Design of coordination strategies in multiagent systems via genetic fuzzy systems

This paper suggests an evolutionary approach to design coordination strategies for multiagent systems. Emphasis is given to auction protocols since they are of utmost importance in many real world applications such as power markets. Power markets are one of the most relevant instances of multiagent systems and finding a profitable bidding strategy is a key issue to preserve system functioning and improve social welfare. Bidding strategies are modeled as fuzzy rule-based systems due to their modeling power, transparency, and ability to naturally handle imprecision in input data, an essential ingredient to a multiagent system act efficiently in practice. Specific genetic operators are suggested in this paper. Evolution of bidding strategies uncovers unknown and unexpected agent behaviors and allows a richer analysis of auction mechanisms and their role as a coordination protocol. Simulation experiments with a typical power market using actual thermal plants data show that the evolutionary, genetic-based design approach evolves strategies that enhance agents profitability when compared with the marginal cost-based strategies commonly adopted     

Genetic algorithms (GAs) to evolve multiple-agent cooperative systems

In this paper, a microrobot soccer-playing game, such as that of MIROSOT (Microrobot World Cup Soccer Tournament), is adopted as a standard test bed for research on multiple-agent cooperative systems. It is considerably complex and requires expertise in several difficult research topics, such as mobile microrobot design, motor control, sensor technology, intelligent strategy planning, etc., to build up a complete system to play the game. In addition, because it is an antagonistic game, it appears ideal to test whether one method is better than other. To date there have been two different kinds of architecture for building such system. One is called vision-based or centralized architecture, and the other is known as robot-based or decentralized architecture. The main difference between them lies in whether there exists a host computer system which responds to data processing and strategy planning, and a global vision system which can view the whole playground and transfer the environment information to the host computer in real time. We believe that the decentralized approach is more advanced, but in the preliminary step of our study, we used the centralized approach because it can lighten any overload of the microrobot design. In this paper, a simplified layer model of the multiple-agent cooperative system is first proposed. Based on such a model, a system for a microrobot soccer-playing game is organized. At the same time a simple genetic algorithm (SGA) is used for the autonomous evolution of cooperative behavior among microrobots. Finally, a computer simulation system is introduced and some simulated results are explained. This work was presented, in part, at the Third International Symposium on Artificial Life and Robotics, Oita, Japan, January 19–21, 1998.     

Consensus seeking in multiagent systems under dynamically changing interaction topologies

This note considers the problem of information consensus among multiple agents in the presence of limited and unreliable information exchange with dynamically changing interaction topologies. Both discrete and continuous update schemes are proposed for information consensus. This note shows that information consensus under dynamically changing interaction topologies can be achieved asymptotically if the union of the directed interaction graphs have a spanning tree frequently enough as the system evolves.     

Fully distributed consensus for fuzzy multiagent systems with jointly connected topology

This paper investigates the problem of fully distributed consensus for polynomial fuzzy multiagent systems (MASs) under jointly connected topologies. First, a polynomial fuzzy modeling method is presented to characterize the error dynamics that is constructed by one leader and multiple followers. Then, using the relative state information and the agents' dynamics, a distributed adaptive protocol is designed to guarantee that MASs under jointly connected topologies can achieve consensus in a fully distributed fashion. Utilizing the Lyapunov technique, a relaxed sufficient criterion is proposed to ensure consensus for fuzzy MASs under jointly connected topologies. Moreover, the adaptive coupling weights between neighboring agents can converge to certain values. The derived condition is transformed into a sum-of-squares form, which can be solved numerically. We provide an example to illustrate the proposed distributed adaptive consensus technique's validity.     

Genetic tuning of fuzzy inference within fuzzy classifier systems

Abstract: In generating a suitable fuzzy classifier system, significant effort is often placed on the determination and the fine tuning of the fuzzy sets. However, in such systems little thought is given to the way in which membership functions are combined within the fuzzy rules. Often traditional fuzzy inference strategies are used which consequently provide no control over how strongly or weakly the inference is applied within these rules. Furthermore such strategies will allow no interaction between grades of membership. A number of theoretical fuzzy inference operators have been proposed for both regression and classification problems but they have not been investigated in the context of real-world applications. In this paper we propose a novel genetic algorithm framework for optimizing the strength of fuzzy inference operators concurrently with the tuning of membership functions for a given fuzzy classifier system. Each fuzzy system is generated using two well-established decision tree algorithms: C4.5 and CHAID. This will enable both classification and regression problems to be addressed within the framework. Each solution generated by the genetic algorithm will produce a set of fuzzy membership functions and also determine how strongly the inference will be applied within each fuzzy rule. We investigate several theoretical proven fuzzy inference techniques (T-norms) in the context of both classification and regression problems. The methodology proposed is applied to a number of real-world data sets in order to determine the effects of the simultaneous tuning of membership functions and inference parameters on the accuracy and robustness of fuzzy classifiers.     

Adding security and trust to multiagent systems

Multiagent systems (MASs) are societies whose individuals are software delegatees (agents) acting on behalf of their owners or delegators (people or organizations). When deployed in an open network such as the Internet, MASs face some trust and security issues. Agents comeand go, and interact with strangers. Assumptions about security and general trustworthiness of agents and their deployers are inadequate in this context. In this paper, the design of a security infrastructure is presented applicable to MASs in general. This design addresses both security threats and trust issues. In this design, there are mechanisms for ensuring secure communication among agents and secure naming and resource location services. And two types of trusts are addressed: trust that agents will not misbehave and trust that agents are really delegatees of whom they claim to be. To establish the first type of trust, deployers of agents are made liable for the actions of their agents; to establish the second type of trust, it is proposed that agents prove that they know secrets that only their delegators know.     

A multiagent architecture for fuzzy modeling

In this paper a hybrid learning system that combines different fuzzy modeling techniques is being investigated. In order to implement the different methods, we propose the use of intelligent agents, which collaborate by means of a multiagent architecture. This approach, involving agents which embody the different problem solving methods, is a potentially useful strategy for enhancing the power of fuzzy modeling systems. ©1999 John Wiley & Sons, Inc.     

Opportunities for multiagent systems and multiagent reinforcement learning in traffic control

The increasing demand for mobility in our society poses various challenges to traffic engineering, computer science in general, and artificial intelligence and multiagent systems in particular. As it is often the case, it is not possible to provide additional capacity, so that a more efficient use of the available transportation infrastructure is necessary. This relates closely to multiagent systems as many problems in traffic management and control are inherently distributed. Also, many actors in a transportation system fit very well the concept of autonomous agents: the driver, the pedestrian, the traffic expert; in some cases, also the intersection and the traffic signal controller can be regarded as an autonomous agent. However, the “agentification” of a transportation system is associated with some challenging issues: the number of agents is high, typically agents are highly adaptive, they react to changes in the environment at individual level but cause an unpredictable collective pattern, and act in a highly coupled environment. Therefore, this domain poses many challenges for standard techniques from multiagent systems such as coordination and learning. This paper has two main objectives: (i) to present problems, methods, approaches and practices in traffic engineering (especially regarding traffic signal control); and (ii) to highlight open problems and challenges so that future research in multiagent systems can address them.     

Modelling social norms in multiagent systems

Social norms are cultural phenomena that naturally emerge in human societies and help to prescribe and proscribe normative patterns of behaviour. In recent times, the discipline of multi-agent systems has been used to model social norms in an artificial society of agents. In this paper we review norms in multi-agent systems and then explore a series of norms in a simulated urban traffic setting. Using game-theoretic concepts we define and offer an account of norm stability. Particularly in small groups, a relatively small number of individuals with cooperative attitude are needed for the norm of cooperation to evolve and be stable. In contrast, in larger populations, a larger proportion of cooperating individuals are required to achieve stability.     

Communication in reactive multiagent robotic systems

Multiple cooperating robots are able to complete many tasks more quickly and reliably than one robot alone. Communication between the robots can multiply their capabilities and effectiveness, but to what extent? In this research, the importance of communication in robotic societies is investigated through experiments on both simulated and real robots. Performance was measured for three different types of communication for three different tasks. The levels of communication are progressively more complex and potentially more expensive to implement. For some tasks, communication can significantly improve performance, but for others inter-agent communication is apparently unnecessary. In cases where communication helps, the lowest level of communication is almost as effective as the more complex type. The bulk of these results are derived from thousands of simulations run with randomly generated initial conditions. The simulation results help determine appropriate parameters for the reactive control system which was ported for tests on Denning mobile robots.     

Integrating trust measures in multiagent systems

Several models have been proposed in the past for representing both reliability and reputation. However, we remark that a crucial point in the practical use of these two measures is represented by the possibility of suitably combining them to support the agent's decision. In the past, we proposed a reliability–reputation model, called RRAF, that allows the user to choose how much importance to give to the reliability with respect to the reputation. However, RRAF shows some limitations, namely: (i) The weight to assign to the reliability versus reputation is arbitrarily set by the user, without considering the system evolution; (ii) the trust measure that an agent a perceives about an agent b is completely independent of the trust measure perceived by each other agent c, while in the reality the trust measures are mutually dependent. In this paper, we propose an extension of RRAF, aiming at facing the limitations above. In particular, we introduce a new trust reputation model, called TRR, that considers, from a mathematical viewpoint, the interdependence among all the trust measures computed in the systems. Moreover, this model dynamically computes a parameter measuring the importance of the reliability with respect to the reputation. Some experiments performed on the well‐known ART(Agent Reputation and Trust) platform show the significant advantages in terms of effectiveness introduced by TRR with respect to RRAF. © 2011 Wiley Periodicals, Inc.     

Strategic best-response learning in multiagent systems

We present a novel and uniform formulation of the problem of reinforcement learning against bounded memory adaptive adversaries in repeated games, and the methodologies to accomplish learning in this novel framework. First we delineate a novel strategic definition of best response that optimises rewards over multiple steps, as opposed to the notion of tactical best response in game theory. We show that the problem of learning a strategic best response reduces to that of learning an optimal policy in a Markov Decision Process (MDP). We deal with both finite and infinite horizon versions of this problem. We adapt an existing Monte Carlo based algorithm for learning optimal policies in such MDPs over finite horizon, in polynomial time. We show that this new efficient algorithm can obtain higher average rewards than a previously known efficient algorithm against some opponents in the contract game. Though this improvement comes at the cost of increased domain knowledge, simple experiments in the Prisoner's Dilemma, and coordination games show that even when no extra domain knowledge (besides that an upper bound on the opponent's memory size is known) is assumed, the error can still be small. We also experiment with a general infinite-horizon learner (using function-approximation to tackle the complexity of history space) against a greedy bounded memory opponent and show that while it can create and exploit opportunities of mutual cooperation in the Prisoner's Dilemma game, it is cautious enough to ensure minimax payoffs in the Rock–Scissors–Paper game.     

Declarative representations of multiagent systems

This paper explores the specification and semantics of multiagent problem-solving systems, focusing on the representations that agents have of each other. It provides a declarative representation for such systems. Several procedural solutions to a well-known test-bed problem are considered, and the requirements they impose on different agents are identified. A study of these requirements yields a representational scheme based on temporal logic for specifying the acting, perceiving, communicating, and reasoning abilities of computational agents. A formal semantics is provided for this scheme. The resulting representation is highly declarative, and useful for describing systems of agents solving problems reactively     

A Markovian jump system approach to consensus of heterogeneous multiagent systems with partially unknown and uncertain attack strategies

The problem of robust leader‐following consensus of heterogeneous multiagent systems subject to deny‐of‐service attacks is investigated, where attack strategies are partially unknown and uncertain to defender. A Markovian jump system approach is proposed, that is, capable of describing the occurrence of different attack strategies, and the occurring probability of each attack strategy is represented by the transition probability of the Markovian jump model. Then, sufficient conditions are derived such that the output tracking performance can be guaranteed. In order to design the controller gains, some slack matrices are introduced, which can provide some design freedom. Finally, it is shown that the controller design results can be applied to the multivehicle position‐tracking system. The simulation results reveal that the consensus performance is much better if one has more statistics information on attacks.     

Controllability of heterogeneous multiagent systems

The existing results on controllability of multiagent systems (MASs) are mostly based on homogeneous nodes. This paper focuses on controllability of heterogeneous MASs, where the agents are modeled as two types. One type is that the agents have the same high‐order dynamics, and the interconnection topologies of the information flow in different orders are supposed to be different; the other type is that the agents have generic linear dynamics, and the dynamics are supposed to be heterogeneous. For the first type, the necessary and sufficient condition for controllability of heterogeneous‐topology system is derived via combination of Laplacian matrices. For the second type, the contribution also has two parts. The first part supposes that the agents have the same dimensional states and proves that controllability of this kind of MASs is equivalent to the controllability of each node and the whole interconnection topology, while the last parameter of the state feedback vector must not be 0. The second part supposes that the agents may have different dimensional states. For this kind of systems, the concept of β‐controllability is proposed. The necessary and sufficient condition for β‐controllability of heterogeneous‐dynamic systems is also derived and it is also proved that the feedback gain vectors have the effect to improve controllability. Different illustrative examples are provided to demonstrate the effectiveness of the theoretical results in this paper.     

Introduction to the special issue on normative multiagent systems

This special issue contains four selected and revised papers from the second international workshop on normative multiagent systems, for short NorMAS07 (Boella et al. (eds) Normative multiagent systems. Dagstuhl seminar proceedings 07122, 2007), held at Schloss Dagstuhl, Germany, in March 2007. At the workshop a shift was identified in the research community from a legal to an interactionist view on normative multiagent systems. In this editorial we discuss the shift, examples, and 10 new challenges in this more dynamic setting, which we use to introduce the papers of this special issue.     

A platform-independent metamodel for multiagent systems

Various agent-oriented methodologies and metamodels exist to design and develop multiagent systems (MAS) in an abstract manner. Frequently, these frameworks specialise on particular parts of the MAS and only few works have been invested to derive a common standardisation. This limits the impact of agent-related systems in commercial applications. In this paper, we present a metamodel for agent systems that abstracts from existing agent-oriented methodologies, programming languages, and platforms and could thus be considered as platform-independent. This metamodel defines the abstract syntax of a proposed domain-specific modelling language for MAS that is currently under development and provides furthermore the base to generate code out of the generated designs. This is done by applying the principles of model-driven development (MDD) and providing two model transformations that allow transforming the generated models into textual code that can be executed with JACK and JADE.     

分布式网络系统的任意预设时间分组一致

研究任意预设时间控制下的多智能体网络系统分组一致性问题.设计非零分组投影参数下任意预设时间控制协议,使得分布式网络系统在物理允许范围内的任意预设时间内迅速实现分组一致,该预设时间与系统参数和初始值都无关系.基于代数图论、李雅普诺夫稳定性和矩阵理论等,分别讨论无向和有向拓扑网络情形下,多智能体系统实现预设时间分组一致的充分条件.独轮车的多智能体系统仿真实验验证了所提方法的有效性.