotomic sequences with the period 2pq over F4 is established. The linear complexity of proposed sequences with the period 2pq is determined. The results show that such sequences have high linear complexity.
We analyze the effect of the propagation of route request packets in ad hoc network routing protocols such as DSR and AODV. So far it has not been clear how the number density of route request packets depends on propagation and hop counts. By stochastic analysis, it is found that the collisions of route request packets can be avoided efficiently by adjusting the number of the relevant nodes in the early stages of propagation.
Yuan TAO Yangdong DENG Shuai MU Zhenzhong ZHANG Mingfa ZHU Limin XIAO Li RUAN
The sparse matrix operation, y ← y+AtAx, where A is a sparse matrix and x and y are dense vectors, is a widely used computing pattern in High Performance Computing (HPC) applications. The pattern poses challenge to efficient solutions because both a matrix and its transposed version are involved. An efficient sparse matrix format, Compressed Sparse Blocks (CSB), has been proposed to provide nearly the same performance for both Ax and Atx. We develop a multithreaded implementation for the CSB format and apply it to solve y ← y+AtAx. Experiments show that our technique outperforms the Compressed Sparse Row (CSR) based solution in POSKI by up to 2.5 fold on over 70% of benchmarking matrices.
Moving objects or more generally foreground objects are the simplest objects in the field of computer vision after the pixel. Indeed, a moving object can be defined by 4 integers only, either two pairs of coordinates or a pair of coordinates and the size. In fixed camera scenes, moving objects (or blobs) can be extracted quite easily but the methods to produce them are not able to tell if a blob corresponds to remaining background noise, a single target or if there is an occlusion between many target which are too close together thus creating a single blob resulting from the fusion of all targets. In this paper we propose an novel method to refine moving object detection results in order to get as many blobs as targets on the scene by using a tracking system for additional information. Knowing if a blob is at proximity of a tracker allows us to remove noise blobs, keep the rest and handle occlusions when there are more than one tracker on a blob. The results show that the refinement is an efficient tool to sort good blobs from noise blobs and accurate enough to perform a tracking based on moving objects. The tracking process is a resolution free system able to reach speed such as 20 000fps even for UHDTV sequences. The refinement process itself is in real time, running at more than 2000fps in difficult situations. Different tests are presented to show the efficiency of the noise removal and the reality of the independence of the refinement tracking system from the resolution of the videos.
Shunsuke YAMAKI Masahide ABE Masayuki KAWAMATA
This paper derives the balanced realizations of second-order analog filters directly from the transfer function. Second-order analog filters are categorized into the following three cases: complex conjugate poles, distinct real poles, and multiple real poles. For each case, simple formulas are derived for the synthesis of the balanced realizations of second-order analog filters. As a result, we obtain closed form expressions of the balanced realizations of second-order analog filters.
Fereidoun H. PANAHI Tomoaki OHTSUKI
In a cognitive radio (CR) network, the channel sensing scheme used to detect the existence of a primary user (PU) directly affects the performances of both CR and PU. However, in practical systems, the CR is prone to sensing errors due to the inefficiency of the sensing scheme. This may yield primary user interference and low system performance. In this paper, we present a learning-based scheme for channel sensing in CR networks. Specifically, we formulate the channel sensing problem as a partially observable Markov decision process (POMDP), where the most likely channel state is derived by a learning process called Fuzzy Q-Learning (FQL). The optimal policy is derived by solving the problem. Simulation results show the effectiveness and efficiency of our proposed scheme.
Jianqiao WANG Yuehua LI Jianfei CHEN
Observed samples in wideband radar are always represented as nonlinear points in high dimensional space. In this paper, we consider the feature selection problem in the scenario of wideband radar target clustering. Inspired by manifold learning, we propose a novel feature selection algorithm, called Local Reconstruction Error Alignment (LREA), to select the features that can best preserve the underlying manifold structure. We first select the features that minimize the reconstruction error in every neighborhood. Then, we apply the alignment technique to extend the local optimal feature sequence to a global unique feature sequence. Experiments demonstrate the effectiveness of our proposed method.
Ruyuan ZHANG Yafeng ZHAN Yukui PEI Jianhua LU
Cooperative spectrum sensing is an effective approach that utilizes spatial diversity gain to improve detection performance. Most studies assume that the background noise is exactly known. However, this is not realistic because of noise uncertainty which will significantly degrade the performance. A novel weighted hard combination algorithm with two thresholds is proposed by dividing the whole range of the local test statistic into three regions called the presence, uncertainty and absence regions, instead of the conventional two regions. The final decision is made by weighted combination at the common receiver. The key innovation is the full utilization of the information contained in the uncertainty region. It is worth pointing out that the weight coefficient and the local target false alarm probability, which determines the two thresholds, are also optimized to minimize the total error rate. Numerical results show this algorithm can significantly improve the detection performance, and is more robust to noise uncertainty than the existing algorithms. Furthermore, the performance of this algorithm is not sensitive to the local target false alarm probability at low SNR. Under sufficiently high SNR condition, this algorithm reduces to the improved one-out-of-N rule. As noise uncertainty is unavoidable, this algorithm is highly practical.
Jun TAKEUCHI Akihiko HIRATA Hiroyuki TAKAHASHI Naoya KUKUTSU Yoshiaki YAMADA Kei KITAMURA Mitsuhiro TESHIMA
This paper presents 10-Gbit/s bidirectional and 20-Gbit/s unidirectional wireless data transmission systems using 120-GHz-band finline orthomode transducers (OMTs). A new finline OMT was fabricated with two improved designs, to adapt it to the data transmission characteristics of the 120-GHz-band wireless link. One improvement is higher isolation between orthogonal ports and the other is lower group delay variation. The measured isolation is more than 59dB at the carrier frequency of the 120-GHz-band wireless link, and the measured group delay variation is 43ps. Using the finline OMT, we developed 10-Gbit/s bidirectional and 20-Gbit/s unidirectional wireless equipment that can transmit two channels of 10-Gbit/s data using polarization multiplexing. With this wireless equipment, we succeeded in 10-Gbit/s bidirectional and 20-Gbit/s unidirectional wireless data transmission, which leads to successful seamless connection to 10 Gigabit Ethernet and 12-ch high definition television signal transmission.
Xiangxue LI Qingji ZHENG Haifeng QIAN Dong ZHENG Kefei CHEN
Given specified parameters, the number of check nodes, the expected girth and the variable node degrees, the Progressive Weight-Growth (PWG) algorithm is proposed to generate high rate low-density parity-check (LDPC) codes. Based on the theoretic foundation that is to investigate the girth impact by adding/removing variable nodes and edges of the Tanner graph, the PWG progressively increases column weights of the parity check matrix without violating the constraints defined by the given parameters. The analysis of the computational complexity and the simulation of code performance show that the LDPC codes by the PWG provide better or comparable performance in comparison with LDPC codes by some well-known methods (e.g., Mackay's random constructions, the PEG algorithm, and the bit-filling algorithm).
Kanshiro KASHIKI I-Te LIN Tomoki SADA Toshihiko KOMINE Shingo WATANABE
This paper describes an analytical study of performance of a proposed signal detection scheme that will allow coexistence of an additional radio communication system (generally, secondary system) in the service area where the existing communication system (primary system) is operated. Its performance characteristics are derived by an analytical method based on stochastic theory, which is subsequently validated by software simulation. The main purpose of the detection scheme is to protect the primary system from the secondary system. In such a situation, the signals of the primary system and secondary system may be simultaneously received in the signal detector. One application of such a scheme is D-to-D (Device-to-Device) communication, whose system concept including the detection scheme is briefly introduced. For improved secondary signal detection, we propose the signal cancellation method of the primary system and the feature detection method of the secondary system signal. We evaluate the performance characteristics of the detection scheme in terms of “probability of correct detection”. We reveal that an undesired random component is produced in the feature detection procedure when two different signals are simultaneously received, which degrades the detection performance. Such undesired component is included in the analytical equations. We also clarify that the cancellation scheme improves the performance, when the power ratio of the primary signal to secondary signal is higher than 20-22dB.
Ha-Nguyen TRAN Yohannes D. ALEMSEGED Hiroshi HARADA
Spectrum sensing is one of the methods to identify available white spaces for secondary usage which was specified by the regulators. However, signal quality to be sensed can plunge to a very low signal-to-noise-ratio due to signal propagation and hence readings from individual sensors will be unreliable. Distributed sensing by the cooperation of multiple sensors is one way to cope with this problem because the diversity gain due to the combining effect of data captured at different position will assist in detecting signals that might otherwise not be detected by a single sensor. In effect, the probability of detection can be improved. We have implemented a distributed sensing system to evaluate the performance of different cooperative sensing algorithms. In this paper we describe our implementation and measurement experience which include the system design, specification of the system, measurement method, the issues and solutions. This paper also confirms the performance enhancement offered by distributed sensing algorithms, and describes several ideas for further enhancement of the sensing quality.
Tsuyoshi SHIMOMURA Teppei OYAMA Hiroyuki SEKI
Television white spaces (TVWS) are locally and/or temporally unused portions of TV bands. After TVWS regulations were passed in the USA, more and more regulators have been considering efficient use of TVWS. Under the condition that the primary user, i.e., terrestrial TV broadcasting system, is not interfered, various secondary users (SUs) may be deployed in TVWS. In Japan, the TVWS regulations started with broadcast-type SUs and small-area broadcasting systems, followed by voice radio. This paper aims to provide useful insights for more efficient utilization of TVWS as one of the options to meet the continuously increasing demand for wireless bandwidth. TVWS availability in Japan is analyzed using graphs and maps. As per the regulations in Japan, for TV broadcasting service, a protection contour is defined to be 51dBµV/m, while the interference contour for SU is defined to be 12.3dBµV/m. We estimate TVWS availability using these two regulation parameters and the minimum separation distances calculated on the basis of the ITU-R P.1546 propagation models. Moreover, we investigate and explain the effect of two important factors on TVWS availability. One is the measures to avoid adjacent channel interference, while the other is whether the SU has client devices with interference ranges beyond the interference area of the master device. Furthermore, possible options to increase available TVWS channels are discussed.
Nhan NGUYEN-THANH Anh T. PHAM Van-Tam NGUYEN
Designing a medium access control (MAC) protocol is a key for implementing any practical wireless network. In general, a MAC protocol is responsible for coordinating users in accessing spectrum resources. Given that a user in cognitive radio(CR) networks do not have priority in accessing spectrum resources, MAC protocols have to perform dynamic spectrum access (DSA) functions, including spectrum sensing, spectrum access, spectrum allocation, spectrum sharing and spectrum mobility, beside conventional control procedure. As a result, designing MAC protocols for CR networks requires more complicated consideration than that needed for conventional/primary wireless network. In this paper, we focus on two major perspectives related to the design of a CR-MAC protocol: dynamic spectrum access functions and network infrastructure. Five DSA functions are reviewed from the point of view of MAC protocol design. In addition, some important factors related to the infrastructure of a CR network including network architecture, control channel management, the number of radios in the CR device and the number of transmission data channels are also discussed. The remaining challenges and open research issues are addressed for future research to aim at obtaining practical CR-MAC protocols.
Xiaoxue YU Yasushi YAMAO Motoharu MATSUURA
Radio over Fiber (RoF) is a promising technology that is suitable for broadband wireless access systems to cover in-building areas and outdoor dead-spots. However, one issue in RoF transmission that should be considered is the nonlinear distortion caused by Electrical/Optical (E/O) converters. Multicarrier RF (Radio Frequency) signal formats such as Orthogonal Frequency Division Multiplexing (OFDM), which are commonly employed in broadband wireless communications, are weak against nonlinearities. To enable the linear transmission of OFDM signal in RoF channel, we propose to employ the Envelop Pulse Width Modulation (EPWM) transmission scheme for RoF channel. Two commonly used E/O converters, Mach-Zehnder modulator and direct-modulation of Distributed Feedback Laser Diode (DFB LD), are employed to validate the proposal. Based on the measured nonlinearities of the E/O converters, they are mathematically modeled and their transmission performance are analyzed. A modified Rapp model is developed for the modeling of the DFB LD. Through simulations and experiments, the proposed scheme is shown to be effective in dealing with the nonlinearities of the E/O converters.
A covariance-based algorithm is proposed to find a barrage jammer suppression filter for surveillance radar with an adaptive array. The conventional adaptive beamformer (ABF) or adaptive sidelobe canceller (ASLC) with auxiliary antennas can be used successfully in sidelobe jammer rejection. When a jammer shares the same bearing with the target of interest, however, those methods inherently cancel the target in their attempt to null the jammer. By exploiting the jammer multipath scattered returns incident from other angles, the proposed algorithm uses only the auto-covariance matrix of the sample data produced by stacking range cell returns in a pulse repetition interval (PRI). It does not require estimation of direction of arrival (DOA) or time difference of arrival (TDOA) of multipath propagation, thus making it applicable to electronic countermeasure (ECM) environments with high power barrage jammers and it provides the victim radar with the ability to null both the sidelobe (sidebeam) and mainlobe (mainbeam) jammers simultaneously. Numeric simulations are provided to evaluate the performance of this filter in the presence of an intensive barrage jammer with jammer-to-signal ratio (JSR) greater than 30dB, and the achieved signal-to-jammer-plus-noise ratio (SJNR) improvement factor (IF) exceeds 46dB.
Van Hai DO Xiong XIAO Eng Siong CHNG Haizhou LI
This paper presents a novel acoustic modeling technique of large vocabulary automatic speech recognition for under-resourced languages by leveraging well-trained acoustic models of other languages (called source languages). The idea is to use source language acoustic model to score the acoustic features of the target language, and then map these scores to the posteriors of the target phones using a classifier. The target phone posteriors are then used for decoding in the usual way of hybrid acoustic modeling. The motivation of such a strategy is that human languages usually share similar phone sets and hence it may be easier to predict the target phone posteriors from the scores generated by source language acoustic models than to train from scratch an under-resourced language acoustic model. The proposed method is evaluated using on the Aurora-4 task with less than 1 hour of training data. Two types of source language acoustic models are considered, i.e. hybrid HMM/MLP and conventional HMM/GMM models. In addition, we also use triphone tied states in the mapping. Our experimental results show that by leveraging well trained Malay and Hungarian acoustic models, we achieved 9.0% word error rate (WER) given 55 minutes of English training data. This is close to the WER of 7.9% obtained by using the full 15 hours of training data and much better than the WER of 14.4% obtained by conventional acoustic modeling techniques with the same 55 minutes of training data.