This paper describes two different types of GaAs-HBT compatible, base-collector diode 0/20-dB step attenuators--diode-linearizer type and harmonics-trap type--for 3.5-GHz-band wireless applications. The two attenuators use an AC-coupled, stacked type diode switch topology featuring high power handling capability with low bias current operation. Compared to a conventional diode switch topology, this topology can improve the capability of more than 6 dB with the same bias current. In addition, successful incorporation of a shunt diode linearizer and second- and third-harmonic traps into the attenuators gives the IM3 distortion improvement of more than 7 dB in the high power ranging from 16 dBm to 18 dBm even in the 20-dB attenuation mode when IM3 distortion levels are basically easy to degrade. Measurement results show that both the attenuators are capable of delivering power handling capability (P0.2 dB) of more than 18 dBm with IM3 levels of less than -35 dBc at an 18-dBm input power while drawing low bias currents of 3.8 mA and 6.8 mA in the thru and attenuation modes from 0/5-V complementary supplies. Measured insertion losses of the linearizer-type and harmonics-trap type attenuators in the thru mode are as low as 1.4 dB and 2.5 dB, respectively.

Yubazaki et al. have proposed "single input rule modules connected type fuzzy reasoning method" (SIRMs method, for short) whose final output is obtained by summarizing the product of the importance degrees and the inference results from single input fuzzy rule module. Another type of single input type fuzzy reasoning method proposed by Hayashi et al. (we call it "Single Input Connected fuzzy reasoning method" (SIC method, for short) in this paper) uses rule modules to each input item as well as SIRMs method. We expect that inference results of SIRMs method and SIC method have monotonicity if the antecedent parts and consequent parts of fuzzy rules in SIRMs rule modules have monotonicity. However, this paper points out that even if fuzzy rules in SIRMs rule modules have monotonicity, the inference results do not necessarily have monotonicity. Moreover, it clarifies the conditions for the monotonicity of inference results by SIRMs method and SIC method.

Estimating the generalization error is one of the key ingredients of supervised learning since a good generalization error estimator can be used for model selection. An unbiased generalization error estimator called the subspace information criterion (SIC) is shown to be useful for model selection, but its range of application is limited to linear learning methods. In this paper, we extend SIC to be applicable to non-linear learning.

In this paper, we propose a learning classifier based on maximum entropy (ME) for resolving zero-anaphora in Chinese text. Besides regular grammatical, lexical, positional and semantic features motivated by previous research on anaphora resolution, we develop two innovative Web-based features for extracting additional semantic information from the Web. The values of the two features can be obtained easily by querying the Web using some patterns. Our study shows that our machine learning approach is able to achieve an accuracy comparable to that of state-of-the-art systems. The Web as a knowledge source can be incorporated effectively into the ME learning framework and significantly improves the performance of our approach.

Repacking is an efficient scheme for bandwidth packing problem (BPP) in centralized networks (CNs), where a central unit allocates bandwidth to the rounding terminals. In this paper, we study its performance by proposing a new formulation of the BPP in the CN, and introducing repacking scheme into next fit algorithm in terms of the online constraint. For the realistic applications, the effect of call demand distribution is also exploited by means of simulation. The results show that the repacking efficiency is significant (e.g. the minimal improvement about 13% over uniform distribution), especially in the scenarios where the small call demands dominate the network.

We propose a new genetic fuzzy discretization method with feature selection for the pattern classification problems. Traditional discretization methods categorize a continuous attribute into a number of bins. Because they are made on crisp discretization, there exists considerable information loss. Fuzzy discretization allows overlapping intervals and reflects linguistic classification. However, the number of intervals, the boundaries of intervals, and the degrees of overlapping are intractable to get optimized and a discretization process increases the total amount of data being transformed. We use a genetic algorithm with feature selection not only to optimize these parameters but also to reduce the amount of transformed data by filtering the unconcerned attributes. Experimental results showed considerable improvement on the classification accuracy over a crisp discretization and a typical fuzzy discretization with feature selection.

We study the problem of optimizing admission control policies in mobile multimedia cellular networks when predictive information regarding movement is available and we evaluate the gains that can be achieved by making such predictive information available to the admission controller. We consider a general class of prediction agents which forecast the number of future handovers and we evaluate the impact on performance of aspects like: whether the prediction refers to incoming and/or outgoing handovers, inaccurate predictions, the anticipation of the prediction and the way that predictions referred to different service classes are aggregated. For the optimization process we propose a novel Reinforcement Learning approach based on the concept of afterstates. The proposed approach, when compared with conventional Reinforcement Learning, yields better solutions and with higher precision. Besides it tackles more efficiently the curse of dimensionality inherent to multimedia scenarios. Numerical results show that the performance gains measured are higher when more specific information is provided about the handover time instants, i.e. when the anticipation time is deterministic instead of stochastic. It is also shown that the utilization of the network is maintained at very high values, even when the highest improvements are observed. We also compare an optimal policy obtained deploying our approach with a previously proposed heuristic prediction scheme, showing that plenty of room for technological innovation exists.

In orthogonal frequency-division multiplex access (OFDMA) uplink, the carrier-frequency offsets (CFOs) between the multiple transmitters and the receiver introduce inter-carrier interference (ICI) and severely degrade the performance. In this paper, based on the perfect estimation of each user's CFO, we propose two low-complexity iterative algorithms to cancel ICI due to CFOs, which are denoted as the basic algorithm and the improved algorithm with decision-feedback equalization (DFE), respectively. For the basic one, two theorems are proposed that yield a sufficient condition for the convergence of iterations. Moreover, the interference-power-evolution (IPE) charts are proposed to evaluate the convergence behavior of this interference cancellation algorithm. Motivated by the IPE chart, the procedure of DFE is introduced into the iterations, which is the basic idea of the improved algorithm. For this improved algorithm, the error-propagation effect are analyzed and suppressed by an efficient stopping criterion. From IPE charts and simulation results, it can be easily observed that the basic algorithm has the same capability of ICI cancellation as the linear optimal minimum mean square error (MMSE) method, but offers lower complexity, while the improved algorithm with DFE outperforms the MMSE method in terms of the bit-error rate (BER) performance.

Frequency-domain equalization (FDE) based on minimum mean square error (MMSE) criterion can replace the conventional rake combining to significantly improve the bit error rate (BER) performance in a frequency-selective fading channel. MMSE-FDE requires an accurate estimate of the channel transfer function and the signal-to-noise power ratio (SNR). Direct application of pilot-assisted channel estimation (CE) degrades the BER performance, since the frequency spectrum of the pilot chip sequence is not constant over the spreading bandwidth. In this paper, we propose a pilot-assisted decision feedback frequency-domain MMSE-CE. The BER performance with the proposed pilot-assisted MMSE-CE in a frequency-selective Rayleigh fading channel is evaluated by computer simulation. It is shown that MMSE-CE always gives a good BER performance irrespective of the choice of the pilot chip sequence and shows a high tracking ability against fading. For a spreading factor SF of 16, the Eb/N0 degradation for BER=10-4 with MMSE-CE from the ideal CE case is as small as 0.9 dB (including an Eb/N0 loss of 0.28 dB due to the pilot insertion).

This paper proposes a co-channel interference (CCI) suppression scheme employing a frequency-domain nulling filter and turbo equalizer for single-carrier uplink time division multiple access (TDMA) systems. In the proposed scheme, after the received signal is transformed into a frequency-domain signal via fast Fourier transform (FFT), CCI from an adjacent cell is suppressed by the nulling filter. Moreover, the proposed scheme employs a soft canceller and minimum mean square error (SC/MMSE) based turbo equalizer to suppress the performance degradation due to inter-symbol interference (ISI) caused by the nulling filter as well as the ISI induced by fading channel. Computer simulation confirms that the proposed scheme is effective in suppression of CCI compared to the conventional linear frequency-domain equalizer.

The traffic with asymmetry between uplink and downlink has recently been getting remarkable on mobile communication systems providing multimedia communication services. In the future mobile communications, the accommodation of asymmetric traffic is essential to realize efficient multimedia mobile communication systems. This paper discusses asymmetric traffic accommodation in CDMA/FDD cellular packet communication systems and proposes its efficient scheme using an adaptive cell sizing technique. In the proposed scheme, each base station autonomously controls its coverage area so that almost the same communication quality can be achieved across the service area under the asymmetric traffic conditions. We present some numerical examples to demonstrate the effectiveness of the proposed scheme by using computer simulation. The simulation results show that, under asymmetric traffic conditions, the proposed scheme can provide fair communication quality across the service area in both links and can improve total transmission capacity in the uplink.

For the optimization design of air circuit breaker (ACB), it is important and necessary to calculate the electro-dynamic repulsion force acting on the movable contact. A method based on 3-D FEM with the equations that describe the relationships among current, magnetic field and repulsion force, which takes the ferromagnet into account, is adopted to calculate the electro-dynamic repulsion force. The method enables one to analyze the factors that affect the electro-dynamic repulsion force, including the number of the movable conductor parallel branches as well as the location of the axis and the shape of the flexible connection. The discussion of the calculation results is also presented in this paper.

In this work particle swarm optimization (PSO) aided with radial basis functions (RBF) has been suggested to carry out multiuser detection (MUD) for synchronous direct sequence code division multiple access (DS-CDMA) systems. The performance of the proposed algorithm is compared to that of other standard suboptimal detectors and genetic algorithm (GA) assisted MUD. It is shown to offer better performance than the others especially if there are many users.

Internet access traffic follows hourly patterns that depend on various factors, such as the periods users stay on-line at the access point (e.g. at home or in the office) or their preferences for applications. The clustering of Internet users may provide important information for traffic engineering and billing. For example, it can be used to set up service differentiation according to hourly behavior, resource optimization based on multi-hour routing and definition of tariffs that promote Internet access in low busy hours. In this work, we propose a methodology for clustering Internet users with similar patterns of Internet utilization, according to their hourly traffic utilization. The methodology resorts to three statistical multivariate analysis techniques: cluster analysis, principal component analysis and discriminant analysis. The methodology is illustrated through measured data from two distinct ISPs, one using a CATV access network and the other an ADSL one, offering distinct traffic contracts. Principal component analysis is used as an exploratory tool. Cluster analysis is used to identify the relevant Internet usage profiles, with the partitioning around medoids and Ward's method being the preferred clustering methods. For the two data sets, these methods lead to the choice of 3 clusters with different hourly traffic utilization profiles. The cluster structure is validated through discriminant analysis. It is also evaluated in terms of several characteristics of the user traffic not used in the cluster analysis, such as the type of applications, the amount of downloaded traffic, the activity duration and the transfer rate, resulting in coherent outcomes.

A hierarchical representation formed by an octree for a volume ray casting is a well-known data structure to skip over transparent regions requiring little preprocessing and storage. However, it accompanies unnecessary comparison and level shift between octants. We propose a new data structure named half-skewed octree, which is an auxiliary octree to support the conventional octree. In preprocessing step, a half-skewed octree selects eight different child octants in each generation step compared with the conventional octree. During rendering, after comparing an octant of the conventional octree with corresponding octant of the half-skewed octree simultaneously at the same level, a ray chooses one of two octants to jump over transparent regions farther away. By this method, we can reduce unnecessary comparison and level shift between octants. Another problem of a conventional octree structure is that it is difficult to determine a distance from the boundary of a transparent octant to opposite boundary. Although we exploit the previously proposed distance template, we cannot expect the acceleration when a ray direction is almost parallel to the octant's boundary. However, our method can solve it without additional operations because a ray selects one octant to leap farther away. As a result, our approach is much faster than the method using conventional octree while preserving image quality and requiring minimal storage.

Detecting edge directions and estimating the exact value of a missing line are currently active research areas in deinterlacing processing. This paper proposes a spatial domain fuzzy rule that is based on an interpolation algorithm, which is suitable to the region with high motion or scene change. The algorithm utilizes fuzzy theory to find the most accurate edge direction with which to interpolate missing pixels. The proposed fuzzy direction oriented interpolator operates by identifying small pixel variations in seven orientations (0°, 45°, -45°, 63°, -63°, 72°, and -72°), while using rules to infer the edge direction. The Bayesian network model selects the most suitable deinterlacing method among three deinterlacing methods and it successively builds approximations of the deinterlaced sequence, by evaluating three methods in each condition. Detection and interpolation results are presented. Experimental results show that the proposed algorithm provides a significant improvement over other existing deinterlacing methods. The proposed algorithm is not only for speed, but also effective for reducing deinterlacing artifacts.

An age-group classification method based on a fusion of different classifiers with different two-dimensional feature extraction algorithms is proposed. Theoretically, an integration of multiple classifiers can provide better performance compared to a single classifier. In this paper, we extract effective features from one sample image using different dimensional reduction methods, construct multiple classifiers in each subspace, and combine them to reduce age-group classification errors. As for the dimensional reduction methods, two-dimensional PCA (2DPCA) and two-dimensional LDA (2DLDA) are used. These algorithms are antisymmetric in the treatment of the rows and the columns of the images. We prepared the row-based and column-based algorithms to make two different classifiers with different error tendencies. By combining these classifiers with different errors, the performance can be improved. Experimental results show that our fusion-based age-group classification method achieves better performance than existing two-dimensional algorithms alone.

In this paper, we propose a predictive block-constrained trellis-coded quantization (BC-TCQ) to quantize cepstral coefficients for distributed speech recognition. For prediction of the cepstral coefficients, the first order auto-regressive (AR) predictor is used. To quantize the prediction error signal effectively, we use the BC-TCQ. The quantization is compared to the split vector quantizers used in the ETSI standard, and is shown to lower cepstral distance and bit rates.

This paper proposes a suboptimal algorithm for the maximum likelihood detection (MLD) in multiple-input multiple-output (MIMO) communications. The proposed algorithm regards transmitted signals as continuous variables in the same way as a common method for the discrete optimization problem, and then searches for candidates of the transmitted signals in the direction of a modified gradient vector of the metric. The vector is almost proportional to the direction of the noise enhancement, from which zero-forcing (ZF) or minimum mean square error (MMSE) algorithms suffer. This method sets the initial guess to the solution by ZF or MMSE algorithms, which can be recursively calculated. Also, the proposed algorithm has the same complexity order as that of conventional suboptimal algorithms. Computer simulations demonstrate that it is much superior in BER performance to the conventional ones.

This paper deals with statistical mechanical characteristics of fuzzy clustering regularized with fuzzy entropy. We obtain the Fermi-Dirac distribution function as a membership function by regularizing the fuzzy c-means with fuzzy entropy. Then we formulate it as a direct annealing clustering, and examine the meanings of Fermi-Dirac function and fuzzy entropy from a statistical mechanical point of view, and show that this fuzzy clustering method is none other than the Fermi-Dirac statistics.