Semantically Enabled Communication Framework for Autonomous Systems

Introduction

Agent technologies have gained a lot of attention in recent years. Moreover, advancement and on going research in multi-agent systems (MAS) domain has gained overwhelming attention in the recent years. A lot of ongoing research exhibits special interest in designing and implementing enterprise based agent software applications that will upkeep-distributed processes through artificial intelligence. Emergent applications of distributed agents are being utilized not only for Internet applications but also in network-based environments. As a breakthrough to the conventional systems, ADS was proposed (Mori et al., 1984). ADS can be defined as a composition of live things that have the autonomy and decentralization abilities to achieve continuous extension, maintenance, and fault tolerance without affecting the overall system. Principles of autonomous controllability and autonomous coordination make ADS one of the most promising technologies for future applications (Mori et al., 1993). In a generic MAS architecture, each agent (Wooldrige, 1998) has inadequate knowledge or abilities and a number of agents collectively work to accomplish a broader goal that is based on decentralization of data and control. In typical MAS communication among agents is not synchronized or centralized, but the agents may require support from one another for fulfillment of their goals. Agents require flexible mechanisms to define multi valued attributes and competencies, and they also require techniques for inferring these expressions either from individuals input or from computational derivations. A MAS is highly appropriate for real-time collaborative decision making process by utilizing the capabilities of autonomous agents (Wooldridge, 2002) (Singh et al., 2005) (Jennings, 2000), that constantly become accustomed to surroundings, and their strength depends on high-level inter-agent communication and negotiation.

Agent communication can be defined as a way of exchanging messages through which the connection among agents is articulated through the use of intermediary signals that can invoke agents to perform some action after interpretation (Tweedale et al., 2007). The information being exchanged among agents is encoded and decoded by using some language that keeps track of sent information when being transmitted and is later decoded by the receivers at arrival end. Agents have not been defined in particular and don’t have a well-accepted definition, but the main purpose is to achieve some goal through exchange of messages and to assist one another. It is almost crucial that agents reach an agreement on some standard prescribed model for the exchange of messages being exchanged. On top of underlying message transport infrastructure a high-level conversation support is required by the agents to have interactions with one another. A well-defined and structured content language is required at the application level, by agents for exchanging messages. In addition, they require distinct exchange patterns in the form of interaction protocols in order to participate in intelligent conversations.

Many multi-agent systems have defined and delivered solutions for encoding the content of messages in content languages. However, existing architectures have several shortcomings that still need to be improved to increase the deployment ratio of the MAS specifically because of non-availability of support for various profiles of semantic languages. The outline of this article is as follows; in section 2 we highlight various MAS and compare them with SAGE. Section 3 describes the communication framework for SAGE. This description is followed by the system architecture, content languages, and explanations of the FIPA-SL encoder and decoder in section 4. Section 5 then describes the functional capabilities of the MAS and proves them with corresponding examples. Section 6 compares the performance evaluation of the efficient encoder and decoder of SAGE to JADE. We summarize the paper with the conclusions and suggestions for potential work in section 7.