Recent years are seeing an increasing need for on-line monitoring of teams of cooperating agents, e.g., for visualization, or performance tracking. However, in monitoring deployed teams, we often cannot rely on the agents to always communicate their state to the monitoring system. This paper presents

Multi-agent systems can benefit from the possibility of broadcasting messages to a wide audience. The audience may include overhearing agents which, unknown to senders, observe conversations and, among other things, pro-actively send suggestions. Current mainstream agent communication languages however lack adequate support for broadcasting. This paper defines the requirements for a form of broadcast that we call channeled multicast, whose distinguishing features include the ability to distinguish streams of messages by their theme, and to address agents by their characteristics. We present an implementation based on multicast IP, called LoudVoice. We show how channeled multicast could be used in alternative to matchmaking, and present in some detail a broadcast-based version of the English Auction Interaction Protocol. Finally, we discuss how we use the ability to overhear conversations in order to build innovative applications and we present a case study which is a testbed for various types of agents and multi-agent systems.

Overhearing is gaining attention as a generic method for cooperative monitoring of distributed, open, multiagent systems. It involves monitoring the routine conversations of agents–who know they are being overheard–to assist the agents, assess their progress, or suggest advice. While there have been several investigations of applications and methods of overhearing, no formal model of overhearing exists. This paper takes steps towards such a model. It first formalizes a conversation system–the set of conversations in a multi-agent system. It then defines a key step in overhearing–conversation recognition– identifying the conversations that took place within a system, given a set of overheard messages. We provide a skeleton algorithm for conversation recognition, and provide instantiations of it for settings involving no message loss, random message loss, and systematic message loss (such as always losing one side of the conversation). We analyze the complexity of these algorithms, and show that the systematic message loss algorithm, which is unique to overhearing, is significantly more efficient then the random loss algorithm (which is intractable). 1.

Implicit Culture is the relation between a set and a group of agents such that the elements of the set behave accordingly...

We propose group communication for agent coordination within "active rooms" and other pervasive computing scenarios featuring strict real-time requirements, inherently unreliable communication, and a large but continuously changing set of context-aware autonomous systems. Messages are exchanged over multicast channels, which may remind of chat rooms where everybody hears everything being told. The issues that have to be faced (e.g., changing users' preferences and locations; performance constraints; redundancies of sensors, actuators, and services; evolving sources of information; the continuous joining and leaving of agents on mobile devices) require the ability of dynamically selecting the "best" agents for providing a service in a given context. Our approach is based on the idea of implicit organization, which refers to the set of all agents willing to play a given role on a given channel; an implicit organization is a special form of team with no explicit formation phase and a single role involved. No middle agent is required by an implicit organization; instead, a set of protocols, designed for unreliable group communication, are used for two purposes: first, to negotiate a coordination policy; second, for actual team coordination. Preconditions and effects of these protocols are formalized by means of the joint intention theory (JIT). We sketch a general computational model for an agent participating to an implicit organization.

We propose group communication for agent coordination within “active rooms” and other pervasive computing scenarios featuring strict real-time requirements, inherently unreliable communication, and a large but continuously changing set of context-aware autonomous systems. Messages are exchanged over multicast channels, which may remind of chat rooms in which everybody hears everything being said. The issues that have to be faced (e.g., changing users’ preferences and locations; performance constraints; redundancies of sensors and actuators; agents on mobile devices continuously joining and leaving) require the ability of dynamically selecting the “best” agents for providing a service in a given context. Our approach is based on the idea of implicit organization, which refers to the set of all agents willing to play a given role on a given channel. An implicit organization is a special form of team with no explicit formation phase and a single role involved. No middle agent is required. A set of protocols, designed for unreliable group communication, are used to negotiate a coordination policy, and for team coordination. We sketch a general computational model for an agent participating to an implicit organization.

Helping behavior in effective teams is achieved via some overlapping “shared mental models ” that are developed and maintained by members of the team. In this paper, we take the perspective that multiparty “proactive ” communication is critical for establishing and maintaining such a shared mental model among teammates, which is the basis for agents to offer proactive help and to achieve coherent teamwork. We first provide formal semantics for multiparty proactive performatives within a team setting. We then examine how such performatives result in updates to mental model of teammates, and how such updates can trigger helpful behaviors from other teammates. We also provide conversation policies for multiparty proactive performatives. 1.

Abstract. Lately, overhearing has gained interest in monitoring multi-agent settings. Previous investigations provided an extensive set of techniques using overhearing. However, most previous investigations rely on a problematic assumption that all inter-agent communications can be overheard. In the real-world settings, it is reasonable to assume that the available overhearing resources will be essentially limited. Thus, overhearing targets should be carefully chosen. We provide a theoretical and empirical study of selective overhearing. In particular, we focus on overhearing hierarchical organizations that are highly popular in the real-world settings. This paper first presents a theoretical approach for modelling overhearing in hierarchical organizations. Then, based on the proposed model, we present experiments in simulating conversations in hierarchical organizations, and empirically examine a set of overhearing strategies particularly suited for such organizations. Based on these extensive experiments, we are able to determine efficient overhearing strategies and isolate the parameters influencing their behavior. 1

In the first year of the project, Bologna (UNIBO) will spend 14 months on WP1, a logic-based model for computees, and 10 months on WP2, modelling interactions between computees. Ferrara (DIFERRARA) will spend 12 months on WP1 and 8 months on WP2.

Abstract. Coordination of task execution, in open distributed multi-agent systems, is often accomplished through communications between agents. Thus, most tasks carry some cost associated with those communications. The work in this article proposes to use overhearing for reducing some of this cost, and therefore reducing the overall cost of task execution. Since in open distributed systems, and in particular in large-scale settings, it is rare for two agents to communicate directly, their communications (as in real-world networks) are routed to their destination through other agents. We allow those intermediate agents to overhear and monitor passing through communications. In doing so, the intermediate agents detect some of the errors generated by communicating agents before they reach their final destination. Filtering those erroneous messages in advance reduces the cost associated with routing them through the system. This work first formalizes this problem, and then proposes algorithms for finding an effective set of filtering agents. Unfortunately, finding an optimal solution for this problem turned out to be intractable. Thus, an efficient heuristic solution is proposed. An empirical simulation of these algorithms shows that the heuristic algorithm achieves similar performance to that of the optimal algorithm, while maintaining efficient run-time complexity. 1