Desenvolvimento de uma arquitetura multiagente holônica reconfigurável aplicada ao cenário de smart parking

Abstract

The characteristics in a smart parking indicate the usage of Multiagent Systems as a trend due to the dynamism of this environment. In the MAS field, the usage of holonic agents (Holonic Multiagent Systems - HMAS) provides to the system a high level of adaptation towards the changes that take place in the environment according to the arrangement of agents. The environment might suffer some changes due to some agent failure and also different organizational settings. Therefore, an architecture designed to assign parking spaces in a smart parking is expected to deal with these changes so that the process of requisition and usage of parking spaces is not affected by the drivers. Thus the present work shows the development of a reconfigurable holonic multiagent architecture applied to the intelligent parking lot called MAPS-HOLO (MultiAgent Parking System - Holonic). We say that our architecture is reconfigurable because it may adapt its structure depending on some flaws from its agents. The MAPS-HOLO architecture will have its structure reconfigured in terms of how the agents are organized and how the HMAS government works, it may vary from a centralized (head-body) to a decentralized (body-body) organization. Besides, the MAPS-HOLO runs in JaCaMo, which is a framework to the development of MAS subdivided into three layers: Jason language (agent programming), Cartago Framework (artifacts and environment programming) and Moise (normative programming of agents). Although the focus of the architecture is parking spaces, the MAPS-HOLO could be extended to work with any allocable resource, as long as it has a given cost C and a time T, for example, a computational resource (grid), bike spaces, task allocation, among others. Our major goal is to develop a reconfigurable holonic multiagent architecture in order to assign parking spaces to drivers. Furthermore, we highlight some additional outcomes that we are looking forward to obtaining: i) Development of recovery plans for agent failures; ii) Application of JaCaMo framework in a holonic system; iii) Analysis and comparison of different government policies in the HMAS. After the development of MAPS-HOLO , we can say the HMAS deal with the parking space assigning process. Further, the architecture handles with the agent’s failures whereas MAPS-HOLO got in average higher results compared to agent cloning provided by JaCaMo. Finally, it is possible to develop an HMAS through JaCaMo. However, some improvements on Jason and Moise are required to a full HMAS development support.