Recent progress in motor control suggests that in controlling unstable systems humans switch intermittently between the passive and active behavior instead of controlling the system in a continuous manner. Traditionally, the models of intermittent control employ the notion of threshold to mimic control switching mechanisms in humans. The notion of noise-driven control activation developed here provides a richer alternative to the conventional threshold-based models of intermittent motor control. We show that the model implementing noise-driven activation matches the experimental data on human balancing of virtual overdamped stick. Our results suggest that the stochasticity of the control activation mechanism is a fundamental property and may play an important role in the dynamics of human-controlled systems.

View full abstract

The author has already presented stochastic optimal tracking control theory with preview for linear continuous-time Markovian jump systems on the finite time interval by state feedback ([21]). He has also presented optimal estimation theory for linear continuoustime Markovian jump systems on the finite time interval ([22]). In this paper we study stochastic optimal tracking control theory with preview for linear continuous-time Markovian jump systems on the finite time interval by output feedback. The necessary and sufficient conditions for the solvability of the stochastic optimal tracking problem with preview by state feedback are given by coupled Riccati differential equations with terminal conditions. In order to decide the gains of output feedback controllers, we need solutions of another type of coupled Riccati differential equations with initial conditions.

View full abstract

In this paper, we are concerned with a problem of optimization of the linear observations which are used in the stationary Kalman-Bucy filter. Especially, we consider the optimization of the gain matrix in the observation. In the previous works of the author, the corresponding problem for discrete-time systems was already considered and the condition of optimality was obtained. This paper is concerned with the case of the continuous-time systems and it is shown that the condition of optimality is the same as the discrete-time case except for the accompanying Lyapunov equation is continuous-type whereas it was discrete-type in the discrete-time problem. We propose a method of solving the set of equations of the Riccati equation for the error covariance and the condition of optimality by a simple recursive algorithm. The results of numerical experiments show the efficiency of the algorithm.

View full abstract

In this article, we are concerned with the non-linear filtering problem for the signal process of Markov type with the observation of discontinuous semimartingales. Through the extended Girsanov transform of observation processes, we consider processes of unnormalized filtering measures so that the filtering problem reduces to analyze Markov processes with generators affected by observations. This way makes it possible to use the Malliavin calculus to show the existence of smooth densities of filtering measures. Processes of unnormalized filtering measures are characterized by linear functional SDE (Zakai equations) whose coefficients are integrodifferential operators. In the case where noises of the signal and the observation are independent, by a simple replacement of solutions, the associated Zakai equation can be transformed to a parabolic integro-differential equation with coefficients affected by the observation. This transform would be useful to prove the uniqueness of solutions to Zakai equations. It is possible to generalize these results to some cases where noises of the signal and the observation are correlated. Under regularity conditions for the existence of stochastic flows, by certain replacement of solutions using stochastic flows, those Zakai equations also can be transformed to parabolic integro-differential equations.

View full abstract

In this paper we study the H_∞ state estimation problems for a class of linear continuous-time systems with impulsive effects, stochastic uncertainties and discretetime observation on the finite time interval. The systems include linear stochastic discrete-time systems. We adopt game theoretic and variational approach, and derive, the impulsive Riccati equation which gives the necessary condition for the solvability of the H_∞ estimation problems, the estimates and the form of H_∞ estimators by calculating the stochastic first variation of the performance index under the dynamics constraint. The stochastic variational approach is equivalent to that for control systems in [22, 23] form the point of view of target tracking.

View full abstract

In this paper we study the optimal state estimation problems for a class of linear discrete-time Markovian jump systems. We adopt maximum likelihood (ML) approach and stochastic variational calculus method to derive forms of dynamic estimators on the fixed time interval. The necessary condition for the solvability of the optimal state estimation problems are given by the coupled Riccati difference equations with initial conditions.

View full abstract

In this paper, a new filter called Gaussian unscented filter (GUF) is proposed to address state filtering problems for nonlinear non-Gaussian state-space models. The GUF is similar to the unscented Kalman filter (UKF) in terms of both using the unscented transformation (UT) technique, however, the GUF is not a Kalman-type filter which provides linear minimum mean state estimates. The GUF is a sigma point realization of a Bayesian filter which sequentially estimates the prior and posterior probability density functions of states. It is therefore capable of handling non-Gaussian noises while the UKF is not. The GUF is also similar to the Gaussian particle filter (GPF) but the number of samples called sigma points necessary in the GUF algorithm is far less than the number of particles required in the Gaussian particle filter (GPF). The GUF is therefore expected to be superior in terms of the calculation cost viewpoint. We compare the state estimation accuracy and calculation time of the GUF with those for the UKF, GPF and the other particle filter using a numerical example.

View full abstract

The dynamics of heart rhythms have been widely studied because of the key aspects in the physiology of living beings. Mathematical modeling of heart rhythms is the objective of many research efforts. Since the component elements exhibit oscillatory behavior, they can be modeled as nonlinear oscillators. Gois et al.[1] have recently proposed a mathematical model to describe heart rhythms considering three-coupled Van der Pol oscillators. In this paper, we constitute the particle filter and the ensemble Kalman filter from this model to detect the cardiac diseases by estimating the coupling parameters. In particular, we use two types of ECG data calculated by the discrete Van der Pol oscillators, which are the initial-time responses including the transient state and excluding the transient state. Comparing the particle filter with the ensemble Kalman filter, we indicate that the latter has better estimation performance.

View full abstract

This paper describes a method for the detection of the state of the indoor environment using the change of the strength of the wireless LAN signals. Data acquisition of the physical location of the target person by a monitoring system is important for the person's safety and reliability. For the monitoring system, camera devices are widely used because of the width of view; however, it is difficult to apply the camera system to monitor the person in the environment because of the privacy concerns. The sensing system that can obtain the wide range sensing area but cannot obtain the shape of the target objects or persons directly is desired. We focus on change of the pattern of the wireless LAN signal strength based on the singular spectrum transformation of the time series of of the wireless LAN signal strength. We apply the singular spectrum transformation to detect the change of the strength of wireless LAN signal patterns. Experimental results show the feasibility of the method.

View full abstract

A signal processor which is suitable for distilling a melody played by a particular instrument from a music ensemble is discussed in this report. The processor separates a monaural music sound into individual tones of every pitch with a particular timbre, combinations of which provide melodies played by an instrument. A mixture of strongly correlated sounds such as the same pitch tones by different instruments can be processed by the proposed technique that is a windowed Wiener filter based on current statistics of the mixture sequentially evaluated in the frequency-domain.

View full abstract

In this study, we propose a method for classifying echocardiograms into different tissue types on the basis of brightness. In general, the echo level between different electrocardiograms changes because the doctor or clinical examiner has to make adjustments according to the imaging conditions. It is therefore difficult to obtain the same echo level (i.e., brightness value) for the same tissue across images. In order to reliably identify tissue types, classification algorithms need to be independent of changes in the echo level. We here propose a system in which tissue can be correctly classified, independent of the overall echo level, using self-organizing maps [1]-[3].

View full abstract

A multiuser Orthogonal Frequency Division Multiplexing (OFDM) system with perfect channel knowledge and limited power at base station is studied to maximize the average system throughput. Data rate and bit error rate (BER) can be adaptively changed with the channel variation. An efficient and low complexity algorithm for subcarrier and power allocation in OFDM systems is proposed in this paper. We consider an OFDM downlink system where each subcarrier is used by only one user. Since our resource allocation problem is non-convex, it is not easy to find an optimal solution. Our approach is to consider a user iterative water filling algorithm (IWF), which is concave, and find a numerical method to approximate the optimal solution for our non convex problem. Our results show that each subcarrier is allocated to a user by taking into account channel gains and interference caused by other users.

View full abstract

This paper deals with a consensus problem for multiagent networks with noise-corrupted measurements. The networks are assumed to have antagonistic interactions that represent a kind of competition between agents. The objective of this paper is to find a condition under which global consensus on the states of the agents is achieved. In particular, a rigorous stopping rule is derived for the consensus algorithm. It gives a certain number of updates so that all of the states get close to the average value of them within a specified tolerance and with a specified probability.

View full abstract

In this paper, H₂ control is considered for sampleddata systems with variable sampling rate, where the H₂ control performance is introduced according to a standard sampled-data setting. A discrete-time H₂ control problem is employed for solving the original problem. Its solvability condition is then established as a parameter-dependent linear matrix inequality. A probabilistic approach is taken for coping with the parameter-dependency.

View full abstract

Distributed optimization methods have been studied recently motivated by emerging applications in smart grid, multi-robot, etc. In most of the studies, convexity and smoothness of the objective and constraint functions are assumed while such assumptions are not always made in practice. This paper presents a distributed particle swarm optimization algorithm based on primal-dual decomposition architectures so that any gradient information is not needed. To investigate the potential of the presented algorithm, some simple numerical examples are provided while comparing with the conventional distributed primal-dual perturbation algorithm.

View full abstract

In this paper, a demand side management (DSM) technique for minimizing the peak load of the electric grid is proposed. The grid optimization problem is formulated and analyzed to find the optimized strategies. A smart power system with multi-residential end-users connected to a single energy source is considered. The end-users attempt to minimize peak-to-average ratio (PAR) of the electric grid by using energy scheduler (ES) as well as their energy production and storage strategies. Some end-users posses renewable energy sources (RESs), distribution generators (DGs), storage devices or both. Users charge their storage devices at low-demand periods and discharge them at high-demand periods. Likewise, end-users produces electricity during the peak hours to minimize their energy cost.

View full abstract

We study the long term portfolio which is able to track a liability. The portfolio optimization problem is defined as the stochastic optimal control problem and the performance criterion is the lower mean square error between the liability and our wealth. We impose constraints for the portfolio weights and obtain the optimal portfolio strategy numerically by solving the Hamilton-Jacobi-Bellman equation applying the quadratic approximation scheme. The numerical simulations using the empirical data provided by Japanese organizations are run under the two types of constraints: the no-short-selling constraint; the upper bound constraint for the portfolio weights. The former demonstrates that the liability tracking ability of our optimal portfolio strategy does not drop if we restrict the short selling. The latter implies the imbalance between the growth rate of the liability and the profitability of the assets.

View full abstract

In this paper, the author numerically investigates the convergence properties of estimators concerned with expectations for a class of fuzzy random sets, where the fuzzy random set is considered as a model of the capricious vague perception of a crisp phenomenon or a crisp random phenomenon. First, fuzzy random sets as vague perceptions of crisp phenomena and their extended one as vague perceptions of random phenomena are reviewed. Secondly applying the standard strong law of large numbers(SLLN) for the random elements in a separable Banach space, the convergence property of estimators for expectations of random fuzzy sets is examined, and finally they are confirmed numerically by simulation studies. The proposed fuzzy random sets are expected to apply to the social system modelings including the mathematical finance, where many data have both the properties of vagueness and randomness.

View full abstract

The survival prediction has become important for patient cancer treatment plan. Previous studies have shown that there is a relation between patient gene expression profile and the survival information. Due to the advancement in microarray technology, high dimensional expressions have been generated. These high dimension microarray data contain many uninformative genes. One of the solutions is to apply a gene selection method to select a small subset of informative genes. A Random Survival Forest method has been selected to perform gene selection for survival prediction. In this method, a score for gene ranking is the key factor to select the smallest possible of gene subsets with lower error rates. Hence, an improved gene selection method using Random Survival Forest has been proposed to obtain gene subsets with a lower error rate. The proposed idea is based on average gene scores for gene ranking. The results showed that the improved gene selection method obtained a lower error rate compared to the standard gene selection methods. Furthermore, the selected subset of genes could assist the predictive ability for survival prediction.

View full abstract

A brain-computer interface (BCI) is a communication system that uses as input the EEG (electroencephalography) of humans, to operate a computer without using a keyboard or the like. The purpose of BCIs is one of the communication tool of handicap users. In this paper we experimented with a hybrid BCI system using EEG of right and left hand motor imagery (MI) and steady-state visual evoked potential (SS-VEP). We examined the feature value performs frequency analysis for the experimental data. Furthermore, we evaluated the linear discriminant method as optimal feature extraction method. The identification rate obtained using a Bayesian classifier. As the result, all subjects was 90 % identification rate by linear discriminant method.

View full abstract

In this paper, the receding horizon optimal control problem for the stochastic logical dynamical systems with finite state is considered. An algebraic form of dynamic programming algorithm for the finite horizon optimal control problem is given. As an application, the control problem of the residual gas fraction (RGF) in engine is described on the stochastic logical dynamical systems framework to improve engine performance and reject imbalances on cycle-to-cycle variation. Based on the underlying physics of RGF, a logical model-based RGF control algorithm is developed with variable valve timing (VVT) as control actuator. Finally, to demonstrate the performances of the proposed control scheme, the experiment is expressed.

View full abstract

This paper discusses the stochastic evolutions of the state trajectories of a stochastic Single Machine Infinite Bus (SMIB) system. The dynamical equation of the SMIB system assumes the structure of a second-order non-linear differential equation. That is described as a power swing equation. After accounting for the stochasticity in the machine swing equation as well as accomplishing the phase space formulation, we are led to a vector stochastic differential equation. Importantly, the Itô stochasticity as well as the Stratonovich stochasticity coincide for the stochastic swing equation, since the Wang-Zakai correction term of the stochastic integral will vanish for the special case. Despite universality of the second-order non-linear machine swing equation in theoretical studies, e.g. power system dynamics, circuits and systems literature, the stochasticity of the machine swing equation is not analysed in a greater detail yet. In this paper, we analyse the stochasticity of the swing equation after utilizing the notion of the stochastic evolution of a scalar function of vector stochastic processes in combination with conditional expectation. Note that the stochastic state vector of the machine swing equation of this paper is a Markov process.

View full abstract

A fair simple car driving simulator was created based on the open source engine TORCS and used in car-following experiments aimed at studying the basic features of human behavior in car driving. Four subjects with different skill in driving real cars participated in these experiments. The subjects were instructed to drive a car without overtaking and losing sight of a lead car driven by computer at a fixed speed. Based on the collected data the distributions of the headway distance, the car velocity, acceleration, and jerk are constructed and compared with the available experimental data for the real traffic flow. A new model for the car-following is proposed to capture the found properties. As the main result, we draw a conclusion that human actions in car driving should be categorized as generalized intermittent control with noise-driven activation. Besides, we hypothesize that the car jerk together with the car acceleration are additional phase variables required for describing the dynamics of car motion governed by human drivers.

View full abstract

In this paper, we present methods of detecting and correcting Doppler biases which cause positioning errors in Doppler-aided Kalman filter-based GNSS (Global Navigation Satellite System) positioning. It is well known that Doppler frequency shift observables are more precise compared with C/A code pseudorange observables. Doppler shifts are transformed into range-rates and utilized to smooth code pseudoranges. And Doppler shift observables are immune to cycle-slip and continuously available as long as the receiver can track the signals. The precise raw Doppler shift observables can be further exploited for RTK (Real Time Kinematic) positioning to provide robust carrier-phase anomaly detection and correction processes [5, 6]. Therefore Doppler shift observables can be utilized on a priority basis in Doppler-aided Kalman filter-based positioning than C/A code pseudoranges. However unexpected Doppler bias errors cause position errors or route departures in the positioning. If the Doppler shifts include bias errors, they can be detected by monitoring the innovation processes in Kalman filter. And then they can be corrected by the exclusion and estimation methods.

View full abstract

In this paper, we present regional ionospheric delay models for global navigation satellite system (GNSS) and estimation methods of the ionospheric delay. The ionospheric delay models derived in this paper are based on the spherical cap harmonic analysis (SCHA). The SCHA coefficients are estimated by the least squares method and the Kalman filter and the ionospheric delay can be calculated from the estimated coefficients. In the experiments, the proposed models by the least squares method and the Kalman filter are compared with Klobuchar model. As a result, the effectiveness of the model estimated by Kalman filter is shown.

View full abstract

In this paper we present a carrier-phase-based PPP algorithm with multiple antennas which are disposed with solid geometrical distances and with the common receivers' clock errors, based on GR models. We present GR equations for multiple antennas, and derive the Kalman filtering formula of very precise point positioning. We also discuss on the algorithms of the so-called GNSS(GPS) Gyro by obtaining baseline vecters among antennas and the rotational matrix. The experimentals are shown by using results of the large triangle of antennas' positions: Kyoto, Okinawa and Hokkaido areas, GNSS Earth Observation Network System (GEONET) of Japan.

View full abstract

In this paper, practical methods for improving GPS-PPP (Global Positioning System–Precise Point Positioning) accuracy in urban canyons are presented and examined in real severe circumstances. In our previous work, the method for improving positioning accuracy is presented by using a barometric pressure sensor. And the usefulness of the method was shown by the experiments such that the number of satellites used in the position calculation was artificially decreased to 3. In this paper, the algorithm proposed in our previous work is tested for the automobile which runs in real severe circumstances such as near the skyscrapers, raised expressways and the feasibility and usefulness of the proposed method with few satellites are shown.

View full abstract

This research investigates a non-invasive pupil reflection measurement system. The proposed measurement system includes a pupil area calculation function by using image recognition from the images captured by a camera, and it also has a control function of displaying both light stimulus and gazing point on the monitor. The proposed system can be used with simple instructions. The experimental results show that the characteristic of the pupil reflection obtained by the proposed system is similar to that by the conventional system.

View full abstract

Generally, kernel support vector machines have several hyper-parameters which have to be determined by appropriate estimation methods. Usually, such hyper-parameters are fixed to same values for all learning samples. In this paper, we propose a novel kernel function based on usual radial basis function. We choose seeds of kernel function from learning samples at random, and we set randomized hyper-parameters for each kernel seed individually.

View full abstract

A new personal identification system is proposed by using the Random Input Password (RIP). This ID system does not require any specific codes to memorize, thus has a potential to flee us from the flood of increasing number of passwords. This RIP is a random string made by individual person by hitting one of the four contact keys by L times. Thus the RIP is different every time. However, the RIP carries peculiarities of individuals and this feature is the essence of this system to propose. At the registration stage, each person register the RIPs by N times consecutively. The identifier is constructed from the registered N RIPs, by averaging each element of 8 parameters. Those 8 parameters, P1, P2, P3, T1, T2, T3, DF, RD, are selected in order to depict characteristics of individuals sufficient to distinguish a specific person from the set of all. The average accuracy evaluated by using 9 individuals was that the Type-I error rate is 23% and Type-II error rate is 19% on the average.

View full abstract

Linear discrete-time stochastic systems with partially unknown system matrices are considered. Assuming that the stochastic systems are partially observable, we present a new class of pseudomeasurements modified from the one given in the preceeding paper which are fictitious and additional observation processes on the unknown entries of the system matrices. Then we propose a new class of iterative methods which identify those unknown entries and the state vectors simultaneously. A numerical example is given to demonstrate the advantages of the proposed approach.

View full abstract

The paper studies a Langevin equation for a non-Markovian Brownian motion with the Basset memory term and a non-Basset one. The exact expression of the velocity correlation function, the diffusion constant and the mean square displacement are obtained analytically. These results provide the baseline information on the theoretical and experimental analysis in complex stochastic molecular motions in viscoelastic media.

View full abstract

Within the rigor typical for physical models a new type non-symmetric diffusion problem is considered and the corresponding Brownian motion implementing such diffusion processes is constructed. As a particular example, random walks with internal causality on a square lattice are studied in detail. By construction, one elementary step of a random walker on the lattice may consist of its two succeeding jumps to the nearest neighboring nodes along the x- and then y-axis or the y- and then x-axis ordered, e.g., clock-wise. It is essential that the second fragment of elementary step is caused by the first one, meaning that the second fragment can arise only if the first one has been implemented, but not vice versa. In particular, if for some reasons the second fragment is blocked, the first one may be not affected, whereas if the first fragment is blocked, the second one cannot be implemented in any case. As demonstrated, on time scales much larger then the duration of one elementary step these random walks are characterized by a diffusion matrix with non-zero anti-symmetric component, which is also justified by numerical simulation.

View full abstract

We analyzed the skill of hula hooping motion using a Bayesian network to build a model of human periodic motion. Many human movements are periodic motions; e.g., walking and running. Therefore, an analysis method for the characterization of a periodic motion is needed. We chose the rotation of a hula hoop as periodic motion because it is easy to discriminate failure or success for this task. We created our model with the intention it be used in rehabilitation and sports instruction. Physical models are popular for motion analysis. However, they require the use of large-scale measurement systems for each application; e.g., three dimensional motion analysis, electromyograph sensing, and force-plate analysis, all require large research laboratories. Physical models are thus not always suitable. We solved this problem by building a causal relation model to focus on important body parts and important phases of movement for instruction using the Bayesian network. We found that the model can determine the skills of periodic motion.

View full abstract

Since each creature interacts with each other through a food chain, an analysis of the food chain is a crucial problem in an ecosystem. In this paper, we consider the plankton-fish system consisting of fish, zooplankton and phytoplankton, and study the influence of the random noise on the stability of the plankton system under fish school by numerical simulations. The qualitative change of the solutions of differential equations by variation of a system parameter is generally called a bifurcation. This paper is concerned with the bifurcation analysis of a stochastic plankton-fish system. As major analytical methods for the stochastic bifurcation, the P-bifurcation (Phenomenological bifurcation) and the D-bifurcation (Dynamical bifurcation) approaches are cited. The P-bifurcation studies a stationary measure corresponding to the one-point motion, while the D-bifurcation considers the stability of invariant measures. Since the P-bifurcation approach is based on the one point-motion, there is a possibility to miss certain branches in bifurcation diagrams. So, we study the bifurcation in the stochastic plankton-fish system using the D-bifurcation. By the numerical simulations, we show that the D-bifurcation does not occur in the stochastic plankton-fish system considered here.

View full abstract

In this paper a generalized hyper gamma distribution (GHGD) is proposed as the most generalized unimodal skewed probability density function (PDF). Two methods of the parameter estimation for the GHGD by the moment and the maximum likelihood estimator are presented. It is also shown that the the GHGD covers three types of important PDFs for the inverse temperature processes of the human heart beat R-R intervals in the context with the Beck-Cohen superstatistics.

View full abstract

The present work reports the results of our experiments aimed at estimating the distribution of the response delay time in human intermittent control over unstable mechanical systems. A novel experimental paradigm: balancing an overdamped inverted pendulum was used; the overdamping eliminates the effects of inertia and, therefore, reduces the dimensionality of the system. The created simulator of balancing a virtual pendulum by a human operator via computer mouse movement makes the pendulum (stick) invisible when the angle between it and the upward position is less than 5^o. It enabled us to measure directly the delay time as the time lag between the moment when the pendulum becomes visible and the moment when a subject starts to move the mouse. Eight persons (male students) were involved in the experiments. The collected experimental data are presented in the form of the delay time histograms. For the analyzed system it is demonstrated, in particular, that (i) the response delay time may be treated as a random variable distributed within a wide interval. Its lower boundary is estimated as less than 50 ms, which corresponds to the limit delay time determined by human physiology. The upper boundary is estimated as 500-600 ms, which is about typical values of the response delay time when a controlled system exhibits complex dynamics. Besides, the obtained results enable us to hypothesize that the response delay in human intermittent control may be determined by cumulative actions of two distinct mechanisms, automatic and international, endowing it with complex nonlinear properties.

View full abstract