Discrete Communication Research Articles

Communication Learning via Backpropagation in Discrete Channels with Unknown Noise

This work focuses on multi-agent reinforcement learning (RL) with inter-agent communication, in which communication is differentiable and optimized through backpropagation. Such differentiable approaches tend to converge more quickly to higher-quality policies compared to techniques that treat communication as actions in a traditional RL framework. However, modern communication networks (e.g., Wi-Fi or Bluetooth) rely on discrete communication channels, for which existing differentiable approaches that consider real-valued messages cannot be directly applied, or require biased gradient estimators. Some works have overcome this problem by treating the message space as an extension of the action space, and use standard RL to optimize message selection, but these methods tend to converge slower and to inferior policies. In this paper, we propose a stochastic message encoding/decoding procedure that makes a discrete communication channel mathematically equivalent to an analog channel with additive noise, through which gradients can be backpropagated. Additionally, we introduce an encryption step for use in noisy channels that forces channel noise to be message-independent, allowing us to compute unbiased derivative estimates even in the presence of unknown channel noise. To the best of our knowledge, this work presents the first differentiable communication learning approach that can compute unbiased derivatives through channels with unknown noise. We demonstrate the effectiveness of our approach in two example multi-robot tasks: a path finding and a collaborative search problem. There, we show that our approach achieves learning speed and performance similar to differentiable communication learning with real-valued messages (i.e., unlimited communication bandwidth), while naturally handling more realistic real-world communication constraints. Content Areas: Multi-Agent Communication, Reinforcement Learning.

Simultaneous Policy and Discrete Communication Learning for Multi-Agent Cooperation

Decentralized multi-agent reinforcement learning has been demonstrated to be an effective solution to large multi-agent control problems. However, agents typically can only make decisions based on local information, resulting in suboptimal performance in partially-observable settings. The addition of a communication channel overcomes this limitation by allowing agents to exchange information. Existing approaches, however, have required agent output size to scale exponentially with the number of message bits, and have been slow to converge to satisfactory policies due to the added difficulty of learning message selection. We propose an independent bitwise message policy parameterization that allows agent output size to scale linearly with information content. Additionally, we leverage aspects of the environment structure to derive a novel policy gradient estimator that is both unbiased and has a lower variance message gradient contribution than typical policy gradient estimators. We evaluate the impact of these two contributions on a collaborative multi-agent robot navigation problem, in which information must be exchanged among agents. We find that both significantly improve sample efficiency and result in improved final policies, and demonstrate the applicability of these techniques by deploying the learned policies on physical robots.

Impulsive-Interaction-Driven Synchronization in an Array of Coupled Neural Networks

This paper deals with the problem of globally exponential synchronization and bipartite synchronization of coupled neural networks with impulsive interactions. Impulsive interaction means that a number of neural networks only communicate with each other at impulsive instants, while the array of neural networks are independent from each other at the remaining time. The advantage of the scheme is that communication cost can be largely reduced when only discrete communication is required. Moreover the communication links between nodes can be either positive or negative at impulsive instants. Using the Lyapunov method combined with some mathematical analysis and average impulsive interval, some efficient criteria are obtained to guarantee synchronization of impulsive coupled neural networks. Finally, the validity of our theoretical results is demonstrated by two numerical examples.

Efficacy of Group Conversation Intervention in Individuals with Severe Profiles of Aphasia.

An estimated 390,000 to 520,000 individuals with severe aphasia (IWSA) currently live in the United States. IWSA experience profound social isolation, which is associated with a wide range of negative health outcomes, including mortality. Treatments for severe aphasia frequently focus on compensatory communication approaches or a discrete communication act rather than on participation-based treatment. The purpose of this study was to determine whether IWSA demonstrated improved performance on standardized language measures, patient-reported outcome measures, and connected speech samples as a result of client-centered conversation group treatment. Results of assessments conducted at pretreatment, posttreatment, and maintenance intervals were variable across participants. All participants demonstrated improvement in at least one of the outcome measures considered. Importantly, none of these measures fully captured how IWSA were able to convey their thoughts in supported conversation. The results lend support for the use of conversation treatment for, and for further study in, this subpopulation of individuals with aphasia.

Improved explicit co-simulation methods incorporating relaxation techniques

Explicit co-simulation is an increasingly popular technique for solver coupling of multidisciplinary systems in a distributed environment. Compared to implicit approaches, explicit coupling schemes are simpler to implement, since explicit methods allow the synchronization of different subdomains at user-defined discrete communication points without a reinitialization of the subsystem solvers. However, a typical disadvantage of explicit schemes is the reduced stability behavior, which is introduced by errors resulting from the approximation of the coupling variables. In the manuscript at hand, improved explicit co-simulation methods incorporating relaxation techniques are considered for the case that the coupling of the subsystems is realized by constitutive laws. We discuss three decomposition techniques, namely force/force coupling, force/displacement coupling and displacement/displacement coupling. Two novel explicit coupling approaches are analyzed. Within the first approach, classical co-simulation methods based on Lagrange polynomials for extrapolating the coupling variables are modified by incorporating a straightforward relaxation approach using only the current and previous coupling variables. Applying the second approach, predicted coupling variables are generated by using a constant or linear acceleration approach for the coupling bodies. With the predicted, current and previous coupling variables, special approximation polynomials are generated also using a relaxation approach. The numerical stability and convergence characteristics will be analyzed for two explicit co-simulation approaches. It will be shown that stability and convergence can be notably improved by a proper choice of the relaxation parameters. The stabilized co-simulation techniques will be verified using several numerical examples.

An Event-Triggered Secondary Control Strategy With Network Delay in Islanded Microgrids

Frequency deviation and powers sharing error exist in a microgrid with multiple droop-controlled distributed generations. To solve the above problems, we propose an event-triggered multistage secondary control. First, to improve the synchronization accuracy, we divide secondary control into multiple stages based on frequency/voltage droop characteristics. The inputs of proportion-integration compensator for these stages are different. The corresponding stage will be activated when the value of frequency/voltage meets a certain “event” (given ranges). Then, we consider the network delay, use a discrete communication scheme to reduce the bandwidth utilization, and further avoid the oversized delay. To stabilize the interloop control with input delay, we derive the robustness criteria. Finally, simulations in RT-LAB confirm the effectiveness of the proposed control strategy.

Distributed Event-triggered Consensus Control for Heterogeneous Multi-agent Systems under Fixed and Switching Topologies

This paper investigates the distributed consensus problem for heterogeneous multi-agent systems via event-triggered scheme. To alleviate the communication burden, a distributed event-triggered consensus algorithm using an open-loop estimate and neighboring information is proposed. In the proposed event-triggered scheme, the control input of each agent is dependent on the estimate and information at discrete instants. Firstly, the novel event-triggered consensus algorithm is proposed for fixed topology, which involves the discrete communication between neighboring agents. Then, for the switching topology, the triggering condition is proposed by virtue of a topology-dependent dwell time approach. Further, the corresponding consensus procedures are given. Finally, the effectiveness of the proposed scheme is illustrated by numerical results.

Robust distributed synchronization of networked linear systems with intermittent information

The problem of synchronization of multiple linear time-invariant systems connected over a network with asynchronous and intermittently available communication events is studied. To solve this problem, we propose a controller with hybrid dynamics, namely, the controller utilizes information transmitted to it during discrete communication events and exhibits continuous dynamics between such events. Due to the additional continuous and discrete dynamics inherent to the interconnected networked systems and communication structure, we use a hybrid systems framework to model and analyze the closed-loop system. The problem of synchronization is then recast as a set stabilization problem and, by employing Lyapunov stability tools for hybrid systems, sufficient conditions for asymptotic stability of the synchronization set are provided. Furthermore, we show that the property of synchronization is robust to perturbations. Numerical examples illustrating the main results are included.

Energy-leak monitoring and correction to enhance stability in the co-simulation of mechanical systems

Non-iterative co-simulation is an increasingly important technique for the simulation of complex mechanical systems. Adopting co-simulation schemes enables the simultaneous use of computational resources and makes it possible to select the most appropriate modelling techniques and algorithms to describe and solve the dynamics of each system component. However, it inherently requires the coupling of different subsystems at discrete communication times, which may compromise the stability of the overall integration process. One of the negative effects of discrete-time communication is the introduction of artificial energy in the system dynamics, which can render the simulation unstable if it accumulates over time. Excess energy can be dissipated introducing virtual damping elements in the subsystem models. The actual amount of damping must be adjusted as the simulation progresses to ensure that all the artificially generated energy is removed from the system while keeping the dynamics realistic. In this paper, we introduce a monitoring framework to keep track of this excess energy, and put forward a dissipation methodology to eliminate it. The ability of this framework to achieve stable non-iterative co-simulation was tested with several mechanical system examples.

It's not all about the Soprano: Rhinolophid bats use multiple acoustic components in echolocation pulses to discriminate between conspecifics and heterospecifics.

Acoustic communication plays a pivotal role in conspecific recognition in numerous animal taxa. Vocalizations must therefore have discrete acoustic signatures to facilitate intra-specific communication and to avoid misidentification. Here we investigate the potential role of echolocation in communication in horseshoe bats. Although it has been demonstrated that echolocation can be used to discriminate among con- and hetero-specifics, the specific acoustic cues used in discrimination are still relatively unknown. Furthermore, the Acoustic Communication Hypothesis proposes that in multispecies assemblages, in which echolocation frequencies are likely to overlap, bats partition acoustic space along several dimensions so that each species occupies a discrete communication domain. Thus, multiple echolocation variables may be used in discrimination. The objective of this study was to investigate the potential of various echolocation variables to function as discriminatory cues in echolocation-based species discrimination. Using habituation–dishabituation playback experiments, we firstly tested the ability of Rhinolophus clivosus to discriminate between echolocation pulses of heterospecifics with either discrete or overlapping frequencies. Secondly, to determine whether R. clivosus could use echolocation variables other than frequency, we investigated its ability to discriminate among echolocation pulses differing in only one manipulated parameter. These test variables were identified by their contribution to the dissimilarity among pulses. Our results suggest that R. clivosus could discriminate readily between species using echolocation pulses with discrete frequencies. When frequencies overlapped, the ability of bats to discriminate was dependant on additional acoustic variables that defined the acoustic space occupied by the test signal. These additional acoustic variables included, but may not be restricted to, sweep rate of the FM and duty cycle. Thus, when echolocation pulses share a similar acoustic domain, bats use several cues to reliably discriminate among heterospecifics.

Joint Source-Channel Secrecy Using Uncoded Schemes: Towards Secure Source Broadcast

This paper investigates a joint source-channel secrecy problem for the Shannon cipher broadcast system. We suppose list secrecy is applied, i.e., a wiretapper is allowed to produce a list of reconstruction sequences and the secrecy is measured by the minimum distortion over the entire list. For discrete communication cases, we propose a permutation-based uncoded scheme, which cascades a random permutation with a symbol-by-symbol mapping. Using this scheme, we derive an inner bound for the admissible region of secret key rate, list rate, wiretapper distortion, and distortions of legitimate users. For the converse part, we easily obtain an outer bound for the admissible region from an existing result. Comparing the outer bound with the inner bound shows that the proposed scheme is optimal under certain conditions. Besides, we extend the proposed scheme to the scalar and vector Gaussian communication scenarios, and characterize the corresponding performance as well. For these two cases, we also propose another uncoded scheme, orthogonal-transform-based scheme, which achieves the same performance as the permutation-based scheme. Interestingly, by introducing the random permutation or the random orthogonal transform into the traditional uncoded scheme, the proposed uncoded schemes, on one hand, provide a certain level of secrecy, and on the other hand, do not lose any performance in terms of the distortions for legitimate users.

Klaus Iohannis’ Presidential Term: Between High Hopes and Disillusionment

The 2014 presidential elections in Romania and the victory of the former mayor of Sibiu, Klaus Iohannis, nurtured many expectations and high hopes. His election marks a rupture with the old political elite and was expected to bring a more western and balanced approach in the head institution of the state. But after almost a year and a half from the start of his presidential term his performance looks rather ambiguous and inconsistent. The president’s selective use of presidential power and his discrete communication style have often put him in a hesitant and defensive position. In this context it is highly unlikely that Klaus Iohannis will pursue important changes and improvements during his term in the domain of minority politics, or extend ethnic minorities collective rights and provisions. Even though Klaus Iohannis belongs to Romania’s ethnic German minority, his presidential agenda makes few references to the empowerment of ethnic minorities. His ‘small steps’ approach might bring surprising results, but so far the first one year and a half of the presidential term has been marked by a rather unconvincing performance.

Seagull Syndrome: Relationships between Patrol Workers and Government Officials in NSW, Australia

This paper explores the nature of relations between public officials and community workers, drawing on empirical data from a study on Indigenous patrols in New South Wales, Australia. Patrol workers interact with public officials from various state entities who are tasked with overseeing funding, carrying out evaluations and, to varying degrees, monitoring the ‘effectiveness’ of local patrol operations. These interactions illuminate several issues regarding the ways in which knowledge about patrols is created, contested and communicated between Indigenous and non-Indigenous domains. The emergent patterns of these relations can be described as ‘seagull syndrome’, which involves the privileging of some types of knowledge over others in decision-making regarding Indigenous affairs, often with disastrous consequences for Indigenous organisations and communities. The paper documents the core features of seagull syndrome with respect to the discrete practices, everyday decision-making and mundane communication between public officials and patrol workers in New South Wales. It considers the implications of seagull syndrome for policy-makers and academics working in the Indigenous justice space and suggests ways to resist or challenge this tendency

ДОСЛІДЖЕННЯ БАГАТОВИМІРНОЇ АНАЛІТИЧНОЇ МОДЕЛІ ДЖЕРЕЛА ЗАВАД У ДИСКРЕТНИХ КАНАЛАХ З'ВЯЗКУ

The method of checking the adequacy of the stochastic models of noise sources in binary discrete communication channels is suggested. The technique is based on a comparison of the theoretical probability characteristics of a discrete channel and corresponding statistical estimates of these characteristics. Bit error probability, average error-free interval length and probability of receiving a block of a certain length with errors and errors with fixed multiplicity were chosen as main indexes of estimated characteristics. The calculations for the four-dimensional model were performed. The ratios of theoretical and statistical estimates of the discrete channel parameters were determined.

The missing link: does tunnelling nanotube-based supercellularity provide a new understanding of chronic and lifestyle diseases?

Tunnelling nanotubes (TNTs) are increasingly recognized as central players in a multitude of cellular mechanisms and diseases. Although their existence and functions in animal organisms are still elusive, emerging evidence suggests that they are involved in developmental processes, tissue regeneration, viral infections or pathogen transfer, stem cell differentiation, immune responses as well as initiation and progression of neurodegenerative disorders and cancer (see Sisakhtnezhad & Khosravi 2015 Eur. J. Cell Biol. 94, 429–443. (doi:10.1016/j.ejcb.2015.06.010)). A broader field of vision, including their striking functional and structural resemblance with nanotube-mediated phenomena found throughout the phylogenetic tree, from plants down to bacteria, points to a universal, conserved and tightly regulated mechanism of cellular assemblies. Based on our initial definition of TNTs as open-ended channels mediating membrane continuity between connected cells (Rustom et al. 2004 Science 303, 1007–1010. (doi:10.1126/science.1093133)), it is suggested that animal tissues represent supercellular assemblies that—besides opening discrete communication pathways—balance diverse stress factors caused by pathological changes or fluctuating physiological and environmental conditions, such as oxidative stress or nutrient shortage. By combining current knowledge about nanotube formation, intercellular transfer and communication phenomena as well as associated molecular pathways, a model evolves, predicting that the linkage between reactive oxygen species, TNT-based supercellularity and the intercellular shuttling of materials will have significant impact on diverse body functions, such as cell survival, redox/metabolic homeostasis and mitochondrial heteroplasmy. It implies that TNTs are intimately linked to the physiological and pathological state of animal cells and represent a central joint element of diverse diseases, such as neurodegenerative disorders, diabetes or cancer.

Analysis of the productivity for anti-interference digital data transmission systems

Background. Efficient use of energy and frequency resource of telecommunications channels is needed.Objective. The main objective is favor of a discrete communication channel productivity analysis through the identification of behavior reliability patterns of the demodulated multiposition signal manipulation by using redundant error-correcting codes.Methods. Estimation of capacity and productivity extremum search is carried out in a gradual increase in the symbol rate. Potential noise immunity given estimated error probability value in the telecommunications channel with white noise. The error probability is achieved by correcting the reliability of the demodulator output using the error-correcting code.Results. Analysis of capacity and productivity without coding, and with coding showed the extremum in the respective energy condition in the channel. At the same time at the point of extremum signals QAM-16 are more productive than the signals QAM-64. This demonstrates the rationality of using signals QAM-64 in wireless systems (e.g., LTE), that are useful to apply in a wide area wireless systems (e.g., UMTS).Conclusions. Efficient use of energy resources and frequency of communication channels occurs at the point of extremum performance in well-defined values of the transfer rate. If the maximum is to the left of it is advisable to increase the symbol rate, and if the maximum is to the right the signal energy can be reduced.

The Multiobjective Constraint Fault-Tolerant Control of Event-Triggered Nonuniform Transmission for Networked T-S Fuzzy System

Considering the actuator saturation, the problem of multiobjective constraint fault-tolerant control is studied for networked Takagi-Sugeno (T-S) fuzzy system with nonuniform transmission period under the discrete event-triggered communication scheme (DETCS). A closed-loop T-S fuzzy model, which includes numerous factors such as actuator saturation, actuator failure, event-triggered condition, nonuniform transmission period, and time-delay, is established by transferring the effect caused by the nonuniform update period on the system performance into the effect on the system time-delay. Two stability criterions and two codesign methods for the networked closed-loop failure T-S fuzzy system are developed based on a discontinuous Lyapunov-Krasovskii function, a kind of reciprocally convex technology, a kind of linear convex combination, Jessen inequality, and Wirtinger’s inequality. A feasible experimental solution is presented to verify the effectiveness of the proposed method.

THE PRODUCTIVITY ANALYSIS OF DISCRETE INFORMATION TRANSFER SYSTEM NEAR THE SHANNON BORDER

Background . The channel capacity and productivity assessments give a chance for system analysis principles development in applied information theory for telecommunication. The central task in the bounds of proposed research is the study of discrete communication channel (DCC) productivity. Objective . The main task of current study is the analysis of discrete communication channel productivity through identification of the regularities in reliability behavior of modulation multi-position keying signals. Integral productivity assessment is the efficiency metric of communication channel resources usage. Methods . Solving the main task of research generates bases on the finding the informational abilities of DCC. These informational abilities of DCC are expressed by channel capacity and productivity per one bit of information. Finally, the interdependence between channel capacity and productivity expressed via the information efficiency parameter ηС is the integral productivity estimation for DCC. The case of multiposition keying with its varieties (QPSK, QAM-16, and QAM-64) creates significantly bigger interest. Results . The model of discrete information transfer system that includes source of discrete information, discrete communication channel with noises, defined method of signal forming and method of transmitted signal processing is proposed. The regularities of the channel capacity change and productivity of a discrete communication channel are established. Conclusions . The hypothesis of the existence of extremes (upper limits) performance of DCС in discrete information transfer system without limitations on the received signals reliability is practically substantiated. Quantitative assessment to the discrete communication channel capacity and productivity extrema are presented. For achieving required reliability is noiseimmunity coding as the recognized tool in the Shannon theory. Key words : information theory; productivity; channel capacity; information efficiency.

Adaptive feedback synchronisation of complex dynamical network with discrete-time communications and delayed nodes

This paper considered the synchronisation of continuous complex dynamical networks with discrete-time communications and delayed nodes. The nodes in the dynamical networks act in the continuous manner, while the communications between nodes are discrete-time; that is, they communicate with others only at discrete time instants. The communication intervals in communication period can be uncertain and variable. By using a piecewise Lyapunov–Krasovskii function to govern the characteristics of the discrete communication instants, we investigate the adaptive feedback synchronisation and a criterion is derived to guarantee the existence of the desired controllers. The globally exponential synchronisation can be achieved by the controllers under the updating laws. Finally, two numerical examples including globally coupled network and nearest-neighbour coupled networks are presented to demonstrate the validity and effectiveness of the proposed control scheme.

Discrete Event-Triggered Robust Fault-Tolerant Control for Nonlinear Networked Control Systems withα-Safety Degree and Actuator Saturation

This paper deals with the discrete event-triggered robust fault-tolerant control problem for uncertain nonlinear networked control systems (NNCSs) withα-safety degree. A discrete event-triggered communication scheme (DETCS) is initially proposed, and a closed-loop fault model is subsequently established for NNCSs with actuator saturation under the DETCS. Based on an appropriately constructed delay-dependent Lyapunov–Krasovskii function, sufficient conditions are derived to guarantee the asymptotic stability of NNCSs under two different event-triggered conditions and are established as the contractively invariant sets of fault tolerance withα-safety degree. Furthermore, codesign methods between the robust fault-tolerant controller and event-triggered weight matrix are also proposed in terms of linear matrix inequality. The simulation shows that the resultant closed-loop fault NNCSs possesses a high safety margin, and an improved dynamic performance, as well as a reduced communication load. A comparative analysis of the two event-triggered conditions is discussed in the experiment section.