In this study, we constructed a minimal model to clarify the role of the animal body and motor neurons in the generation of synchronized periodic motions. The novel contribution of this study is the analysis of the phenomenon of adaptive muscle synchronization that occurs only by inter-muscular interaction through the body dynamics. The proposed model is based on the primitive local feedback in animal muscles, which is called the stretch reflex. We conducted theoretical analysis and simulations using a single muscle model and its combinations (an antagonistic muscle pair and a limb model). The results of the theoretical analysis revealed the necessary and sufficient condition for limit cycle generation with resonance frequencies. Without any explicit neural connection between the muscle modules, the proposed simple model suggests that musculoskeletal systems can automatically generate resonant oscillation. Moreover, the resonance behavior is adaptive to changes in the body parameters of these systems.
An application of nonlinear control to enhance the operation of dynamical systems is found in the literature. In particular, recent research achieved harmonics reduction by driving a single-phase AC/AC converter to a chaotic regime. However, the periodic symmetry of the output voltage is broken during this process. This paper proposes two single-phase AC/AC converters with a symmetry-recovery feature that allows the converters to work either in a chaotic or stable regime. The proposed method, which is based on the extended delay feedback control, found a direct impact to recover the symmetry on the distribution of the power density through the time-delayed input.
The need for safe and secure networked motion control systems have been growing with the increase in the number of cyberattacks against critical infrastructure. This paper proposes three methods of tamper detection: the maximum delay (MaxD), minimum delay (MinD), and data thresholding (DT) methods, for a networked motion control system. The system has redundant forward network paths from a controller to an actuator and a path selector on the actuator side. The selector chooses an appropriate path without tampering. Error compensation schemes in the path change are also proposed. In the experiments using the three redundant forward paths, the proposed tamper detection methods are compared. Experimental results show that the proposed methods can achieve stable operation of the system even if one of the paths is exposed to data tampering attacks, while each method has advantages and drawbacks.
This paper proposes a remote congestion controller with the butterfly-shaped perfect delay compensator (PDC) for time-delay compensation in active queue management (AQM) supporting transmission control protocol (TCP) flows. The butterfly-shaped PDC does not need any time-delay model of TCP/AQM network. Simulations show that the proposed controller with the butterfly-shaped PDC effectively stabilizes the TCP/AQM network even if the system includes time-varying delays.
This paper focuses on an image sensor communication system that uses an LED as the transmitter and a high-speed image sensor (camera) as the receiver. Communication in this scheme depends on the quality of images transmitted from the LED to the sensor. If the image becomes unfocused on the way to the receiver, the LED luminance that make up the signal cannot be detected, so the receiver cannot demodulate the signal data. To overcome this problem, this study proposes a novel demodulation scheme to recover data from a degraded image, based on a maximum likelihood decoding (MLD) algorithm. The proposed method creates template images that imitate all possible blinking patterns produced by the LED transmitter, and then calculates the Euclidean distances between pixels in the captured image and the pseudo images for all possible blinking patterns. Finally, the algorithm chooses the image template with the smallest Euclidian distance from the received signal as the recovered data. Though an exhaustive set of image templates must be prepared for the proposed MLD, the number of templates depends on the number of LEDs on the transmitter. Thus, the computational complexity of this method increases as the number of transmitter LEDs increases. To reduce the computational complexity of the proposed MLD algorithm, the binary differential evolution (BDE) algorithm is used, which is a swarm intelligence technique. Computer simulations are used to evaluate the BDE algorithm's usefulness for reducing computational complexity and improving the BER of the communication system.
In a wireless propagation channel, the channel for each user is independent. By using this characteristic, in orthogonal frequency division multiple access (OFDMA) systems, the multiuser diversity (MUDiv) gain is obtained. However, at high Doppler frequency, since channel state information (CSI) is changed between the head and latter of packets due to the fast fading channel, the system performance is degraded. Moreover, in this case, when the feedback delay is occurred, CSI is more changed and the system performance is more degraded. To solve these problems, in this paper, we propose the compensation of the deteriorated CSI allocation due to the fast fading channel and the feedback delay based on the turbo decision direct for MUDiv/OFDMA systems.
Electrons possess both wave and particle natures. In this paper, we represent an electron on multi-stage coupled electron waveguides both by a wave function and by a probabilistic particle. We show first that time evolution of the wave function can be obtained by the combined use of the tight-binding method and two-port circuit theory. The wave function is in variable-separable form when wave propagation on the coupled electron waveguides is single mode. We present secondly that the variable separation simplifies the general Langevin equation describing behavior of the probabilistic particle. According to these two schemes, wave functions and sample particle trajectories on single and two-stage coupled electron waveguides were exemplified.
A pattern recognition system is trained by using a training data set composed of input data and corresponding desired output data. After the training, the performance of the system is evaluated from certain perspectives. One is the misclassification rate (MCR) for a test data set, which is a data set separated from the training data set used in the training. The strength against noise, i.e., the noise robustness, is also an important performance measure. The noise robustness of a system is estimated by testing the MCR for a data set in which the inputs are corrupted by artificial noise. However, this test procedure can be computationally expensive, because a large number of corrupt inputs have to be created in order to cover the variability of the noise and the classification procedure has to be run for all of them. In this paper, based on a perturbative approximation method, we derive an effective test method for the noise robustness of pattern recognition systems based on deep neural networks. We demonstrate the validity of our method through numerical experiments using the MNIST data set and show that our method is much faster than the conventional expensive test method.
Optical reservoir computing (RC) with delayed feedback is expected to achieve high-speed data processing. However, in the parallel RC framework, the digital pre-process limits the actual processing speed. An analog-based, simple pre-processing method was developed and implemented in an optical RC architecture to overcome the bottleneck, and the performance in calculating two types of parallel task was evaluated. One was two independent benchmark tasks; the other was an integrative multi-input and multi-output odor identification task. We successfully demonstrated that these two parallel tasks can be physically processed with the network parameters optimized to maximize the RC performance. These results strongly suggest the potential of the optical reservoir system for future high-speed multimodal data processing applications.
Electroencephalogram (EEG) related to music preferences can be recorded when a participant has to evaluate preference score when listening to music excerpts. Whether EEG is related to decision-making regarding music preference remains unknown. In the present study, we separated listening-to-music and evaluating-preference (thinking) periods. Participants evaluated preferences using six-point Likert scales. F3γ power during the thinking period was significantly correlated with preference. The mean discrimination rate was around 70% using F3γ power. The results suggest that F3γ power may be related to deciding music preference.
This paper proposes a framework for the numerical design of continuous-time dynamical systems that bind desirably configured unstable periodic orbits (UPOs) into a chaotic attractor. The proposed numerical framework is comprised of the following four steps:(a) construction of a chaos-generating template structure consisting of a set of trajectories including explicitly embedded UPOs,(b) topology-preserving deformation of the template structure according to the desired configuration of UPOs,(c) assignment of attracting properties to the deformed template structure, and(d) function approximation that yields a vector-field function of a dynamical system that generates a chaotic attractor with desired UPOs. This paper elaborates on each step of the framework and presents a reference implementation, supported with numerical examples. From the viewpoint of numerical vector-field design, our proposal intends to extend the functionality of (stable) limit-cycle generators by introducing transitivity among configured UPOs that is intrinsic to chaotic systems.
We describe a method for tracking bifurcation curves from only time-series datasets. We apply a tracking algorithm to unknown systems based on the reconstruction of bifurcation diagrams that can estimate the parameter space and oscillatory patterns when parameters change. Therefore, this method can track the bifurcation curves of unknown systems from only measured time-series datasets, whereas the target systems in previous studies are known. By tracking bifurcation curves, we can obtain bifurcation points with increased accuracy as compared with bifurcation diagrams plotted by brute-force methods. In this paper, we present the results of numerical experiments in which bifurcation curves of a Hénon map as an unknown system are tracked from only several time-series datasets.