A rotator graph was proposed as a topology for interconnection networks of parallel computers, and it is promising because of its small diameter and small degree. However, a rotator graph is a directed graph that sometimes behaves harmfully when it is applied to actual problems. A bi-rotator graph is obtained by making each edge of a rotator graph bi-directional. In a bi-rotator graph, average distance is improved against a rotator graph with the same number of nodes. In this paper, we give an algorithm for the container problem in bi-rotator graphs with its evaluation results. The solution achieves some fault tolerance such as file distribution based information dispersal technique. The algorithm is of polynomial order of n for an n-bi-rotator graph. It is based on recursion and divided into two cases according to the position of the destination node. The time complexity of the algorithm and the maximum length of paths obtained are estimated to be O(n3) and 4n-5, respectively. Average performance of the algorithm is also evaluated by computer experiments.

In this paper, the 1-D real-valued discrete Gabor transform (RDGT) proposed in our previous work and its relationship with the complex-valued discrete Gabor transform (CDGT) are briefly reviewed. Block time-recursive RDGT algorithms for the efficient and fast computation of the 1-D RDGT coefficients and for the fast reconstruction of the original signal from the coefficients are then developed in both the critical sampling case and the oversampling case. Unified parallel lattice structures for the implementation of the algorithms are studied. And the computational complexity analysis and comparison show that the proposed algorithms provide a more efficient and faster approach for the computation of the discrete Gabor transforms.

This paper focuses on the underdetermined blind source separation (BSS) of three speech signals mixed in a real environment from measurements provided by two sensors. To date, solutions to the underdetermined BSS problem have mainly been based on the assumption that the speech signals are sufficiently sparse. They involve designing binary masks that extract signals at time-frequency points where only one signal was assumed to exist. The major issue encountered in previous work relates to the occurrence of distortion, which affects a separated signal with loud musical noise. To overcome this problem, we propose combining sparseness with the use of an estimated mixing matrix. First, we use a geometrical approach to detect when only one source is active and to perform a preliminary separation with a time-frequency mask. This information is then used to estimate the mixing matrix, which allows us to improve our separation. Experimental results show that this combination of time-frequency mask and mixing matrix estimation provides separated signals of better quality (less distortion, less musical noise) than those extracted without using the estimated mixing matrix in reverberant conditions where the reverberant time (TR) was 130 ms and 200 ms. Furthermore, informal listening tests clearly show that musical noise is deeply lowered by the proposed method comparatively to the classical approaches.

It is not always possible to find a global parse for an input sentence owing to problems such as errors of a sentence, incompleteness of lexicon and grammar. Partial parsing is an alternative approach to respond to these problems. Partial parsing techniques try to recover syntactic information efficiently and reliably by sacrificing completeness and depth of analysis. One of the difficulties in partial parsing is how the grammar might be automatically extracted. In this paper we present a method of automatically extracting partial parsing rules from a tree-annotated corpus using the decision tree method. Our goal is deterministic global parsing using partial parsing rules, in other words, to extract partial parsing rules with higher accuracy and broader expansion. First, we define a rule template that enables to learn a subtree for a given substring, so that the resultant rules can be more specific and stricter to apply. Second, rule candidates extracted from a training corpus are enriched with contextual and lexical information using the decision tree method and verified through cross-validation. Last, we underspecify non-deterministic rules by merging substructures with ambiguity in those rules. The learned grammar is similar to phrase structure grammar with contextual and lexical information, but allows building structures of depth one or more. Thanks to automatic learning, the partial parsing rules can be consistent and domain-independent. Partial parsing with this grammar processes an input sentence deterministically using longest-match heuristics, and recursively applies rules to an input sentence. The experiments showed that the partial parser using automatically extracted rules is not only accurate and efficient but also achieves reasonable coverage for Korean.

Data accessed by many sites are replicated in distributed environments for performance and availability. In this paper, replication schemes are examined in parallel image convolution processing. This paper presents a system architecture that we have developed with CORBA (Common Object Request Broker Architecture) for the processing. Employing CORBA enables us to make use of a cluster of workstations, each of which has a different level of computing power. The paper also describes a parallel and distributed image convolution processing model using replicas stored in a network of workstations, and reports some experimental results showing that our analytical model can agree with practical situations.

This paper presents a technique for designing a dielectric Electronically Steerable Parasitic Array Radiator (Espar) antenna to achieve miniaturization of the conventional Espar antenna. The antenna's size is reduced by immersing the central active element in a dielectric cylinder, mounting the surrounding planar parasitic elements at the circumference of the cylinder, and decreasing the radius of the ground skirt to that of the parasitic elements. An example of a polycarbonate (εr = 2.9 + j0.006) Espar antenna operating at 2.484 GHz is optimised by using a genetic algorithm in conjunction with an FEM-based cost function. The designed antenna generates a half-power beam width of 78and a main lobe that elevates at an angle of only 5from the horizontal plane. The designed antenna is also fabricated and measured. Good agreement between the measurement and simulation results is obtained. We reduce the size of the designed Espar antenna to 1/8 the size of its conventional counterpart while achieving a 12improvement in half-power beam width.

Frequency domain blind source separation has the great advantage that the complicated convolution in time domain becomes multiple efficient multiplications in frequency domain. However, the inherent ambiguity of permutation of ICA becomes an important problem that the separated signals at different frequencies may be permuted in order. Mapping the separated signal at each frequency to a target source remains to be a difficult problem. In this paper, we first discuss the inter-frequency correlation based method, and propose a new method using the continuity in power between adjacent frequency components of same source. The proposed method also implicitly utilizes the information of inter-frequency correlation, as such has better performance than the previous method.

A new technique called the Dual Parasitic Beam (DPB) technique is proposed to suppress grating lobes in a rectangular waveguide broad wall transverse slot array. This technique involves an extra layer of parasitic strip dipoles that generate the DPB to suppress the grating lobes without opposing the main beam of the original slot linear array. A full wave EM analysis in Method of Moments (MoM) is conducted to compute the coupling excitation coefficients as well as the far field patterns of the slot and dipole currents. Analysis shows that a suitable dimension and arrangement of dipoles are needed to get a desired level of dipole excitations to meet the grating suppression condition. It is found that the grating lobes can be suppressed as much as 15 dB in the presence of the parasitic dipoles. Experiments are conducted to confirm the computed results.

To guarantee the high reliability of video services, video servers usually adopt parity-encoding techniques in which data blocks and their associated parity blocks form a parity group. For real-time video service, all the blocks in a parity group are prefetched in order to cope with a possible disk failure, thereby incurring a buffering overhead. In this paper, we propose a new scheme called Round-level Parity Grouping (RPG) to reduce the buffer overhead while restoring VBR video streams in the presence of a faulty disk. RPG allows variable parity group sizes so that the exact amount of data is retrieved during each round. Based on RPG, we have developed a storage allocation algorithm for effective buffer management. Experimental results show that our proposed scheme reduces the buffer requirement by 20% to 25%.

High-Speed High-Resolution Comparators are integral parts of very high-speed high-resolution Analog-to-Digital Converters (ADC). Parallel successive-approximation and flash ADCS can boost conversion rates while providing high resolution, provided that accurate and fast offset-cancelled comparators could be implemented. Moreover, accurate offset cancellation is needed in accurate gain stages of other types of high speed ADCs as well. This has never been easy and creates a bottle neck for high-speed high-resolution ADCs. The reason is that conventional offset cancellation methods, suffer either from inaccurate cancellation or from slow operation. Hence, either speed or accuracy is compromised. This is due to the trade off of gain (accuracy) for bandwidth (speed) in conventional methods. Here, we introduce a new offset cancellation method which satisfies the need for both high-speed and accurate offset cancellation simultaneously.

A simple high-speed low input impedance CMOS current comparator is presented. The circuit uses improved Wilson current-mirror to perform subtraction. Negative feedback is employed to reduce the input impedance of the circuit. HSPICE is used to verify the circuit performance with standard 0.5 µm CMOS technology. Simulation results demonstrate propagation delay of 1.02 ns, average power consumption of 0.9 mW, and input impedance of 137 Ω for 0.1 µA input current at the supply voltage of 3 V.

This paper presents a long length discrete Hartley transform (DHT) design with a new hardware efficient distributed arithmetic (DA) approach. The new DA design approach not only explores the constant property of coefficients as the conventional DA, but also exploits its cyclic property. To efficiently apply this approach to long length DHT, we first decompose the long length DHT algorithm to short ones using the prime factor algorithm (PFA), and further reformulate it by using Agarwal-Cooley algorithm such that all the partitioned short DHT still consists of the cyclic property. Besides, we also exploit the scheme of computation sharing on the content of ROM to reduce the hardware cost with the trade-off in slowing down the computing speeds. Comparing with the existing designs shows that the proposed design can reduce the area-delay product from 52% to 91% according to a 0.35 µm CMOS cell library.

Recently, Das et al. proposed a dynamic ID-based verifier-free password authentication scheme using smart cards. To resist the ID-theft attack, the user's login ID is dynamically generated and one-time used. Herein, we demonstrate that Das et al.'s scheme is vulnerable to an impersonation attack, in which the adversary can easily impersonate any user to login the server at any time. Furthermore, we also show several minor weaknesses of Das et al.'s scheme.

A new edge-coloring algorithm for bipartite graphs is presented. This algorithm, based on the framework of the O(m log d + (m/d) log (m/d) log d) algorithm by Makino-Takabatake-Fujishige and the O(m log m) one by Alon, finds an optimal edge-coloring of a bipartite graph with m edges and maximum degree d in O(m log d + (m/d) log (m/d)) time. This algorithm does not require elaborate data structures, which the best known O(m log d) algorithm due to Cole-Ost-Schirra depends on.

We present a simple decoding algorithm that modifies soft bit-flipping algorithm for decoding LDPC codes. In our method, a new parameter is explored to distinguish the variables (symbols) belonging to the same number of unsatisfied constraints. A token is also assigned in the method to avoid repeated flipping of the same variable, rather than using a constant taboo length. Our scheme shows a similar computational load as the taboo-based algorithm, while having a similar decoding performance as the belief propagation algorithm.

The problems with the CV characterization on very leaky (thin) nitrided oxide are mainly due to the measurement precision and MOS gate dielectric model accuracy. By doing S-parameter measurement at RF frequency and using simple but reasonably accurate model, we can obtain proper CV curves for very thin nitrided gate dielectrics. Regarding the measurement frequency we propose a systematic method to find a frequency range in which we can select measurement frequencies for all biases to obtain a full CV curve. Moreover, we formulated the first order relationship between the measurement frequency range and the test structure design for CV characterization. With the established formulae, we redesigned the test structures and verified that the formulae can be used as a guideline for the test structure design for RFCV measurements.

"Can one walk to infinity on Gaussian primes taking steps of bounded length?" We adopted computational techniques to probe into this open problem. We propose an efficient method to search for the farthest point reachable from the origin, which can be parallelized easily, and have confirmed the existence of a moat of width k =, whereas the best previous result was k = due to Gethner et al. The amount of computation needed for k = is about 5000 times larger than that for k =. A refinement of Vardi's estimate for the farthest distance reachable from the origin is proposed. The proposed estimate incorporates discreteness into Vardi's that is based on percolation theory.

In this paper, we propose a new reinforcement learning scheme called CHQ that could efficiently acquire appropriate policies under partially observable Markov decision processes (POMDP) involving probabilistic state transitions, that frequently occurs in multi-agent systems in which each agent independently takes a probabilistic action based on a partial observation of the underlying environment. A key idea of CHQ is to extend the HQ-learning proposed by Wiering et al. in such a way that it could learn the activation order of the MDP subtasks as well as an appropriate policy under each MDP subtask. The goodness of the proposed scheme is experimentally evaluated. The result of experiments implies that it can acquire a deterministic policy with a sufficiently high success rate, even if the given task is POMDP with probabilistic state transitions.

Let G = (V,E) be a connected graph such that each edge e ∈ E and each vertex v ∈ V are weighted by nonnegative reals w(e) and h(v), respectively. Let r be a vertex designated as a root, and p be a positive integer. The minmax rooted-subtree cover problem (MRSC) asks to find a partition X = {X1,X2,...,Xp of V and a set of p subtrees T1,T2,...,Tp such that each Ti contains Xi∪{r} so as to minimize the maximum cost of the subtrees, where the cost of Ti is defined to be the sum of the weights of edges in Ti and the weights of vertices in Xi. Similarly, the minmax rooted-cycle cover problem (MRCC) asks to find a partition X = {X1,X2,...,Xp} of V and a set of p cycles C1,C2,...,Cp such that Ci contains Xi∪{r} so as to minimize the maximum cost of the cycles, where the cost of Ci is defined analogously with the MRSC. In this paper, we first propose a (3-2/(p+1))-approximation algorithm to the MRSC with a general graph G, and we then give a (6-4/(p+1))-approximation algorithm to the MRCC with a metric (G,w).

We report the empirically obtained conditions for the fusion splicing with photonic crystal fibers (PCF) having large mode areas. By controlling the arc-power and the arc-time of a conventional electric-arc fusion splicer, the splicing loss between two PCFs could be lowered down to 0.2 dB in average. For the splicing PCF with a conventional single mode fiber (SMF), the loss was increased due to the modal field mismatch, but still below 0.45 dB in average. The tensile strength was weakened by the splicing from 2.83 GPa down to 1.04 GPa for the PCF-PCF case and 0.89 GPa for the PCF-SMF one.