The online buffer management problem formulates the problem of queuing policies of network switches supporting QoS (Quality of Service) guarantee. In this paper, we consider one of the most standard models, called multi-queue switches model. In this model, Albers et al. gave a lower bound $\frac{e}{e-1}$, and Azar et al. gave an upper bound $\frac{e}{e-1}$ on the competitive ratio when m, the number of input ports, is large. They are tight, but there still remains a gap for small m. In this paper, we consider the case where m = 2, namely, a switch is equipped with two ports, which is called a bicordal buffer model. We propose an online algorithm called Segmental Greedy Algorithm (SG) and show that its competitive ratio is at most $\frac{16}{13} (\simeq 1.231)$, improving the previous upper bound by $\frac{9}{7} (\simeq 1.286)$. This matches the lower bound given by Schmidt.

View full abstract

It has been known that testing of reversible circuits is relatively easier than conventional irreversible circuits in the sense that few test vectors are needed to cover all stuck-at faults. This paper shows, however, that it is NP-hard to generate a minimum complete test set for stuck-at faults on the wires of a reversible circuit using a polynomial time reduction from 3SAT to the problem. We also show non-trivial lower bounds for the size of a minimum complete test set.

View full abstract

Reversal complexity has been studied as a fundamental computational resource along with time and space complexity. We revisit it by contrasting with access complexity which we introduce in this study. First, we study the structure of space bounded reversal and access complexity classes. We characterize the complexity classes L, P and PSPACE in terms of space bounded reversal and access complexity classes. We also show that the difference between polynomial space bounded reversal and access complexity is related with the L versus P problem. Moreover, we introduce a theory of memory access patterns, which is an abstracted structure of the order of memory accesses during a random access computation, and extend the notion of reversal and access complexity for general random access computational models. Then, we give probabilistic analyses of reversal and access complexity for almost all memory access patterns. In particular, we prove that almost all memory access patterns have ω(log n) reversal complexity while all languages in L are computable within O(log n) reversal complexity.

View full abstract

With the development of digital TV system, how to display the NTSC signal in digital TV system is a problem. De-interlacing is an algorithm to solve it. In previous papers, using motion compensation (MC) method for de-interlacing needs lots of computation complexity and it is not easy to implement in hardware. In this paper, a content adaptive de-interlacing algorithm is proposed. Our algorithm is based on the motion adaptive (MA) method which combines the advantages of intra-field and inter-field method. We propose a block type decision mechanism to predict the video content instead of a blind processing with MC method throughout the entire frame. Additionally, in intra-field method, we propose the edge-base adaptive weight average (EAWA) method to achieve a better performance and smooth the edge and stripe. In order to demonstrate our algorithm, we implement the de-interlacing system on the DSP platform with thorough complexity analysis. Compared to MC method, we not only achieve higher video quality in objective and subjective view, but also consume lower computation power. From the profiling on CPU run-time analysis, the proposed algorithm is only one-fifth of MC method. At the DSP demonstration board, the saving ratio is about 54% to 96%.

View full abstract

Task preemption is a critical mechanism for building an effective multi-tasking environment on dynamically reconfigurable processors. When a task is preempted, its necessary state information must be correctly preserved in order for the task to be resumed later. Not only do coarse-grained Dynamically Reconfigurable Processing Array (DRPAs) devices have different architectures using a variety of development tools, but the great amount of state data of hardware tasks executing on such devices are usually distributed on many different storage elements. To address these difficulties, this paper aims at studying a general method for capturing the state data of hardware tasks targeting coarse-grained DRPAs. Based on resource usage, algorithms for identifying preemption points and inserting preemption states subject to user-specified preemption latency are proposed. Moreover, a modification to automatically incorporate proposed steps into the system design flow is also discussed. The performance degradation caused by additional preemption states is minimized by allowing preemption only at predefined points where demanded resources are small. The evaluation result using a model based on NEC Electronics' DRP-1 shows that the proposed method can produce preemption points satisfying a given preemption latency with reasonable hardware overhead (from 6% to 15%).

View full abstract

Proofs written in algebraic specification languages are called proof scores. The proof score approach to design verification is attractive because it provides a flexible way to prove that designs for systems satisfy properties. Thus far, however, the approach has focused on safety properties. In this paper, we describe a way to verify that designs for systems satisfy liveness properties with the approach. A mutual exclusion protocol using a queue is used as an example. We describe the design verification and explain how it is verified that the protocol satisfies the lockout freedom property.

View full abstract

In recent years, heterogeneous devices have been employed frequently in mobile storage systems because a combination of such devices can supply a synergistically useful storage solution by taking advantage of each device. One important design constraint in heterogeneous storage systems is to mitigate I/O performance degradation stemming from the difference between access times of different devices. To this end, there has not been much work to devise proper buffer cache management algorithms. This paper presents a novel buffer cache management algorithm which considers both I/O cost per device and workload patterns in mobile computing systems with a heterogeneous storage pair of a hard disk and a NAND flash memory. In order to minimize the total I/O cost under varying workload patterns, the proposed algorithm employs a dynamic cache partitioning technique over different devices and manages each partition according to request patterns and I/O types along with the temporal locality. Trace-based simulations show that the proposed algorithm reduces the total I/O cost and flash write count significantly over the existing buffer cache algorithms on typical mobile traces.

View full abstract

In this study we propose a manufacturing control framework based on radio-frequency identification (RFID) technology and a distributed information system to construct a mass-customization production process in a loosely coupled shop-floor control environment. On the basis of this framework, we developed RFID middleware and an integrated information system for tracking and controlling the manufacturing process flow. A bicycle manufacturer was used to demonstrate the prototype system. The findings of this study were that the proposed framework can improve the visibility and traceability of the manufacturing process as well as enhance process quality control and real-time production pedigree access. Using this framework, an enterprise can easily integrate an RFID-based system into its manufacturing environment to facilitate mass customization and a just-in-time production model.

View full abstract

The mixture modeling framework is widely used in many applications. In this paper, we propose a component reduction technique, that collapses a Gaussian mixture model into a Gaussian mixture with fewer components. The EM (Expectation-Maximization) algorithm is usually used to fit a mixture model to data. Our algorithm is derived by extending mixture model learning using the EM-algorithm. In this extension, a difficulty arises from the fact that some crucial quantities cannot be evaluated analytically. We overcome this difficulty by introducing an effective approximation. The effectiveness of our algorithm is demonstrated by applying it to a simple synthetic component reduction task and a phoneme clustering problem.

View full abstract

A new online, unsupervised voice activity detection (VAD) method is proposed. The method is based on a feature derived from high-order statistics (HOS), enhanced by a second metric based on normalized autocorrelation peaks to improve its robustness to non-Gaussian noises. This feature is also oriented for discriminating between close-talk and far-field speech, thus providing a VAD method in the context of human-to-human interaction independent of the energy level. The classification is done by an online variation of the Expectation-Maximization (EM) algorithm, to track and adapt to noise variations in the speech signal. Performance of the proposed method is evaluated on an in-house data and on CENSREC-1-C, a publicly available database used for VAD in the context of automatic speech recognition (ASR). On both test sets, the proposed method outperforms a simple energy-based algorithm and is shown to be more robust against the change in speech sparsity, SNR variability and the noise type.

View full abstract

This paper proposes a distributed stream multiplexing architecture for video codec LSIs with multi-chip configuration. This distributed architecture utilizes a built-in media multiplexing unit with an external stream input and inter-chip communication interfaces. Parallel protocol processing, with an autonomous inter-chip control mechanism to mix and concatenate packets through daisy-chained transfer paths, provides a complete multi-chip stream output at the end of the chain. Dispensing with external post-processing devices contributes to both high throughput and downsizing of high-end video codec systems. It is configurable for parallel encoding of super high-resolution video, multi-view/-angled HDTV vision and multiple HDTV programs. The architecture was successfully implemented in a fabricated single-chip MPEG-2 422P@HL codec LSI and utilized for the development of a super high-resolution video codec system.

View full abstract

Retinopathy of Prematurity (ROP) is an infant disease characterized by increased dilation and tortuosity of the retinal blood vessels. Automatic tortuosity evaluation from retinal digital images is very useful to facilitate an ophthalmologist in the ROP screening and to prevent childhood blindness. This paper proposes a method to automatically classify the image into tortuous and non-tortuous. The process imitates expert ophthalmologists' screening by searching for clearly tortuous vessel segments. First, a skeleton of the retinal blood vessels is extracted from the original infant retinal image using a series of morphological operators. Next, we propose to partition the blood vessels recursively using an adaptive linear interpolation scheme. Finally, the tortuosity is calculated based on the curvature of the resulting vessel segments. The retinal images are then classified into two classes using segments characterized by the highest tortuosity. For an optimal set of training parameters the prediction is as high as 100%.

View full abstract

In this paper, we analyze existing vulnerabilities in hand-over for mobile WiMAX networks. To overcome these vulnerabilities, we propose a secure handover protocol that guarantees mutual authentication and forward/backward secrecy in handover. We present a formal analysis of our protocol using a logic-based formal method.

View full abstract

This paper proposes a novel algorithm for image feature extraction-the dual two-dimensional fuzzy class preserving projections ((2D)²FCPP). The main advantages of (2D)²FCPP over two-dimensional locality preserving projections (2DLPP) are: (1) utilizing the fuzzy assignation mechanisms to construct the weight matrix, which can improve the classification results; (2) incorporating 2DLPP and alternative 2DLPP to get a more efficient dimensionality reduction method-(2D)²LPP.

View full abstract

This letter presents a novel approach for traffic light detection in a video frame captured by an in-vehicle camera. The algorithm consists of rotated principal component analysis (RPCA), modified amplitude thresholding with respect to the histograms of the PC planes and final filtering with a neural network. The proposed algorithm achieves an average detection rate of 96% and is very robust to variations in the image quality.

View full abstract

This work proposes new features to improve the pathological voice quality classification performance. They are the means, the variances, and the perturbations of the higher-order statistics (HOS) such as the skewness and the kurtosis. The HOS-based features show meaningful differences among normal, grade 1, grade 2, and grade 3 voices classified in the GRBAS scale. The jitter, the shimmer, the harmonic-to-noise ratio (HNR), and the variance of the short-time energy are utilized as the conventional features. The performances are measured by the classification and regression tree (CART) method. Specifically, the CART-based method by utilizing both the conventional features and the HOS-based ones shows its effectiveness in the pathological voice quality measurement, with the classification accuracy of 87.8%.

View full abstract

In digital speech communication over noisy high packet loss rate wireless channels, improving the overall performance of the realtime speech coding and transmission system is of great importance. A novel joint speech coding and transmission algorithm is proposed by fully exploiting the correlation between speech coding, channel coding and the transmission process. The proposed algorithm requires no algorithm delay and less bandwidth expansion while greatly enhancing the error correcting performance and the reconstructed speech quality compared with conventional algorithms. Simulations show that the residual error rate is reduced by 84.36% and the MOS (Mean Opinion Score) is improved over 38.86%.

View full abstract

Improving the overall performance of reliable speech communication in ultrashort wave radios over very noisy channels is of great importance and practical use. An iterative joint source-channel (de-)coding and (de-)modulation (JSCCM) algorithm is proposed for ITU-T Rec.G.729EV by both exploiting the residual redundancy and passing soft information throughout the receiver while introducing a systematic global iteration process. Being fully compatible with existing transmitter structure, the proposed algorithm does not introduce additional bandwidth expansion and transmission delay. Simulations show substantial error correcting performance and synthesized speech quality improvement over conventional separate designed systems in delay and bandwidth constraint channels by using the JSCCM algorithm.

View full abstract

In this letter, we propose a cache organization that substantially reduces the memory bandwidth of motion compensation (MC) in the H.264/AVC decoders. To reduce duplicated memory accesses to P and B pictures, we employ a four-way set-associative cache in which its index bits are composed of horizontal and vertical address bits of the frame buffer and each line stores an 8 × 2 pixel data in the reference frames. Moreover, we alleviate the data fragmentation problem by selecting its line size that equals the minimum access size of the DDR SDRAM. The bandwidth of the optimized cache averaged over five QCIF IBBP image sequences requires only 129% of the essential bandwidth of an H.264/AVC MC.

View full abstract

A rotation invariant texture analysis technique is proposed with a novel combination of Radon Transform (RT) and Hidden Markov Models (HMM). Features of any texture are extracted during RT which due to its inherent property captures all the directional properties of a certain texture. HMMs are used for classification purpose. One HMM is trained for each texture on its feature vector which preserves the rotational invariance of feature vector in a more compact and useful form. Once all the HMMs have been trained, testing is done by picking any of these textures at any arbitrary orientation. The best percentage of correct classification (PCC) is above 98 % carried out on sixty texture of Brodatz album.

View full abstract