This study investigates the effect of frame intervals on the accuracy of the Lucas–Kanade optical flow estimates for high-frame-rate (HFR) videos, with a view to realizing accurate HFR-video-based optical flow estimation. For 512 512 pixels videos of patterned objects moving at different speeds and captured at 1000 frames per second, the averages and standard deviations of the estimated optical flows were determined as accuracy measures for frame intervals of 1–40 ms. The results showed that the accuracy was highest when the displacement between frames was around 0.6 pixel/frame. This common property indicates that accurate optical flow estimation for HFR videos can be realized by varying frame intervals according to the motion field: a small frame interval for high-speed objects and a large frame interval for low-speed objects.

In this letter, we apply dynamic software updating to long-lived applications on the DDS middleware while minimizing service interruption and satisfying Quality of Service (QoS) requirements. We dynamically updated applications which run on a commercial DDS implementation to demonstrate the applicability of our approach to dynamic updating. The results show that our update system does not impose an undue performance overhead–all patches could be injected in less than 350 ms and the maximum CPU usage is less than 17%. In addition, the overhead on application throughput due to dynamic updates ranged from 0 to at most 8% and the deadline QoS of the application was satisfied while updating.

In this letter, we present the evaluation of an option-based learning algorithm, developed to perform a conflict-free allocation of calls among cars in a multi-car elevator system. We evaluate its performance in terms of the service time, its flexibility in the task-allocation, and the load balancing.

In a large queuing system, the effect of the ratio of the filled data on the queue and waiting time from the head of a queue to the service gate are important factors for process efficiency because they are too large to ignore. However, many research works assumed that the factors can be considered to be negligible according to the queuing theory. Thus, the existing queuing models are not applicable to the design of large-scale systems. Such a system could be used as a product classification center for a home delivery service. In this paper, we propose a tree-queue model for large-scale systems that is more adaptive to efficient processes compared to existing models. We analyze and design a mean waiting time equation related to the ratio of the filled data in the queue. Based on simulations, the proposed model demonstrated improvement in process-efficiency, and it is more suitable to realistic system modeling than other compared models for large-scale systems.

Non-negative matrix factorization (NMF) is widely used for monaural musical sound source separation because of its efficiency and good performance. However, an additional clustering process is required because the musical sound mixture is separated into more signals than the number of musical tracks during NMF separation. In the conventional method, manual clustering or training-based clustering is performed with an additional learning process. Recently, a clustering algorithm based on the mel-frequency cepstrum coefficient (MFCC) was proposed for unsupervised clustering. However, MFCC clustering supplies limited information for clustering. In this paper, we propose various timbre features for unsupervised clustering and a clustering algorithm with these features. Simulation experiments are carried out using various musical sound mixtures. The results indicate that the proposed method improves clustering performance, as compared to conventional MFCC-based clustering.

We propose a network coordinated opportunistic beamforming (NC-OBF) protocol for downlink K-cell networks with M-antenna base stations (BSs). In the NC-OBF scheme, based on pseudo-randomly generated BF vectors, a user scheduling strategy is introduced, where each BS opportunistically selects a set of mobile stations (MSs) whose desired signals generate the minimum interference to the other MSs. Its performance is then analyzed in terms of degrees-of-freedom (DoFs). As our achievability result, it is shown that KM DoFs are achievable if the number N of MSs in a cell scales at least as SNRKM-1, where SNR denotes the received signal-to-noise ratio. Furthermore, by deriving the corresponding upper bound on the DoFs, it is shown that the NC-OBF scheme is DoF-optimal. Note that the proposed scheme does not require the global channel state information and dimension expansion, thereby resulting in easier implementation.

In Digital Library (DL) applications, digital book clustering is an important and urgent research task. However, it is difficult to conduct effectively because of the great length of digital books. To do the correct clustering for digital books, a novel method based on probabilistic topic model is proposed. Firstly, we build a topic model named LDAC. The main goal of LDAC topic modeling is to effectively extract topics from digital books. Subsequently, Gibbs sampling is applied for parameter inference. Once the model parameters are learned, each book is assigned to the cluster which maximizes the posterior probability. Experimental results demonstrate that our approach based on LDAC is able to achieve significant improvement as compared to the related methods.

In this letter, a novel resource allocation method is proposed for Discrete Fourier Transform Spread Orthogonal Frequency Division Multiple Access (DFT-s-OFDMA) systems in Long Term Evolution (LTE). The proposed method is developed based on a minimal metric loss criterion and performs better than the commonly used Recursive Maximum Expansion (RME) method.

A new codebook mapping algorithm for artificial bandwidth extension (ABE) is introduced in this paper. We design a wideband line spectrum pair (LSP) codebook which is coupled with the same index as the LSP codebook of a narrowband speech codec. The received narrowband LSP codebook indices are used to directly induce wideband LSP codewords. Thus, the proposed scheme eliminates codebook search processing to estimate the wideband spectrum envelope. We apply the proposed scheme to bandwidth extension in adaptive multi-rate (AMR) compressed domain. Its performance is assessed via the perceptual evaluation of speech quality (PESQ), informal listening tests, and weighted million operations per second (WMOPS) calculations.

This paper investigates the application of inter-cell interference coordination (ICIC) in heterogeneous networks for the LTE-Advanced downlink where picocells are overlaid onto macrocells. In LTE-Advanced, in order to perform ICIC, almost blank subframes (ABSs) are employed, where only the cell-specific reference signal (CRS) is transmitted to protect the subframes in the picocells from severe interference from the macrocells. Furthermore, multicast/broadcast over single-frequency network (MBSFN) subframes are employed to reduce the interference of the CRS on the data channel, although the control channel still suffers from interference from the CRS. When the cell range expansion (CRE), which offload the UEs from macrocells to picocells, is used to improve the system performance, the influence from the CRS increases. In order to assess the influence, the required CRE bias to improve the data channel is investigated based on a system-level simulation under various conditions such as the number of picocells, the protected subframe ratio, and the user distribution. The simulation results show that the cell-edge user throughput is improved with the CRE bias of more than 8 dB, employing ABSs. Furthermore, simulation results show that one dominant source of interference is observed for the sets of user equipment (UEs) connected to the picocells via CRE with such a bias value. Based on observation, the influence that the CRS has on the control channel, i.e., physical control format indicator channel (PCFICH), and physical downlink control channel (PDCCH) is investigated based on a link-level simulation combined with a system-level simulation. The simulation results show that protecting the PCFICH is very important compared to protecting the PDCCH, since the block error rate (BLER) performance of the PCFICH becomes worse than the required BLER of 10-3 to support various conditions, although the BLER performance of the PDCCH can exceed the required BLER of 10-2 by spanning the PDCCH over three OFDM symbols.

Initialize and weak-program erasing scheme is proposed to achieve high-performance and high-reliability Ferroelectric (Fe-) NAND flash solid-state drive (SSD). Bit-by-bit erase VTH control is achieved by the proposed erasing scheme and history effects in Fe-NAND is also suppressed. History effects change the future erase VTH shift characteristics by the past program voltage. The proposed erasing scheme decreases VTH shift variation due to history effects from ±40% to ±2% and the erase VTH distribution width is reduced from over 0.4 V to 0.045 V. As a result, the read and VPASS disturbance decrease by 42% and 37%, respectively. The proposed erasing scheme is immune to VTH variations and voltage stress. The proposed erasing scheme also suppresses the power and bandwidth degradation of SSD.

This paper proposes optical node architectures for the single-layer optical cross-connect (OXC) and hierarchical OXC (HOXC) that utilize dedicated add/drop switches for originating/terminating traffic at a node. For both single-layer OXC and HOXC, three architectures with different restrictions on add/drop capabilities are presented. The performance of the proposed architectures is compared through numerical experiments. The architectures significantly reduce total switch scale and minimize necessary switch size while attaining colorless, directionless and contentionless capabilities.

We describe a multi-step approach for automatic segmentation of the femoral head and the acetabulum in the hip joint from three dimensional (3D) CT images. Our segmentation method consists of the following steps: 1) construction of the valley-emphasized image by subtracting valleys from the original images; 2) initial segmentation of the bone regions by using conventional techniques including the initial threshold and binary morphological operations from the valley-emphasized image; 3) further segmentation of the bone regions by using the iterative adaptive classification with the initial segmentation result; 4) detection of the rough bone boundaries based on the segmented bone regions; 5) 3D reconstruction of the bone surface using the rough bone boundaries obtained in step 4) by a network of triangles; 6) correction of all vertices of the 3D bone surface based on the normal direction of vertices; 7) adjustment of the bone surface based on the corrected vertices. We evaluated our approach on 35 CT patient data sets. Our experimental results show that our segmentation algorithm is more accurate and robust against noise than other conventional approaches for automatic segmentation of the femoral head and the acetabulum. Average root-mean-square (RMS) distance from manual reference segmentations created by experienced users was approximately 0.68 mm (in-plane resolution of the CT data).

The Viterbi algorithm is widely used for decoding of the convolutional codes. The trace-back method is preferable to the register exchange method because of lower power consumption especially for convolutional codes with many states. A drawback of the conventional trace-back is that it generally requires long latency to obtain the decoded data. In this paper, a method of the trace-back with source states instead of decision bits is proposed which reduces the number of memory accesses. The dedicated memory is also presented which supports the proposed trace-back method. The reduced memory accesses result in smaller power consumption and a shorer decode latency than the conventional method.

This paper presents a novel method for improving the readability of automatic speech recognition (ASR) results for classroom lectures. Because speech in a classroom is spontaneous and contains many ill-formed utterances with various disfluencies, the ASR result should be edited to improve the readability before presenting it to users, by applying some operations such as removing disfluencies, determining sentence boundaries, inserting punctuation marks and repairing dropped words. Owing to the presence of many kinds of domain-dependent words and casual styles, even state-of-the-art recognizers can only achieve a 30-50% word error rate for speech in classroom lectures. Therefore, a method for improving the readability of ASR results is needed to make it robust to recognition errors. We can use multiple hypotheses instead of the single-best hypothesis as a method to achieve a robust response to recognition errors. However, if the multiple hypotheses are represented by a lattice (or a confusion network), it is difficult to utilize sentence-level knowledge, such as chunking and dependency parsing, which are imperative for determining the discourse structure and therefore imperative for improving readability. In this paper, we propose a novel algorithm that infers clean, readable transcripts from spontaneous multiple hypotheses represented by a confusion network while integrating sentence-level knowledge. Automatic and manual evaluations showed that using multiple hypotheses and sentence-level knowledge is effective to improve the readability of ASR results, while preserving the understandability.

This paper discusses a control system that employs a power line to transfer signals to control the motion of a single machine, and explores the influence of packet losses on the quality of the control. As an example of a controlled system, a controller with a rotary inverted pendulum as a controlled object, is considered. The feedback loop in between is the power line. The control performance is evaluated in the power line cyclostationary noise environment and compared against the performance in a stationary noise environment. As a result, it is confirmed that the power line and its cyclostationary noise features present an advantage against transmission in a channel with stationary noise.

A 65 nm self synchronous field programmable gate array (SSFPGA) which uses autonomous gate-level power gating with minimal control circuitry overhead for energy minimum operation is presented. The use of self synchronous signaling allows the FPGA to operate at voltages down to 370 mV without any parameter tuning. We show both 2.6x total energy reduction and 6.4x performance improvement at the same time for energy minimum operation compared to the non-power gated SSFPGA, and compared to the latest research 1.8x improvement in power-delay product (PDP) and 2x performance improvement. When compared to a synchronous FPGA in a similar process we are able to show up to 84.6x PDP improvement. We also show energy minimum operation for maximum throughput on the power gated SSFPGA is achieved at 0.6 V, 27 fJ/operation at 264 MHz.

Many studies and systems that incorporate elements such as “pleasure” and “fun” in the game to improve a learner's motivation have been developed in the field of learning environments. However, few are the studies of situations where many learners gather at a single computer and participate in a game-based learning environment (GBLE), and where the GBLE designs the learning process by controlling the interactions between learners such as competition, collaboration, and learning by teaching. Therefore, the purpose of this study is to propose a framework of educational control that induces and activates interaction between learners intentionally to create a learning opportunity that is based on the knowledge understanding model of each learner. In this paper, we explain the design philosophy and the framework of our GBLE called “Who becomes the king in the country of mathematics?” from a game viewpoint and describe the method of learning support control in the learning environment. In addition, we report the results of the learning experiment with our GBLE, which we carried out in a junior high school, and include some comments by a principal and a teacher. From the results of the experiment and some comments, we noticed that a game may play a significant role in weakening the learning relationship among students and creating new relationships in the world of the game. Furthermore, we discovered that learning support control of the GBLE has led to activation of the interaction between learners to some extent.

The development of ontology at the instance level requires the extraction of the terms defining the instances from various data sources. These instances then are linked to the concepts of the ontology, and relationships are created between these instances for the next step. However, before establishing links among data, ontology engineers must classify terms or instances from a web document into an ontology concept. The tool for help ontology engineer in this task is called ontology population. The present research is not suitable for ontology development applications, such as long time processing or analyzing large or noisy data sets. OntoPop system introduces a methodology to solve these problems, which comprises two parts. First, we select meaningful features from syntactic relations, which can produce more significant features than any other method. Second, we differentiate feature meaning and reduce noise based on latent semantic analysis. Experimental evaluation demonstrates that the OntoPop works well, significantly out-performing the accuracy of 49.64%, a learning accuracy of 76.93%, and executes time of 5.46 second/instance.

This paper presents a memory-efficient VLSI architecture for output probability computations (OPCs) of continuous hidden Markov models (HMMs) and likelihood score computations (LSCs). These computations are the most time consuming part of HMM-based isolated word recognition systems. We demonstrate multiple fast store-based block parallel processing (MultipleFastStoreBPP) for OPCs and LSCs and present a VLSI architecture that supports it. Compared with conventional fast store-based block parallel processing (FastStoreBPP) and stream-based block parallel processing (StreamBPP) architectures, the proposed architecture requires fewer registers and less processing time. The processing elements (PEs) used in the FastStoreBPP and StreamBPP architectures are identical to those used in the MultipleFastStoreBPP architecture. From a VLSI architectural viewpoint, a comparison shows that the proposed architecture is an improvement over the others, through efficient use of PEs and registers for storing input feature vectors.