Event-driven Algorithms Research Articles

Scalable solutions for implementing satellite-based sensor networks

The importance of sensor networks in the field of IoT cannot be overstated, especially with the growing demand for IoT solutions. The new Internet protocols such as IPv6 and 5G mobile provide high-capacity communication networks, demonstrating the relevance and potential applications of IoT in the future. However, with new demands come new challenges. This paper aims to explore the challenges of Satellite-based Sensor Networks, a new topic with various applications. This article explores the challenges and constraints of implementing a scalable satellite-based sensor network. We examined the hardware and software limitations of the Iridium satellite SBD protocol and proposed tailored solutions using design flow algorithms to tackle flexibility, power consumption, bandwidth, and scalability issues. Satellite-based sensor networks face challenges in data transmission and sensor lifespan. In this study, we introduce event-driven and compression algorithms to address these challenges. By implementing clustering, we achieved a significant increase in the number of nodes within each cluster. Our proposed methods were validated using simulated datasets derived from temperature and location data. A unique data transmission method, based on event count, was introduced. Notably, the Differential Huffman compression algorithm resulted in an 81 % reduction in data transmission compared to traditional methods. This demonstrates about 62 % improvement in efficiency over existing techniques. Our approach not only paves the way for future advancements in this domain but also highlights the untapped potential of satellite-based sensor networks.

Fully Distributed Event-Driven Adaptive Consensus of Unknown Linear Systems.

This article considers the consensus problem of unknown linear multiagent systems (MASs) through adaptive event-driven control in leader-follower and leaderless networks. The proposed event-driven algorithms do not involve any global information related to the network communication structure and rely only on local information exchange to achieve consensus on MASs and are therefore fully distributed. Furthermore, the constraint of continuous communication among the agents is eliminated in terms of control law updates and triggering state monitoring. Another desirable aspect of this article is that the design process of the control algorithms is independent of the parameters of each agent's dynamics and thus does not require precise information about the dynamics of MASs. We further exclude the Zeno behavior of each agent by proving the existence of a strict positive lower bound between any two adjacent events. Finally, the effectiveness of the proposed adaptive event-driven algorithms is verified by a simulation example.

Translational and rotational non-Gaussianities in homogeneous freely evolving granular gases.

The importance of roughness in the modeling of granular gases has been increasingly considered in recent years. In this paper, a freely evolving homogeneous granular gas of inelastic and rough hard disks or spheres is studied under the assumptions of the Boltzmann kinetic equation. The homogeneous cooling state is studied from a theoretical point of view using a Sonine approximation, in contrast to a previous Maxwellian approach. A general theoretical description is done in terms of d_{t} translational and d_{r} rotational degrees of freedom, which accounts for the cases of spheres (d_{t}=d_{r}=3) and disks (d_{t}=2,d_{r}=1) within a unified framework. The non-Gaussianities of the velocity distribution function of this state are determined by means of the first nontrivial cumulants and by the derivation of non-Maxwellian high-velocity tails. The results are validated by computer simulations using direct simulation Monte Carlo and event-driven molecular dynamics algorithms.

ES-ImageNet: A Million Event-Stream Classification Dataset for Spiking Neural Networks.

With event-driven algorithms, especially spiking neural networks (SNNs), achieving continuous improvement in neuromorphic vision processing, a more challenging event-stream dataset is urgently needed. However, it is well-known that creating an ES-dataset is a time-consuming and costly task with neuromorphic cameras like dynamic vision sensors (DVS). In this work, we propose a fast and effective algorithm termed Omnidirectional Discrete Gradient (ODG) to convert the popular computer vision dataset ILSVRC2012 into its event-stream (ES) version, generating about 1,300,000 frame-based images into ES-samples in 1,000 categories. In this way, we propose an ES-dataset called ES-ImageNet, which is dozens of times larger than other neuromorphic classification datasets at present and completely generated by the software. The ODG algorithm implements image motion to generate local value changes with discrete gradient information in different directions, providing a low-cost and high-speed method for converting frame-based images into event streams, along with Edge-Integral to reconstruct the high-quality images from event streams. Furthermore, we analyze the statistics of ES-ImageNet in multiple ways, and a performance benchmark of the dataset is also provided using both famous deep neural network algorithms and spiking neural network algorithms. We believe that this work shall provide a new large-scale benchmark dataset for SNNs and neuromorphic vision.

High-parallelism Inception-like Spiking Neural Networks for Unsupervised Feature Learning

Spiking Neural Networks (SNNs) are brain-inspired, event-driven machine learning algorithms that have been widely recognized in producing ultra-high-energy-efficient hardware. Among existing SNNs, unsupervised SNNs based on synaptic plasticity, especially Spike-Timing-Dependent Plasticity (STDP), are considered to have great potential in imitating the learning process of the biological brain. Nevertheless, the existing STDP-based SNNs have limitations in constrained learning capability and/or slow learning speed. Most STDP-based SNNs adopted a slow-learning Fully-Connected (FC) architectures and used a sub-optimal vote-based scheme for spike decoding. In this paper, we overcome these limitations with: 1) a design of high-parallelism network architecture, inspired by the Inception module in Artificial Neural Networks (ANNs); 2) use of a Vote-for-All (VFA) decoding layer as a replacement to the standard vote-based spike decoding scheme, to reduce the information loss in spike decoding and, 3) a proposed adaptive repolarization (resetting) mechanism that accelerates SNNs’ learning by enhancing spiking activities. Our experimental results on two established benchmark datasets (MNIST/EMNIST) show that our network architecture resulted in superior performance compared to the widely used FC architecture and a more advanced Locally-Connected (LC) architecture, and that our SNN achieved competitive results with state-of-the-art unsupervised SNNs (95.64%/80.11% accuracy on the MNIST/EMNISE dataset) while having superior learning efficiency and robustness against hardware damage. Our SNN achieved great classification accuracy with only hundreds of training iterations, and random destruction of large numbers of synapses or neurons only led to negligible performance degradation.

Sepia, Tarsier, and Chameleon: A Modular C++ Framework for Event-Based Computer Vision.

This paper introduces an new open-source, header-only and modular C++ framework to facilitate the implementation of event-driven algorithms. The framework relies on three independent components: sepia (file IO), tarsier (algorithms), and chameleon (display). Our benchmarks show that algorithms implemented with tarsier are faster and have a lower latency than identical implementations in other state-of-the-art frameworks, thanks to static polymorphism (compile-time pipeline assembly). The observer pattern used throughout the framework encourages implementations that better reflect the event-driven nature of the algorithms and the way they process events, easing future translation to neuromorphic hardware. The framework integrates drivers to communicate with the DVS, the DAVIS, the Opal Kelly ATIS, and the CCam ATIS.

Spiking Neural Networks Hardware Implementations and Challenges

Neuromorphic computing is henceforth a major research field for both academic and industrial actors. As opposed to Von Neumann machines, brain-inspired processors aim at bringing closer the memory and the computational elements to efficiently evaluate machine learning algorithms. Recently, spiking neural networks, a generation of cognitive algorithms employing computational primitives mimicking neuron and synapse operational principles, have become an important part of deep learning. They are expected to improve the computational performance and efficiency of neural networks, but they are best suited for hardware able to support their temporal dynamics. In this survey, we present the state of the art of hardware implementations of spiking neural networks and the current trends in algorithm elaboration from model selection to training mechanisms. The scope of existing solutions is extensive; we thus present the general framework and study on a case-by-case basis the relevant particularities. We describe the strategies employed to leverage the characteristics of these event-driven algorithms at the hardware level and discuss their related advantages and challenges.

Poker-DVS and MNIST-DVS. Their History, How They Were Made, and Other Details.

This article reports on two databases for event-driven object recognition using a Dynamic Vision Sensor (DVS). The first, which we call Poker-DVS and is being released together with this article, was obtained by browsing specially made poker card decks in front of a DVS camera for 2–4 s. Each card appeared on the screen for about 20–30 ms. The poker pips were tracked and isolated off-line to constitute the 131-recording Poker-DVS database. The second database, which we call MNIST-DVS and which was released in December 2013, consists of a set of 30,000 DVS camera recordings obtained by displaying 10,000 moving symbols from the standard MNIST 70,000-picture database on an LCD monitor for about 2–3 s each. Each of the 10,000 symbols was displayed at three different scales, so that event-driven object recognition algorithms could easily be tested for different object sizes. This article tells the story behind both databases, covering, among other aspects, details of how they work and the reasons for their creation. We provide not only the databases with corresponding scripts, but also the scripts and data used to generate the figures shown in this article (as Supplementary Material).

Event‐Driven Top‐ k Queries in Sensor Networks with Multi Microenvironments

In the emerging area of sensor-based systems, a significant challenge is to develop reliable, energy efficient methods to extract useful information from the distributed sensory data. Existing top-k query algorithms of sensor networks are only applicable to the sensor networks with single Microenvironment (ME). Besides, these approaches rely on using raw sensor readings, which handling these raw sensory data requires large amount of data transmission and is memory-consuming. In this paper we highlight the multiple MEs in sensor networks and propose a novel event-driven approximate top-k query framework based on fuzzy method. Firstly, membership function of fuzzy method is introduced to describe the global potential event confidence. Subsequently, non-uniform membership degree subranges based linguistic variables instead of numeric sensor readings are used for in-network top-k event fusion. Also two distributed event-driven approximate top-k query algorithms are devised. Theoretical analysis as well as extensive simulations based on synthetic and real data sets show that our framework reduces the data transmission significantly and the query results are more interpretable with quality guarantees.

Chaotic dynamics of one-dimensional systems with periodic boundary conditions

We provide appropriate tools for the analysis of dynamics and chaos for one-dimensional systems with periodic boundary conditions. Our approach allows for the investigation of the dependence of the largest Lyapunov exponent on various initial conditions of the system. The method employs an effective approach for defining the phase-space distance appropriate for systems with periodic boundaries and allows for an unambiguous test-orbit rescaling in the phase space required to calculate the Lyapunov exponents. We elucidate our technique by applying it to investigate the chaotic dynamics of a one-dimensional plasma with periodic boundaries. Exact analytic expressions are derived for the electric field and potential using Ewald sums, thereby making it possible to follow the time evolution of the plasma in simulations without any special treatment of the boundary. By employing a set of event-driven algorithms, we calculate the largest Lyapunov exponent, the radial distribution function, and the pressure by following the evolution of the system in phase space without resorting to numerical manipulation of the equations of motion. Simulation results are presented and analyzed for the one-dimensional plasma with a view to examining the dynamical and chaotic behavior exhibited by small and large versions of the system.

Simulation of Cell Signaling Communications Using Event-Driven Algorithms

In this paper we present two implementations of event-driven algorithms for simulating molecular dynamics using the Omnet++ Simulation Framework and its Future Event Set (FES) implementation. The first one uses a cell-linked list algorithm. The second one extends the cell-linked list algorithm incorporating a Verlet neighbor list algorithm. We also present results and compare both algorithms over a set of different scenarios. Finally, we discuss the advantages of using the Omnet++ Simulation Framework and the implemented algorithm for simulating cell-signaling communications.

Efficient Sampling in Event-Driven Algorithms for Reaction-Diffusion Processes

AbstractIn event-driven algorithms for simulation of diffusing, colliding, and reacting particles, new positions and events are sampled from the cumulative distribution function (CDF) of a probability distribution. The distribution is sampled frequently and it is important for the efficiency of the algorithm that the sampling is fast. The CDF is known analytically or computed numerically. Analytical formulas are sometimes rather complicated making them difficult to evaluate. The CDF may be stored in a table for interpolation or computed directly when it is needed. Different alternatives are compared for chemically reacting molecules moving by Brownian diffusion in two and three dimensions. The best strategy depends on the dimension of the problem, the length of the time interval, the density of the particles, and the number of different reactions.

Sinusoidal shaking in event-driven simulations

Event-driven algorithms are a powerful and efficient simulation method that can be used to numerically study the behavior of particulate systems, such as granular materials or powders. Its superior efficiency originates from only considering analytically calculable points in time, where physically relevant events occur. This, however, is only possible for pairs of trajectories whose relative distance can be expressed as a polynomial of order less than or equal to four. For more complex trajectories, like a sinusoidal motion of a wall, the collision times have to be calculated numerically, which dramatically reduces the efficiency. In this paper we present a tabulation method implemented in an event-driven algorithm, which is able to efficiently handle the one-dimensional motion of a periodically moving wall. By demonstrating this for a sinusoidally moving wall, we enhance the comparability of simulations with (usually sinusoidally driven) experiments. Because granular materials or powders are systems which are easily accessible through both experiments and simulations, our work enhances the growing field of statistical physics far from equilibrium.

Economic dispatch of virtual power plants in an event-driven service-oriented framework using standards-based communications

The accelerated evolution of power systems and the associated new trends, such as Smart Grids, has stimulated the development of various approaches in handling the problems of increased electricity demands, power market (de)regulation and power systems reliability. One of the new key concepts is the virtual power plant (VPP), which represents a controllable portfolio of distributed energy resources (DERs). Such a portfolio can be operated in various modes, each with a set of unique control requirements. An open framework providing robust solution for large scale DERs integration and control is one of the key issues in Smart Grid development. This paper proposes an approach for solving this problem by utilizing standards-based power system communications, application modeling based on event-driven information services and algorithms for optimized VPP control. The applicability of the proposed technical solution is demonstrated and analyzed via simulation of the developed economic dispatch algorithm.

Embedding a process plan in function blocks for adaptive machining

This paper presents a function block enabled approach towards adaptive process planning and machining. A two-layer structure of supervisory planning and operation planning is proposed to separate generic data from machine-specific ones. The supervisory planning is only performed once, in advance, at the shop level to generate machine-neutral process plans, whereas the operation planning is carried out at runtime at the machine level to determine machine-specific operations. Such adaptive decision making is facilitated by event-driven algorithms embedded in the function blocks. It is expected that the new approach can greatly enhance the dynamism of fluctuating job-shop machining operations.

Asynchronous Event-Driven Particle Algorithms

We present, in a unifying way, the main components of three asynchronous event-driven algorithms for simulating physical systems of interacting particles. The first example, hard-particle molecular dynamics, is well known. We also present a recently developed diffusion kinetic Monte Carlo algorithm, as well as a novel stochastic molecular-dynamics algorithm that builds on the Direct Simulation Monte Carlo. We explain how to effectively combine asynchronous event-driven with classical time-driven or with synchronous event-driven handling. Finally, we discuss some promises and challenges for event-driven simulation of realistic physical systems.

A Generalized Linear Integrate-and-Fire Neural Model Produces Diverse Spiking Behaviors

For simulations of neural networks, there is a trade-off between the size of the network that can be simulated and the complexity of the model used for individual neurons. In this study, we describe a generalization of the leaky integrate-and-fire model that produces a wide variety of spiking behaviors while still being analytically solvable between firings. For different parameter values, the model produces spiking or bursting, tonic, phasic or adapting responses, depolarizing or hyperpolarizing after potentials and so forth. The model consists of a diagonalizable set of linear differential equations describing the time evolution of membrane potential, a variable threshold, and an arbitrary number of firing-induced currents. Each of these variables is modified by an update rule when the potential reaches threshold. The variables used are intuitive and have biological significance. The model's rich behavior does not come from the differential equations, which are linear, but rather from complex update rules. This single-neuron model can be implemented using algorithms similar to the standard integrate-and-fire model. It is a natural match with event-driven algorithms for which the firing times are obtained as a solution of a polynomial equation.

Time-driven simulation with fully asynchronous pulse coupling

Event Abstract Back to Event Time-driven simulation with fully asynchronous pulse coupling Alexander Hanuschkin1*, Suzanne Kunkel1, Moritz Helias1, Abigail Morrison2 and Markus Diesmann1, 2 1 BCCN Freiburg, Germany 2 RIKEN Brain Science Institute, Japan Artificial synchrony can be introduced by discrete time simulation of neuronal networks, since they typically constrain spike times to a grid determined by the computational step size [1]. Event-driven algorithms avoid this problem but are computationally demanding, both with respect to calculating future spike times and to event management, particularly for large network sizes. To address this problem, Morrison et al. [2] developed a general method of handling off-grid spiking in combination with exact subthreshold integration in globally time-driven simulations [3,4] using interpolation to approximate threshold crossings. In the framework of event-driven simulations, Brette [5] presented an elegant method of calculating spike times of integrate-and-fire neurons with exponentially decaying currents. The prediction of the next threshold crossing of the membrane potential is reformulated into a root finding problem of the equivalent polynomial. Here, we show that this scheme can also be implemented in the time-driven environment of NEST [4]. Additionally, we extended the model of Morrison et al. [2] by replacing the interpolation of threshold crossings with the computationally more expensive, but numerically more exact, Newton-Raphson technique. We compare the accuracy of the three approaches in single-neuron simulations and the efficiency in a balanced random network of 12,000 neurons [6]. For small input and output rates the polynomial method is more efficient than the interpolated one whereas for high input or output rates the interpolated method is more efficient than the polynomial one. The direct numerical method based on Newton-Raphson root finding outperforms both the interpolation and the polynomial approach at all input/output rates. Partially funded by DIP F1.2, BMBF Grant 01GQ0420 to the BCCN Freiburg, and EU Grant 15879 (FACETS).

Transport coefficients for dense hard-disk systems

A study of the transport coefficients of a system of elastic hard disks based on the use of Helfand-Einstein expressions is reported. The self-diffusion, the viscosity, and the heat conductivity are examined with averaging techniques especially appropriate for event-driven molecular dynamics algorithms with periodic boundary conditions. The density and size dependence of the results are analyzed, and comparison with the predictions from Enskog's theory is carried out. In particular, the behavior of the transport coefficients in the vicinity of the fluid-solid transition is investigated and a striking power law divergence of the viscosity with density is obtained in this region, while all other examined transport coefficients show a drop in that density range in relation to the Enskog's prediction. Finally, the deviations are related to shear band instabilities and the concept of dilatancy.

Neighbor list collision-driven molecular dynamics simulation for nonspherical hard particles.II. Applications to ellipses and ellipsoids

We apply the algorithm presented in the first part of this series of papers to systems of hard ellipses and ellipsoids. The theoretical machinery needed to treat such particles, including the overlap potentials, is developed in full detail. We describe an algorithm for predicting the time of collision for two moving ellipses or ellipsoids. We present performance results for our implementation of the algorithm, demonstrating that for dense systems of very aspherical ellipsoids the novel techniques of using neighbor lists and bounding sphere complexes, offer as much as two orders of magnitude improvement in efficiency over direct adaptations of traditional event-driven molecular dynamics algorithms. The practical utility of the algorithm is demonstrated by presenting several interesting physical applications, including the generation of jammed packings inside spherical containers, the study of contact force chains in jammed packings, and melting the densest-known equilibrium crystals of prolate spheroids.