This paper studies digital return maps defined on a set of points. Depending on parameters and initial states, the maps can generate a variety of periodic orbits. In order to analyze the periodic orbits, we present two feature quantities. The first and second quantities can characterize the plentifulness and a variety of periodic orbits, respectively. Constructing a feature plane of the two quantities, the dynamics of the map can be visualized. Using the feature quantities and feature plane, we investigate various periodic orbits in digital return maps derived from a basic class of cellular automata.
This paper studies super-stable periodic orbits and related phenomena in a simple switched dynamical system. The system repeats integrate-and-fire behavior between a constant threshold and a periodic base signal. The base signal consists of two components: the fundamental and higher frequency triangular signals. First, we demonstrate “chaos + chaos = order” such that the system exhibits chaos if the base signal is either of the two components whereas the system exhibits a super-stable periodic orbit if the base signal consists of both components. This phenomenon is confirmed experimentally and is explained theoretically. Second, we extract two key parameters and show that the system can exhibit a variety of super-stable periodic orbits. Applying a mapping procedure, existence regions of basic super-stable periodic orbits are calculated precisely.
Artificial neural networks with stochastic state transitions and calculations, such as Boltzmann machines, have excelled over other machine learning approaches in various benchmark tasks. The networks often achieve better results than deterministic neural networks of similar sizes, but they require implementation of nonlinear continuous functions for probabilistic density functions, thus resulting in an increase in computational effort. The architecture size of cutting-edge artificial neural networks are ever-growing; therefore, they require dedicated hardware. Conversely, asynchronous cellular automaton-based neuron models have been investigated to model the highly nonlinear dynamics of biological neurons. They are special types of cellular automata and are implemented as small asynchronous sequential logic circuits. In this study, we propose a new type of asynchronous network of cellular automaton-based neuron for the efficient implementation of Boltzmann machines. Experimental comparisons demonstrate that the proposed approach achieves comparable or better performances in such benchmark tasks as image classification and generation while it requiring much less computational resources than traditional implementation approaches.
Inspired by the augmented Lorenz equations, we have designed a star network of Rössler oscillators, referred to as augmented Rössler equations, in which each Rössler oscillator is coupled with the other oscillators via a single variable y as the central node of the whole network. We investigate the dynamical nature of the augmented Rössler equations in terms of the bifurcation diagram of a single augmented Rössler oscillator and the chaotic synchronizability of coupled augmented Rössler oscillators, and show that intermittent synchronization between identical augmented Rössler oscillators as well as partial synchronization between nonidentical ones can be achieved via direct coupling of the central nodes. We also show that nonidentical augmented Rössler oscillators coupled via intermittent mutual diffusive coupling exhibit partial synchrony, despite the intermittency of the diffusive coupling. We discuss possible application of such synchronous behavior in terms of chaos-based secret key distribution.
In recent years, many researchers have become interested in methods for mitigating traffic congestion by optimizing traffic signal parameters. To mitigate traffic congestion over a widespread area, a method using an advanced genetic algorithm and a traffic simulator has been proposed (Nishihara, T., et al., “The Verification with Real-World Road Network on Optimization of Traffic Signal Parameters using Multi-Element Genetic Algorithms”, ITS World Congress, 2012). However, this method consumes considerable time when simulating traffic flow. This paper proposes a method that reduces the processing time of the simulator by using a neural network.
Information communication networks are rapidly growing recently. Because of increase of communication overhead, traditional routing protocols using global information of the network, such as topology of the whole network or traffic demands between most of pairs of routers, are facing difficulty in reliable routing. To alleviate this, distributed routing protocols relying only on local observables of the network attract much attention recently. The non-requirement of global knowledge of the whole network largely reduces communication overhead of these protocols. However, the lack of knowledge can also be a significant drawback of them because they cannot promptly respond to traffic changes that occur on out of their local scopes. It means that network resources cannot be utilized sufficiently. To solve the problem, here, by extending an existing distributed routing protocol called ARAS, we propose a novel routing protocol that utilizes multiple paths in parallel. The protocol adaptively modulates packet allocation ratio to paths based only on local observables of the network. Multipath routing, however, easily give flapping of packet allocation due to competition among multiple routers. We study competition between routers and provide ways to suppress the flapping. We show validity of the proposal using a network simulation where prompt response of packet reallocation is required.
Recently, various mobile communication systems have been widely deployed, and mobile traffic is increasing. However, the bandwidth available for mobile communications is limited, hence the scarcity of radio resources in mobile communications is a serious problem. As an approach to solve this problem, cognitive wireless communication models have been proposed. These model search for vacant time slots in multi-channel wireless communication systems. Although previous studies have shown that frequency utilization efficiency can be improved by multi-armed bandit algorithms, channels are assumed to be independent. However, channels used in 2.4 GHz wireless LANs (such as IEEE802.11b or IEEE802.11g) are not independent because these channels overlap with adjacent channels. In this paper, we propose an extended multi-armed bandit algorithm that uses continuous-valued rewards, which is applicable to wireless communication systems with overlapping channels. We show the effectiveness of the proposed method by experimental demonstrations.
MIMO has been recognized as a promising way to provide high speed data transmission. However, since the number of pilot symbols is proportionally depending on the number of transmit antenna, the total transmission rate of MIMO would be degraded. To solve this problem, the channel estimation method using virtual pilot signal(VPS) with a few pilot signals has been proposed. However, since the conventional method identifies the channel state information(CSI) iteratively, the complexity is a seriously considerable work. Previously, we have proposed the time-frequency interferometry (TFI)-OFDM to achieve a superior BER performance with a small number of pilot symbols without any iterative channel estimation. However, the TFI method is still needed a large number of pilot signals. To mitigate the problem of the TFI method, in this paper, we propose a novel channel compensation method based on VPS and TFI for MIMO without increasing the system complexity. The proposed method has a low complexity with improving BER performance to compare with the conventional method. From the simulation results, the proposed method can achieve the improved BER performance of 1.5dB gain compared with the conventional method for 4×4 and 8×8 MIMO systems.
Combinatorial optimization problems consist of static problems such as the traveling salesman problem and the quadratic assignment problem, and dynamic problems such as the packet routing problem and the traffic flow control problem. In static combinatorial optimization problems, the search space for the solution does not change over time and, therefore, neither does the optimal solution. On the contrary, in dynamic combinatorial optimization problems, the search space always changes. Thus, there is no guarantee that the optimal solution for one iteration also applies to the next iteration. In the context of dynamic combinatorial optimization problems, we propose in this paper a heuristic routing method that uses chaotic neurodynamics and degree information to solve the packet routing problem. Numerical experiments showed that the proposed method improves average arrival rate by approximately 130% over the conventional shortest hop method.
Inductive Search for solving global optimization problems has attracted much attention because it showed the best performance at the First International Contest on Evolutionary Optimisation (1st ICEO). However, the details of this method are not clear. We therefore investigated the details of the method by analyzing codes, and we point out problems and notes of the method. In order to overcome the problems, we propose a modified inductive search by using a deterministic one-dimensional global search. Finally, we evaluate the performance among the original method, our implemented method and that of the proposed method.
This paper investigates the implementation of compressive sensing (CS) for stepped frequency continuous wave ground penetrating radar (SFCW GPR) imaging system. Previous works in this field mostly focus on reducing the frequency samples in the measurement. While this approach enables faster scanning speed, we consider reducing spatial sampling is more efficient in reducing the data acquisition time in the GPR survey over a very large area. In this study we propose a data acquisition method and CS algorithm. A two-step sampling scheme is presented. In the first step, the spatial sampling was directly conducted in the data acquisition process. In the second step, the frequency sampling was conducted offline during the signal processing. Full frequency information was used in pre-processing to suppress the noise and clutter in the experiment data. To solve the sparse-data problem, some CS algorithms are compared and a modified Bayesian approach based on fast relevance vector machine (RVM) is proposed. The performance of the proposed CS-GPR system is analyzed using the real experimental GPR data set which contains non ideal conditions, e.g. high level clutter, not truly sparse targets, and inaccurate estimation of wave velocity in the medium. Using the proposed data acquisition and CS algorithm, even with these non ideal conditions, CS can give clear and stable results with high probability detection of the target.
A Redox Flow Battery (RFB) is one of the promising energy storage systems in power grid. An RFB has many advantages such as a quick response, a large capacity, and a scalability. Due to these advantages, an RFB can operate in mixed time scales. Actually, it has been demonstrated that an RFB can be used for load leveling, compensating sag, and smoothing the output of the renewable sources. An analysis on transient behaviors of an RFB is a key issue for these applications. An RFB is governed by electrical, chemical, and fluid dynamics. The hybrid structure makes the analysis difficult. To analyze transient behaviors of an RFB, the exact model is necessary. In this paper, we focus on a change in a concentration of ions in the electrolyte, and simulate the change with a model which is mainly based on chemical kinetics. The simulation results introduces transient behaviors of an RFB in a response to a load variation. There are found three kinds of typical transient behaviors including oscillations. As results, it is clarified that the complex transient behaviors, due to slow and fast dynamics in the system, arise by the quick response to load.