A real-world sensing application, which senses and analyzes the situation in the real world via sensor devices, has attracted increasing attention for providing new services to mobile users. In this paper, for targeting the application, we propose a highly adaptive resource control method that can control the amount of local computing resources, which is provided by the Mobile Edge Computing technology and the amount of network resources necessary for service provisioning. A basic idea is to express various information, such as the amount of sensor information and the amount of user access, into a simple potential field, and then update the potential field in a short cycle in a self-organized manner. Numerical results show that our resource control method based on the potential field is adaptive for the movement of users.

Virtualization technology can readily change the placement of an execution environment, such as a virtual machine (VM), in accordance with the current statuses. To effectively manage system resources in the virtualized infrastructure, the elasticity of a virtualized infrastructure has attracted significant attention for determining on which server VM should be deployed. However, there has been little attention to how long it takes to relocate VMs and in what order VMs migrate. It is difficult to determine the migration order of VMs in consideration of the residual resources of the destination server and network topology. In this paper, we propose two methods for scheduling the parallel migration of multiple VMs to reduce the impact on performance degradation due to VM migrations. We first formulate an integer linear programming (ILP) model to determine the VM migration order. To solve the problem in an acceptable time for large infrastructures, we also design a heuristic method based on dependencies between VM migrations and network bandwidth. Simulation results demonstrate that the heuristic-based method obtained the solutions in a few seconds for the settings with small differences with the optimal solutions obtained from the ILP-based method.

SDI (Software Defined Infrastructure) provides virtualized infrastructures to customers by slicing computing resources and network resources. One of the important problems for deploying SDI framework is to control the assignment of physical resources to a virtual network against changes of traffic demand and service demand. For this problem, VNE (Virtual Network Embedding) problem that maps a virtual network to physical resources has been addressed, but a centralized calculation was assumed. It is difficult to adopt the centralized approaches as a size of infrastructure increases and a number of VN requests increases. This is because the identification of current situation becomes more complicated. In this paper, we present a VNE method that works with a little information for the large, complicated, and uncertain SDI frameworks. To achieve this, the proposed method applies biological “Yuragi” principle. The term Yuragi is a Japanese word whose English translation means a small perturbation to the system. Yuragi is a mechanism of adaptability of organisms and is often expressed as attractor selection model. This paper develops Yuragi-based VNE method that deals with node attribute and has a generality of a performance objective and runs in multi-slice environments. Simulation results show that the Yuragi-based method decreases VN migrations by about 29% than a heuristic method to adapt the fluctuations in resource requirements.

With the increasing diversity of network services and applications, caching technologies for content delivery networks (CDNs) and information-centric networking (ICN) have become beneficial to improve the service quality such as communication latency by storing content data on network nodes near to users. Cache performance depends on the memory size at each node as well as the request distribution of content, and it becomes an important issue to analyze the performance of current caching mechanisms for designing efficient cache systems. In general, the request distribution in content dissemination networks is heavy-tailed, containing many objects called one-timers, which are accessed only once and are not hit while in the cache. Moreover, one-timers may diminish the whole cache performance by causing inefficient cache replacement of more popular content. To solve this problem, effective caching mechanisms having separate queues for one-timers, such as 2Q and adaptive replacement caching (ARC), have been proposed. In this paper, we focus on analyzing the 2Q and ARC mechanisms and propose approximation models that can statistically analyze the influence of memory size and request distribution on the cache performance. Furthermore, we evaluate the accuracy of both approximation models.

As various multimedia services are being provided on networks, broadband traffic is growing as well. Reducing traffic is important because power consumption in networks has been increasing year by year. Meanwhile, content caching is expected to reduce data traffic by storing content replicas on the network nodes and it is beneficial in view of energy efficiency. However, in order to realize an energy efficient network, it is necessary to allocate content replicas effectively in consideration of both power consumption of content caching and transmission of traffic. In this paper, we propose a design method which derives the energy efficient cache locations for content dissemination. Furthermore, we demonstrate the effectiveness of our proposed method.

This paper proposes a new auxiliary-graph-based green-lightpath provisioning algorithm for WDM networks that use optical regenerators. Energy efficiency in communication networks has become a critical issue recently since the total energy consumption of networks has been increasing year by year due to the growth of the Internet traffic. A lightpath introduces optical cut-through, which suppresses energy-inefficient IP packet processing at the intermediate nodes, and thus the total energy consumption of networks. An algorithm for dynamic lightpath establishment is a key factor in achieving high energy efficiency. When calculating a path, the energy consumption of optical regenerators, which are used at the intermediate nodes to guarantee optical signal quality, should be considered. To the best of our knowledge, however, the use of optical regenerators is not considered in previous studies on green-lightpath provisioning. The proposed auxiliary-graph-based algorithm can easily calculate the most energy efficient path considering the energy consumption of optical regenerators since the problem is transformed into the minimum weight path problem on the proposed auxiliary graph. Computer simulations show that our proposal reduces the blocking probability and the total energy consumption of a network.

With the progress of large-scaled networks, a service platform should manage effectively enormous amounts of diversified information to ensure stable operations of the platform systems in a resource-constrained environment. We propose a resource control mechanism based on a feedback control law to enable effective resource management and the stable operation of platform systems. Furthermore, we introduce the application cases for a session admission control system and a packet sampling control system, and demonstrate the effectiveness of the proposed mechanism in simulations and prototype evaluations.

This paper introduces a 51.2Gops, 1.0GB/s-DMA single-chip multiprocessor integrating quadruple cores and proposes a power integrity analysis. Our multiprocessor is designed to decode MP@HL streams without any dedicated circuits. To achieve such high performance, data throughput as well as processing capability is important, requiring a large number of high speed I/Os. However, this makes for a high level of power supply noise. We then applied an interface timing margin analysis tool that took power supply noise into account, and succeeded in putting reasonable restrictions on LSI design, as well as that for the printed circuit board. As a result, we succeeded in operating the processor at 533MHz with the 2ch 64bit main memory IF at 266MHz and 64bit system bus at 178MHz.

A rectangular cutting stock problem for flawed and connectable resources is a kind of a rectangular cutting stock problem where orders are assigned to resources. Each order has two types of size, width and a total length. The aim of the problem is that resources are divided into pieces, called "lumps", and the sum of length of the lumps whose width equals to a width of order satisfies a total length of the order, is decided. A resource is cut by "sets" which are combinations of fixed cutting blade for length direction, and cut end to end for width direction. Moreover, there are features that are flawed that cannot be assigned to orders existing on resources, and the connection that enables to cut several parts of resources by the same set is permitted. This research proposes the method that consists of two phases. In first phase, the rectangles called "available areas" which are generated by avoiding flaws and connecting, are created by deciding positions of set change according to an expected yield ratio and a number of set changes. In the second phase, assignment for each available area is done from an order with a larger width by using a branch and bound method in order to achieve a higher yield ratio. At this time, an effective degree of a set pattern is calculated for each available area. The effective degree means a rate of improvement of a yield ratio per length of an assigned target order. It is possible to decide an assignment rapidly by searching a set pattern that has a large effective degree earlier. The proposed method has been applied to a real cutting stock problem, and the effectiveness is confirmed.

The performance of several algorithms for the quantization of the mel-generalized cepstral coefficients is studied. First, the objective and subjective performance of two-stage vector quantization (VQ) is measured. It is shown that the subjective quality for the mel-generalized cepstral coefficients is higher than that for LSP. Secondly, interframe prediction is introduced in the encoding of mel-generalized cepstral coefficients. By utilizing interframe moving average (MA) prediction, the mel-generalized cepstral coefficients can be encoded more efficiently than LSP in terms of cepstral distortion. Finally, we implement a CELP coder based on mel-generalized cepstral analysis in which mel-generalized cepstral coefficients are quantized using MA prediction. This coder has a higher objective quality than conventional CELP.

We describe an approach to voice characteristics conversion for an HMM-based text-to-speech synthesis system. Since this speech synthesis system uses phoneme HMMs as speech units, voice characteristics conversion is achieved by changing the HMM parameters appropriately. To transform the voice characteristics of synthesized speech to the target speaker, we applied the maximum a posteriori estimation and vector field smoothing (MAP/VFS) algorithm to the phoneme HMMs. Using 5 or 8 sentences as adaptation data, speech samples synthesized from a set of adapted tied triphone HMMs, which have approximately 2,000 distributions, are judged to be closer to the target speaker by 79.7% or 90.6%, respectively, in an ABX listening test.

This paper presents a new text-to-speech synthesis system based on HMM which includes dynamic features, i.e., delta and delta-delta parameters of speech. The system uses triphone HMMs as the synthesis units. The triphone HMMs share less than 2,000 clustered states, each of which is modelled by a single Gaussian distribution. For a given text to be synthesized, a sentence HMM is constructed by concatenating the triphone HMMs. Speech parameters are generated from the sentence HMM in such a way that the output probability is maximized. The speech signal is synthesized directly from the obtained parameters using the mel log spectral approximation (MLSA) filter. Without dynamic features, the discontinuity of the generated speech spectra causes glitches in the synthesized speech. On the other hand, with dynamic features, the synthesized speech becomes quite smooth and natural even if the number of clustered states is small.

Chinese all syllables recognition is described. Chinese all syllables recognition is divided into base syllable recognition disregarding the tones and 4 tones recognition. For base syllable recognition, we used a combination of two multisegment vector quantization (MSVQ) classifiers based on different features (instantaneous and transitional features of speech). For the tones recognition, the vector quantization (VQ) classifier is first used, and is comparable to a multilayer perceptron (MLP) classifier. Next, a combination of a distortion based classifier (VQ) and a discriminant based classifier (MLP) is proposed. An evaluation has been carried out using the standard Chinese syllable database CRDB, and experimental results have shown that the combined classifiers can improve the recognition performance. The recognition accuracy for base syllable and tones is 96.48% and 99.82% respectively.

This paper proposes a technique for estimating the harmonic frequencies based on instantaneous frequency (IF) of a speech signal. The main problem is how to decompose the speech signal into the harmonic components. For this purpose, we use a set of bandpass-filters, each of whose center frequencies changes with time in such a way that it tracks the instantaneous frequency of its output. As a result, the outputs of the band-pass filters become the harmonic components, and the instantaneous frequencies of the harmonics are accurately estimated. To evaluate the effectiveness of the approach, we apply it to pitch extraction. The pitch extraction is simply accomplished by selecting the correct fundamental frequency out of the harmonic frequencies. The most significant feature of the pitch extractor is that the extracted pitch contour is smooth and it requires no post-processing such as nonlinear filtering or any smoothing processes.

We propose a CELP coder based on mel-cepstral analysis. In the coder, since the transfer functions of perceptual weighting and postfiltering are defined through mel-cepstral coefficients, the effects of perceptual weighting and postfiltering should fit with the characteristics of the human auditory sensation. We use a basic CELP structure without adaptive codebook, and the subjective speech quality of the proposed coder in terms of the opinion equivalent Q is measured and compared with that of the conventional CELP coder. It is shown that the improvement of more than 1.8 dB is achieved by the proposed coder over the conventional CELP coder.

This paper proposes an algorithm for speech parameter generation from HMMs which include the dynamic features. The performance of speech recognition based on HMMs has been improved by introducing the dynamic features of speech. Thus we surmise that, if there is a method for speech parameter generation from HMMs which include the dynamic features, it will be useful for speech synthesis by rule. It is shown that the parameter generation from HMMs using the dynamic features results in searching for the optimum state sequence and solving a set of linear equations for each possible state sequence. We derive a fast algorithm for the solution by the analogy of the RLS algorithm for adaptive filtering. We also show the effect of incorporating the dynamic features by an example of speech parameter generation.

This paper proposes an algorithm for speech parameter generation from continuous mixture HMMs which include dynamic features, i.e., delta and delta-delta parameters of speech. We showthatthe parameter generation from HMMs using the dynamic features results in searching for the optimal state sequence and solving a set of linear equations for each possible state sequence. Tosolve the problem, we derivea fast algorithm on the analogy of the RLS algorithm for adaptive #ltering. We show that the generated speech parameter vectors re#ect not only the means of static and dynamic feature vectors but also the covariances of those. An example presenting e#ectiveness of the proposed algorithm in speech synthesis is given. 1. INTRODUCTION The hidden Markov models #HMMs# can model sequences of speech spectra with well-de#ned algorithms, and have successfully been applied to speech recognition systems. From these facts, we surmise that HMMs are also useful for speech synthesis. Actually, some at...

This paper proposes an algorithm for adaptive cepstral analysis based on the UELS (unbiased estimation of log spectrum). In the UELS, the model spectrum is represented by cepstral coefficients and the mean square of the inverse filter output is minimized with respect to the cepstral coefficients. By introducing an instantaneous gradient estimate of the criterion in a similar manner of the LMS algorithm, we develop an adaptive cepstral analysis algorithm. In the analysis system, an IIR adaptive filter whose coefficients are given by cepstral coefficients is realized using the log magnitude approximation (LMA) filter. The filter approximates an exponential transfer function and its stability is guaranteed for approximation of speech spectra. To implement the M th order cepstral analysis, the algorithm requires O(M) operations per sample. It is shown that the algorithm has fast convergence properties in comparison with the LMS algorithm. Several examples of the adaptive cepstral analysis for synthetic signal and natural speech are shown to demonstrate the effectiveness of the algorithm.

We propose an ADPCM coder which uses a backward adaptive predictor based on the adaptive mel-cepstral analysis. The spectrum represented by the mel-cepstral coefficients has frequency resolution similar to that of the human ear which has high resolution at low frequencies. In the coder, since the transfer functions of noise shaping and postfiltering are also defined through the mel-cepstral coefficients, the effects of nose shaping and postfiltering should fit with characteristics of the human auditory sensation. We incorporate a pitch predictor into the ADPCM coder, and evaluate the speech quality based on objective and subjective performance tests. It is shown that the coder at 16 kb/s can produce a high quality speech comparable with that of the CCITT G.721 ADPCM coder at 32 kb/s with no algorithmic delay.>

The authors describe a mel-cepstral analysis method and its adaptive algorithm. In the proposed method, the authors apply the criterion used in the unbiased estimation of log spectrum to the spectral model represented by the mel-cepstral coefficients. To solve the nonlinear minimization problem involved in the method, they give an iterative algorithm whose convergence is guaranteed. Furthermore, they derive an adaptive algorithm for the mel-cepstral analysis by introducing an instantaneous estimate for gradient of the criterion. The adaptive mel-cepstral analysis system is implemented with an IIR adaptive filter which has an exponential transfer function, and whose stability is guaranteed. The authors also present examples of speech analysis and results of an isolated word recognition experiment.>

The authors improve the modeling of the conventional continuous hidden Markov model (HMM) for speech recognition in two aspects. One is to use new functions to model the phonetic duration distribution. These functions can well approximate the unsymmetrical distribution and have relatively simple forms for computation. The other aspect is to use a proportional coefficient to adjust dimensional effects of the output density functions and the phonetic duration functions in the HMM. Using these new techniques, the authors got 7.8% improvement of the recognition correct rate in the vowel recognition experiments of the continuous speech.>

A technique for designing two-dimensional (2-D) digital filters which can approximate arbitrary magnitude functions is proposed. A 2-D spectral factorization technique is used to obtain the recursively computable and stable system with nonsymmetric half-plane support from a given 2-D magnitude function. A new class of realizable 2-D digital filters referred to as 2-D log magnitude approximation (2-D LMA) filters are used to approximate the system obtained by the 2-D factorization. The design procedure is straightforward and computationally efficient. A simple stability condition which guarantees the stability of the designed 2-D LMA filter is given. An efficient network structure of 2-D LMA filters for parallel implementation is also discussed.>

A technique for designing IIR (infinite impulse response) filters with a complex-valued frequency response is proposed. The complex logarithmic error, whose real and imaginary parts represent the log magnitude and phase errors, respectively, is minimized in the Chebyshev sense using an iterative weighted-least-squares (WLS) technique in the frequency domain. Both the log magnitude and phase errors of the resulting filter become nearly equiripple. Filter coefficients at each iteration are efficiently computed using a fast recursive algorithm for a set of linear equations which is derived from the WLS problem.>

An adaptive cepstral analysis method based on an unbiased estimation of the log spectrum is proposed. In the method, an infinite impulse response adaptive filter whose coefficients are given by cepstral coefficients is realized using the log magnitude approximation (LMA) filter. To implement the Mth-order cepstral analysis, the algorithm requires O(M) operations per sample. It is shown that the algorithm has fast convergence properties in comparison with the least-mean-square algorithm. A real-time analysis system is implemented with a general-purpose digital signal processor, and an example of natural speech analysis is shown to demonstrate the convergence.>

A technique is proposed for designing autoregressive-moving average (ARMA) digital filters to have an arbitrary log magnitude frequency response. The technique is based on an iterative weighted least-squares (WLS) approach in the frequency domain. A weight updating procedure is introduced to obtain a least-squares approximation to the given log magnitude function using the WLS approach. Filter coefficients are efficiently calculated using a fast recursive algorithm for a set of linear equations derived from the WLS problem. The technique is also extended to equiripple approximation with a minor modification of the weight updating procedure.>

This paper presents a new technique of cepstral analysis synthesis on the mel frequency scale, the log spectrum on the mel frequency scale (the mel log spectrum) is considered to be an effective representation of the spectral envelope of speech. This analysis synthesis system uses the mel log spectrum approximation (MLSA) filter which was devised for the cepstral synthesis on the mel frequency scale. The filter coefficients are easily obtained through a simple linear transform from the mel cepstrum defined as the Fourier cosine coefficients of the mel log spectral envelope of speech. The MLSA filter has a low coefficient sensitivity and a good coefficient quantization characteristics. The spectral distortion caused by interpolation of the filter parameters of two successive frames is small. Accordingly, the data rate of this system is very low. The same quality speech is synthesized at 60-70 % of data rates in the conventional cepstral vocoder or the LPC vocoder.

Natural and intelligible Japanese was synthesized from consonant-vowel (CV) syllable parameters. Japanese has approximately one hundred CV type syllables which are represented by Japanese alphabet. Each CV syllable is analyzed and stored in the form of cepstrum corresponding to a true log spectral envelope. The speech synthesis system converts character strings into the cepstral parameters with an appropriate dynamic transition from one syllable to the next, and generates a smooth pitch pattern from some pitch informations. In this system, the log magnitude approximation (LMA) filter is employed as parametric synthesizer. The LMA filter generates an acoustic waveform from the cepstrum in real time.

By using the log magnitude approximation filter as the speech synthesizer, a low bit rate cepstral vocoder was obtained. The synthesis filter is of minimum phase and of pole-zero type, and this vocoder can provide a better spectral fit than that provided by all-pole model. The filter can be easily obtained without transforming the cepstrum to an impulse response and the computation for the filter coefficients is non-iterative. The spectral envelope information is transmitted in a form of cepstral values. Since the coefficient sensitivities of the synthesis filter is sufficiently small, a fairly coarse quantization of the cepstrum is allowed. A reasonable speech quality vocoder was realized at the overall bit rate of 1770 bits/s.