A smoothing method for minimum phone error linear regression (MPELR) is proposed in this paper. We show that the objective function for minimum phone error (MPE) can be combined with a prior mean distribution. When the prior mean distribution is based on maximum likelihood (ML) estimates, the proposed method is the same as the previous smoothing technique for MPELR. Instead of ML estimates, maximum a posteriori (MAP) parameter estimate is used to define the mode of prior mean distribution to improve the performance of MPELR. Experiments on a large vocabulary speech recognition task show that the proposed method can obtain 8.4% relative reduction in word error rate when the amount of data is limited, while retaining the same asymptotic performance as conventional MPELR. When compared with discriminative maximum a posteriori linear regression (DMAPLR), the proposed method shows improvement except for the case of limited adaptation data for supervised adaptation.

Metaheuristic methods have been studied for combinational optimization problems for some time. Recently, a Consultant-Guided Search (CGS) has been proposed as a metaheuristic method for the Traveling Salesperson Problem (TSP). This approach is an algorithm in which a virtual person called a client creates a solution based on consultation with a virtual person called a consultant. In this research, we propose a parallel algorithm which uses the Ant Colony System (ACS) to create a solution with a consultant in a Consultant-Guided Search, and calculate an approximation solution for the TSP. Finally, we execute a computer experiment using the benchmark problems (TSPLIB). Our algorithm provides a solution with less than 2% error rate for problem instances using less than 2000 cities.

The use of cooperative nodes is effective for enhancing the reliability of wireless data transmission between a source and a destination by means of transmit diversity effect. However, in its application to wireless multi-hop networks, how to form cooperative node candidates and how to select multiple cooperative nodes out of them have not been well investigated. In this paper, we propose a multiple cooperative node selection method based on a criterion composed of “quality” and “angle” metrics, which can select and order adequate cooperative nodes. Computer simulation results show that the proposed method can effectively reduce the packet error rate without any knowledge on node location.

We propose a novel IDDQ outlier screening flow through a two-phase approach: a clustering-based filtering and an estimation-based current-threshold determination. In the proposed flow, a clustering technique first filters out chips that have high IDDQ current. Then, in the current-threshold determination phase, device-parameters of the unfiltered chips are estimated based on measured IDDQ currents through Bayesian inference. The estimated device-parameters will further be used to determine a statistical leakage current distribution for each test pattern and to calculate a and suitable current-threshold. Numerical experiments using a virtual wafer show that our proposed technique is 14 times more accurate than the neighbor nearest residual (NNR) method and can achieve 80% of the test escape in the case of small leakage faults whose ratios of leakage fault sizes to the nominal IDDQ current are above 40%.

When deploying a new mobile technology such as LTE, it is crucial to identify the factors that affect the radio network in terms of capacity and quality of service. In this context, network coverage is arguably the single most influential factor. This work presents a metaheuristic-optimization approach to automatically adapt the signal losses due to clutter, based on a set of field measurements. The optimization procedure is performed regionally, enabling the calculation of accurate radio-propagation predictions. The evaluation of the proposed approach is carried out on three different regions in Slovenia, where Telekom Slovenije, d.d., provides LTE coverage. The results show radio-propagation predictions of improved quality and the benefits of the presented approach over manual methods, both in terms of problem size and solution accuracy.

The increasing demand of indoor location based service (LBS) has promoted the development of localization techniques. As an important alternative, fingerprinting localization technique can achieve higher localization accuracy than traditional trilateration and triangulation algorithms. However, it is computational expensive to construct the fingerprint database in the offline phase, which limits its applications. In this paper, we propose an efficient indoor positioning system that uses a new empirical propagation model, called regional propagation model (RPM), which is based on the cluster based propagation model theory. The system first collects the sparse fingerprints at some certain reference points (RPs) in the whole testing scenario. Then affinity propagation clustering algorithm operates on the sparse fingerprints to automatically divide the whole scenario into several clusters or sub-regions. The parameters of RPM are obtained in the next step and are further used to recover the entire fingerprint database. Finally, the location estimation is obtained through the weighted k-nearest neighbor algorithm (WkNN) in the online localization phase. We also theoretically analyze the localization accuracy of the proposed algorithm. The numerical results demonstrate that the proposed propagation model can predict the received signal strength (RSS) values more accurately than other models. Furthermore, experiments also show that the proposed positioning system achieves higher localization accuracy than other existing systems while cutting workload of fingerprint calibration by more than 50% in the offline phase.

We propose a speaker adaptation method based on the probabilistic principal component analysis (PPCA) of acoustic models. We define a training matrix which is represented in a two-way array and decompose the training models by PPCA to construct bases. In the two-way array representation, each training model is represented as a matrix and the columns of each training matrix are treated as training vectors. We formulate the adaptation equation in the maximum a posteriori (MAP) framework using the bases and the prior.

Non-Volatile Memories (NVMs) are considered as promising memory technologies for Last-Level Cache (LLC) due to their low leakage and high density. However, NVMs have some drawbacks such as high dynamic energy in modifying NVM cells, long latency for write operation, and limited write endurance. A number of approaches have been proposed to overcome these drawbacks. But very little attention is paid to consider the cache coherency issue. In this letter, we suggest a new cache coherence protocol to reduce the write operations of the LLC. In our protocol, the block data of the LLC is updated only if the cache block is written-back from a private cache, which leads to avoiding useless write operations in the LLC. The simulation results show that our protocol provides 27.1% energy savings and 26.3% lifetime improvements in STT-RAM at maximum.

With the development of personal electronic equipment, the use of a smartphone with a tri-axial accelerometer to detect human physical activity is becoming popular. In this paper, we propose a new feature based on FFT for activity recognition from tri-axial acceleration signals. To improve the classification performance, two fusion methods, minimal distance optimization (MDO) and variance contribution ranking (VCR), are proposed. The new proposed feature achieves a recognition rate of 92.41%, which outperforms six traditional time- or frequency-domain features. Furthermore, the proposed fusion methods effectively improve the recognition rates. In particular, the average accuracy based on class fusion VCR (CFVCR) is 97.01%, which results in an improvement in accuracy of 4.14% compared with the results without any fusion. Experiments confirm the effectiveness of the new proposed feature and fusion methods.

Integrated InP polarization converters based on half-ridge structure are studied numerically. We demonstrate that the fabrication tolerance of the half-ridge structure can be extended significantly by introducing a slope at the ridge side and optimizing the thickness of the residual InGaAsP layer. High polarization conversion over 90% is achieved with the broad range of the waveguide width from 705 to 915~nm, corresponding to a factor-of-two or larger improvement in the fabrication tolerance compared with that of the conventional polarization converters. Finally we present a simple fabrication procedure of this newly proposed structure, where the thickness of the residual InGaAsP layer is controlled precisely by using a thin etch-stop layer.

The paper presents ultra-high-capacity transmission technologies based on multi-core space-division-multiplexing. In order to realize high-capacity multi-core fiber (MCF) transmission, investigation of low crosstalk fiber and connection technology is important, and high-density signal generation using multilevel modulation and crosstalk management are also key technologies. 1Pb/s multi-core fiber transmission experiment using space-division-multiplexing is also described.

Detailed high capacity vector maps must be compressed effectively for transmission or storage in Web GIS (geographic information system) and mobile GIS applications. In this paper, we present a polyline compression method that consists of polyline feature-based hybrid simplification and second derivative-based data compression. The polyline hybrid simplification function detects the feature points from a polyline using DP, SF, and TF algorithms, and divides the polyline into sectors using these feature points. It then simplifies the sectors using an algorithm to determine the minimum area difference among the DP, SF, and TF results. The polyline data compression method segments the second derivatives of the simplified polylines into integer and fractional parts. The integer parts are compressed using the minimum bounding box of the layer to determine the broad position of the object. The fractional parts are compressed using hierarchical precision levels. Experimental results verify that our method has higher simplification and compression efficiency than conventional methods and produces good quality compressed maps.

To implement a wideband bandpass filter with improved skirt-selectivity and out-band characteristics, a new parallel-coupled three-line unit with two short-circuited stubs symmetrically-loaded at the center line is proposed. Unlike most traditional ones, the passband of the proposed parallel-coupled three-line structure is based on the cross-coupling between non-adjacent lines rather than the direct-coupling between adjacent ones, whereas a pair of attenuation poles is found in the stopbands. After revealing its work mechanism, an efficient filter-design-scheme is correspondingly proposed for the presented structure. Firstly, based on a chebyshev-filter synthesis theory, a wideband passband filter consisting of a parallel-coupled two-line and two short-circuited stubs loaded at the input- and output- ports is designed. Furthermore, by putting a properly-designed 3/4-wavelength stepped-impedance resonator (SIR) in between the parallel-coupled two lines, two attenuation poles are then realized at the frequencies very close to the cutoff ones. Accordingly, the roll-off characteristics of the filter are significantly-improved to greater than 100,dB/GHz. Furthermore, two-section open-ended stubs are used to replace the short-circuited ones to realize a pair of extra attenuation poles in stopbands. To validate the proposed techniques, a wideband filter with a bandwidth of 3--5,GHz (Fractional bandwidth (FBW) $= (5,GHz-3,GHz)/4,GHz =50%)$ was designed, simulated, fabricated and measured. The measured responses of the filter agree well with the simulation and theoretical ones, which validates the effectiveness of the newly-proposed three-line unit and the corresponding design scheme.

The main purpose of this paper is to apply the boundary integral equation (BIE) method to the analysis of spoof localized surface plasmons (spoof LSPs) excited in a perfectly conducting cylinder with longitudinal corrugations. Frequency domain BIE schemes based on electric field integral equation (EFIE), magnetic field integral equation (MFIE) and combined field integral equation (CFIE) formulations are used to solve two-dimensional electromagnetic (EM) problems of scattering from the cylinder illuminated by a transverse electric plane wave. In this approach effects of spoof LSPs are included in the secondary surface current and charge densities resulting from the interaction between the plane wave and the cylinder. Numerical results obtained with the BIE schemes are validated by comparison with that of a recently proposed modal solution based on the metamaterial approximation.

In this paper, the function expansion based topology optimization is employed to the automatic optimization of the waveguide dispersion property, and the optimum design of low-dispersion slow-light photonic crystal waveguides is demonstrated. In order to realize low-dispersion and large group index, an objective function to be optimized is expressed by the weighted sum of the objective functions for the desired group index and the low-dispersion property, and weighting coefficients are updated through the optimization process.

Performance suveyrance of CPML (Convolutional PML) for FDTD (Finite-Difference Time-Domain) method in cylindrical coordinate system was carried out. The CPML was placed perpendicularly to the radial axis and designed to absorb diverging or converging waves. To be able to analyze microstructured optical fibers and disk/ring resonators we introduced finite axial wavenumbers into the FDTD formulation. We investigated the dependence of reflectivity upon CPML's constituteve parameters such as $kappa$ and $sigma$ for various curvature radii and the axial wavenumbers. As a result of evaluation we found that the reflectivity gradually increased togather with the increase of the wavenumber. We also confirmed that the absorption performance was of the similar order for the converging waves and the diverging ones provided that their curvature radii were the same.

Blur distortion is a common artifact in image communication and affects the perceived sharpness of a digital image. In this paper, we capitalize on the mathematical knowledge of Gaussian convolution and propose a strategy to minimally reblur test images. From the reblur algorithm, synthetic reblur images are created. We propose a new blind blur metric which makes use of the reblur images to produce blur scores. Compared to other no-reference blur assessments, the proposed method has the advantages of fast computation and training-free operation. Experiment results also show that the proposed method can produce blur scores which are highly correlated with human perception of blurriness.

In recent years, computation offloading, through which applications on a mobile device can offload their computations onto more resource-rich clouds, has emerged as a promising technique to reduce battery consumption as well as augment the devices' limited computation and memory capabilities. In order for computation offloading to be energy-efficient, an accurate estimate of battery consumption is required to decide between local processing and computation offloading. In this paper, we propose a novel technique for estimating battery consumption without requiring detailed information about the mobile application's internal structure or its execution behavior. In our approach, the relationship is derived between variables that affect battery consumption (i.e., the input to the application, the transmitted data, and resource status) and the actual consumed energy from the application's past run history. We evaluated the performance of the proposed technique using two different types of mobile applications over different wireless network environments such as 3G, Wi-Fi, and LTE. The experimental results show that our technique can provide tolerable estimation accuracy and thus make correct decisions between local processing and computation offloading.

Coalition Structure Generation (CSG) is a main research issue in the domain of coalition games. A majority of existing works assume that the value of a coalition is independent of others in the coalition structure. Recently, there has been interest in a more realistic settings, where the value of a coalition is affected by the formation of other coalitions. This effect is known as externality. The focus of this paper is to make use of Maximum Satisfiability (MaxSAT) to solve the CSG problem where externalities may exist. In order to reduce the exponentially growing number of possible solutions in the CSG problem, we follow the previous works by representing the CSG problem as sets of rules in MC-nets (without externalities) and embedded MC-nets (with externalities). Specifically, enlightened by the previous MC-net-based algorithms exploiting the constraints among rule relations to solve the CSG problem, we encode such constraints into weighted partial MaxSAT (WPM) formulas. Experimental results demonstrate that an off-the-shelf MaxSAT solver achieves significant improvements compared to the previous algorithm for the same set of problem instances.

With the advances in wireless Internet and mobile positioning technology, location-based services (LBSs) have become popular. In LBSs, users must send their exact locations in order to use the services, but they may be subject to several privacy threats. To solve this problem, query processing algorithms based on a cloaking method have been proposed. The algorithms use spatial cloaking methods to blur the user's exact location in a region satisfying the required privacy threshold (k). With the cloaked region, an LBS server can execute a spatial query processing algorithm preserving their privacy. However, the existing algorithms cannot provide good query processing performance. To resolve this problem, we, in this paper, propose a k-NN query processing algorithm based on network Voronoi diagram for spatial networks. Therefore, our algorithm can reduce network expansion overhead and share the information of the expanded road network. In order to demonstrate the efficiency of our algorithms, we have conducted extensive performance evaluations. The results show that our algorithm achieves better performance on retrieval time than the existing algorithms, such as PSNN and kRNN. This is because our k-NN query processing algorithm can greatly reduce a network expansion cost for retrieving k POIs.