Communication In Multi-agent Systems Research Articles

A Cost-Aware Multi-Agent System for Black-Box Design Space Exploration

Abstract Effective coordination of design teams must account for the influence of costs incurred while searching for the best design solutions. This paper introduces a cost-aware multi-agent system (MAS), a theoretical model to 1) explain how individuals in a team should search, assuming that they are all rational utility-maximizing decision-makers, and 2) study the impact of cost on the search performance of both individual agents and the system. First, we develop a new multi-agent Bayesian Optimization framework accounting for information exchange among agents to support their decisions on where to sample in search. Second, we employ a reinforcement learning approach based on the multi-agent deep deterministic policy gradient for training MAS to identify where agents cannot sample due to design constraints. Third, we propose a new cost-aware stopping criterion for each agent to determine when costs outweigh potential gains in search as a criterion to stop. Our results indicate that cost has a more significant impact on MAS communication in complex design problems than in simple ones. For example, when searching in complex design spaces, some agents could initially have low-performance gains, thus stopping prematurely due to negative payoffs, even if those agents could perform better in the later stage of the search. Therefore, global-local communication becomes more critical in such situations for the entire system to converge. The proposed model can serve as a benchmark for empirical studies to quantitatively gauge how humans would rationally make design decisions in a team.

An effective approach for reducing data redundancy in multi-agent system communication

The redundancy of the data is an active research topic. While an agent works in a multi-agent system, the number of messages between them increases. This is due to the fact that the functionalities data depends on other agents in terms of functional requirements. Typically, only one agent in a multi-agent system is responsible for accessing a database instead of replicating the database on each agent. A database is stored on multiple agents rather than a single agent to avoid a single point of failure. In this approach, the system has a higher load because one agent is responsible for all agent queries and must send duplicate messages to multiple agents, resulting in redundant data. In this research, we present Multi-Agent System for Commodity Data (MASCD) framework, the multi-agent system based communication using the distributed hash system, to reduce data redundancy in multi-agent system communication. Our anticipated method demonstrated how we divided the database names and efficiently distributed data to each agent. The database splitting is based on manufacturer names or product names. We utilize a table based on prime numbers. Through the hash function, we ascertain the index of the agent granted access to the relevant data. Each agent is accountable for its data. We use a Distributed Hash Table for efficient querying that stores data as key-value pairs. Each agent maintains a Finger Table containing the next and previous nodes for agent communication purposes. Using FIPA messages, we demonstrated how an agent could interact optimally. In conclusion, we illustrate the application of the proposed approach through a case study of mobile phones and university information systems.

Time- and Event-triggered Communication for Multi-agent Systems – Part I: General Framework

Time- and Event-triggered Communication for Multi-agent Systems – Part I: General Framework

Time- and Event-triggered Communication for Multi-agent Systems – Part II: Digital Implementation and Resilience

Time- and Event-triggered Communication for Multi-agent Systems – Part II: Digital Implementation and Resilience

Model checking combined trust and commitments in Multi-Agent Systems

Trust and social commitments have been studied with different objectives for communication in Multi-Agent Systems (MASs) separately. The purpose of this paper is to present the first logical framework to explore the relationship between two key concepts in autonomous MASs, namely trust and commitments among agents, along with its model checking algorithms. An analysis is performed on this relationship, with a specific emphasis on the perspectives of semantics and model checking. The analysis is carried out within the framework of Trust Computation Tree Logic with Commitments (CTLC) logic, an extended logic built on top of the standard branching temporal Computation Tree Logic (CTL) by including modalities for reasoning about commitments and their fulfillment. Moreover, the study employs a trust operator to facilitate trust reasoning. We propose thus Trust Computation Tree Logic with Commitments (TCTLC), a new logic able to express properties about trust and commitments simultaneously. Such rich properties cannot be expressed in other languages. We introduce the semantics using the extended formalism of interpreted systems. We propose some postulates and proofs that show how we can reason about combined trust and commitments in the logic proposed. New Binary Decision Diagram (BDD)-based algorithms for our logic are presented and implemented as additional libraries of the Model Checker for Multi-Agent Systems (MCMAS), the default model checker of MASs. Furthermore, we prove that although TCTLC extends CTL, its model checking algorithm remains polynomial (P-complete) with respect to the size of the model and the length of the formula. Similarly, the space complexity for concurrent programs also remains unchanged, which is polynomial (PSPACE-complete) with respect to the size of the components within these programs. Finally, to evaluate the proposed technique, we report the experimental results using an industrial case study and compare the results with those obtained on relevant benchmarks.

Deep Reinforcement Learning Based Task-Oriented Communication in Multi-Agent Systems

Deep Reinforcement Learning Based Task-Oriented Communication in Multi-Agent Systems

Adaptive fault-tolerant control of heterogeneous MASs based on dynamic event-triggered communication

This paper describes a solution to the problem of fault-tolerant control (FTC) problem for leader-following heterogeneous linear multi-agent systems (MASs) subject to loss of effectiveness and time-varying additive actuator faults. The main goal of the proposed FTC approach is to maintain the behaviour of the heterogeneous MASs in spite of actuator faults and to save on bandwidth required for the MASs communication by using a dynamic event-triggered communication mechanism. Firstly, a fault estimation observer, which deals with linear systems with matched disturbances, is constructed to estimate simultaneously the actuator faults and the followers' states. Then, an FTC with dynamic event-triggered communication is proposed. The latter is based on a dynamic internal reference for each agent and exploits the estimated faults. Thus, it is at first shown that the FTC problem can be indirectly solved through the consensus tracking of the defined internal references. Then, a sufficient condition of consensus tracking for the internal references with a dynamic event-triggered communication is proposed. Finally, a sufficient and necessary condition is derived for the FTC of linear heterogeneous multi-agent systems. The effectiveness of the proposed approach is illustrated with a numerical simulation.

Interaction history as a source of compositionality in emergent communication

Abstract In this paper, we explore interaction history as a particular source of pressure for achieving emergent compositional communication in multi-agent systems. We propose a training regime implementing template transfer, the idea of carrying over learned biases across contexts. In the presented method, a sender-receiver dyad is first trained with a disentangled pair of objectives, and then the receiver is transferred to train a new sender with a standard objective. Unlike other methods (e.g. the obverter algorithm), the template transfer approach does not require imposing inductive biases on the architecture of the agents. We experimentally show the emergence of compositional communication using topographical similarity, zero-shot generalization and context-independence as evaluation metrics. The presented approach is connected to an important line of work in semiotics and developmental psycholinguistics: it supports a conjecture that compositional communication is scaffolded on simpler communication protocols.

A New Distributed Energy Management Strategy for Smart Grid With Stochastic Wind Power

In this article, a distributed energy management strategy is proposed for smart grid with stochastic wind power. The different Weibull probability distribution function (PDF) of wind power during different operation intervals, coupled with ramp rate limits of thermal generators across operation intervals, is considered in this operation cost optimization problem. Under multiagent system communication framework, we first decompose the problem into subproblems associated to operation interval. For each operation interval, parameters of Weibull PDF are estimated by maximum likelihood method according to wind speed obtained by wind speed forecast model of committed wind turbine. Then, a consensus-based alternating direction method of multipliers (ADMM) is proposed to get the optimal operation plan for all committed generators, where Armijo line search algorithm is employed to adaptively obtain the step size in the iterations. The simulation results show that the proposed consensus-based ADMM outperforms distributed projected gradient method in previous study on energy management with wind power; the proposed distributed strategy is valid for short-term energy management for smart grid with stochastic wind power.

Design of one type of linear network prediction controller for multi-agent system

In this paper, to solve network delay and network tracking control problems in multi-agent system communication, a design method of network prediction controller was introduced based on state difference estimation and output tracking error. This design method not only effectively compensates for influences of network delay on the system but also ensures stability of the closed-loop system and realizes the same tracking performance among multi-agents. The effectiveness of the proposed method was proven by simulation experiment.The key innovation in the paper is that the influence of network delay on the system was actively compensated and a network prediction tracking control mechanism was proposed to guarantee the stability of the closed-loop system.The proposed method achieved the same tracking with the local tracking control system under certain conditions.

A Novel Agent-Based Distributed Power Flow Solver for Smart Grids

The goal of this paper is to present a novel agent based distributed power flow solver for general unbalanced radial distribution systems utilizing multiagent system communication. The distributed solver makes use of intelligent agents, which possess three key characteristics, namely autonomy, local view, and decentralization. Comprehensive models of distribution systems will be considered, including lines, switching devices, voltage regulators, transformers, shunt capacitors, distributed energy storage systems, and different load types. The distributed intelligent agents will use the backward/forward sweep technique to iteratively solve the power flow. A distinguishing feature of the new solver is that it deals with the problem from a completely distributed perspective. The proposed solution algorithm is evaluated on three standard IEEE distribution test systems with very promising results.

The Formal Semantics of a Domain-Specific Modeling Language for Semantic Web Enabled Multi-Agent Systems

Development of agent systems is without question a complex task when autonomous, reactive and proactive characteristics of agents are considered. Furthermore, internal agent behavior model and interaction within the agent organizations become even more complex and hard to implement when new requirements and interactions for new agent environments such as the Semantic Web are taken into account. We believe that the use of both domain specific modeling and a Domain-specific Modeling Language (DSML) may provide the required abstraction and support a more fruitful methodology for the development of Multi-agent Systems (MASs) especially when they are working on the Semantic Web environment. Although syntax definition based on a metamodel is an essential part of a modeling language, an additional and required part would be the determination and implementation of DSML constraints that constitute the (formal) semantics which cannot be defined solely with a metamodel. Hence, in this paper, formal semantics of a MAS DSML called Semantic Web enabled Multi-agent Systems (SEA_ML) is introduced. SEA_ML is a modeling language for agent systems that specifically takes into account the interactions of semantic web agents with semantic web services. What is more, SEA_ML also supports the modeling of semantic agents from their internals to MAS perspective. Based on the defined abstract and concrete syntax definitions, we first give the formal representation of SEA_ML's semantics and then discuss its use on MAS validation. In order to define and implement semantics of SEA_ML, we employ Alloy language which is declarative and has a strong description capability originating from both relational and first-order logic in order to easily define complex structures and behaviors of these systems. Differentiating from similar contributions of other researchers on formal semantics definition for MAS development languages, SEA_ML's semantics, presented in this paper, defines both static and dynamic aspects of the interaction between software agents and semantic web services, in addition to the definition of the semantics already required for agent internals and MAS communication. Implementation with Alloy makes definition of SEA_ML's semantics to include relations and sets with a simple notation for MAS model definitions. We discuss how the automatic analysis and hence checking of SEA_ML models can be realized with the defined semantics. Design of an agent-based electronic barter system is exemplified in order to give some flavor of the use of SEA_ML's formal semantics. Lessons learned during the development of such a MAS DSML semantics are also reported in this paper.

English

English

Evaluation for Acquiring Method for Agents’ Actions Using Pheromone Communication in Multi-Agent System

This paper has presented acquiring method for agents' actions using Ant Colony System (ACS) in multi-agent system. ACS is one of powerful meta-heuristics algorithms and some researchers have reported the effectiveness of some applications with the algorithm. I have developed agents with proposed method in a multi-agent system. The final goal of this research is an achievement of co-operations for hetero-agent in multi-agent systems. In this research for implementation for one type of agents is the first step of this goal. Then I have reported results of evaluation experiments using these agents. Index Terms—Machine learning, multi-agent system, swarm intelligence, ant colony optimization. I. I NTRODUCTION Recently, some researchers have reported the effectiveness of systems installed swarm intelligence algorithms (1)-(3). Ant Colony Optimization (ACO) and Ant Colony System (ACS) have become a very successful and widely used in some applications. These algorithms have been used in programs for network routing, traffic control programs, Traveling Salesman Problem (TSP) and so on (4)-(6). In real ants' feeding actions, they are able to find the shortest path from a food source to their nest by pheromone information. Pheromone is one of chemical materials and ants deposit pheromone on a path between a food source to their nest and it become one candidate solution. In a case that other ants trail a path deposited pheromone, the candidate is reinforced. Moreover pheromone disappears into air as time progresses. Ants have done these processes iteratively. The behavior of real ants has inspired ACO and ACS. The system based on ACO and ACS are used artificial ants cooperate to the solution of a problem by exchanging information via pheromone.

SecuredTrust: A Dynamic Trust Computation Model for Secured Communication in Multiagent Systems

Security and privacy issues have become critically important with the fast expansion of multiagent systems. Most network applications such as pervasive computing, grid computing, and P2P networks can be viewed as multiagent systems which are open, anonymous, and dynamic in nature. Such characteristics of multiagent systems introduce vulnerabilities and threats to providing secured communication. One feasible way to minimize the threats is to evaluate the trust and reputation of the interacting agents. Many trust/reputation models have done so, but they fail to properly evaluate trust when malicious agents start to behave in an unpredictable way. Moreover, these models are ineffective in providing quick response to a malicious agent's oscillating behavior. Another aspect of multiagent systems which is becoming critical for sustaining good service quality is the even distribution of workload among service providing agents. Most trust/reputation models have not yet addressed this issue. So, to cope with the strategically altering behavior of malicious agents and to distribute workload as evenly as possible among service providers; we present in this paper a dynamic trust computation model called SecuredTrust.” In this paper, we first analyze the different factors related to evaluating the trust of an agent and then propose a comprehensive quantitative model for measuring such trust. We also propose a novel load-balancing algorithm based on the different factors defined in our model. Simulation results indicate that our model compared to other existing models can effectively cope with strategic behavioral change of malicious agents and at the same time efficiently distribute workload among the service providing agents under stable condition.

Communication Security in MAS with XML Security Specifications

With widely applications of Multi-Agent System (MAS) on internet and intranet, how to ensure security regarding multi-agent communication is a big issue. Traditional cryptographic technologies support a coarse granularity security. This paper adopts XML security technology to provide a security mechanism for MAS communication. With the combination of XML security specifications and MAS, the paper presents a fine-grained security model for MAS communication. The model proposed can satisfy the security requirements of MAS communication based on XML data format.

Revealing bullying patterns in multi-agent systems

Multi-agent systems communication is arguably difficult to be designed. An equal balance of communication among agents is usually desirable. When this does not happen, it affects the quality of service, causing poor performances in terms of response times. The aim of this work is to detect undesirable patterns of communication in these types of systems, using the metaphor of bullying. Though observed in social environments, it can also be applied to a multi-agent system scenario: it consists of some agents’ continuous communications toward the same agents. This causes an unbalanced communication, overloading the receiver, which results in higher response times. Some metrics have been designed to measure the agents’ communication, and some rules classify agents accordingly into several patterns regarding the bullying metaphor. Furthermore, the experimental results prove that the metrics based on the bullying metaphor are strongly related with the quality of service of multi-agent systems. The experimental results also show that the causes of a bad quality of service can be figured out by means of this bullying metaphor and their metrics and classification rules.

Agent communication and social concepts

The importance of agent communication for multi-agent systems is, of course, beyond any doubt. However, the importance of research in this area seems less obvious. After several years with lots of discussions about standard agent communication languages and possible semantics for them it seems people have the feeling that all the issues in this area are settled. Despite some criticism on FIPA ACL this seems to be the de facto standard for agent communication. Especially after the JADE platform (which is probably the widest used in academic circles) was made FIPA compliant. However, there is still a wide gap between being able to parse and generate messages that conform to the FIPA ACL standard and being able to perform meaningful conversations.Inordertoforceorevenjustsupportagentstoperformmeaningfulconversations some form of a shared semantics of the communication process is needed. Since it is impossible to verify compliance of agents based on internal structures of the agents (that cannot be inspected) the semantics should be based on concepts that are externally observable. Hence the growing interest in the use of social concepts that can be observed and verified outside the agents. In thecontextofthese developmentswe are happyto presentthis specialissue.The papers selected for this special issue on agent communication are based on presentations at the Agent Communication workshop of 2004, held in New York. They clearly indicate the general trend in the past few years towards the use of social concepts in defining the semantics of agent communication. Especially “social commitments” and deontic concepts such as “obligations” seem to become a central element in this respect. In three of the four papers in this special issue social commitments form the basis of the theory discussed in that paper. The fourth paper is based on the use of obligations and permissions. The paper of Fornara et. al. discusses the use of “institutions” as an abstract set of rules that describe how social commitments are changed based on communication

Flexible Communication of Agents based on FIPA-ACL

Communication in multi-agent systems is an important subject of the current research. In this paper, the syntax and semantics of a multi-agent programming language, called ECCS, are defined. We focus specially on the communication of agents. The main contribution of this paper is a new and flexible way of communication of agents. We finally work out a well known protocol as an example.

Formal ReSpecT

Logic-based languages have already proved to be effective to build individual agents and to enable inter-agent communication in multi-agent systems. Also, logic tuple centres have shown that logic-based languages can be effectively exploited to rule inter-agent communication so as to build social behaviours. In this paper, we formally define the notion of logic tuple centre as well as the operational semantics of the logic-based language ReSpecT for the behaviour specification of logic tuple centres. For this purpose, we exploit a formal framework for asynchronous systems allowing coordination media to be represented in a separate and independent way with respect to the coordinated entities. As a by-product, this shows that a logic-based approach may be effectively exploited for the coordination of heterogeneous agents of different sorts and technologies.