Experimentation of Negotiation Protocols for Consensus Problems in Smart Parking Systems

摘要

A smart city uses emergent technologies for improving the services that will contribute to make the citizens' daily life more comfortable and convenient. Among several strand offered by a smart city, the smart parking systems focus the transportation and parking of vehicles problems, providing intelligent solutions based on ICT technologies, and particularly artificial intelligence techniques. In this context, a cyber-physical system, based on multi-agent systems, was developed for an intelligent parking system for car and bicycles. This multi-agent based system consists of a community of distributed, intelligent and autonomous agents, representing the parking spots and drivers, which cooperate to reach their objectives. In such systems, the global system behaviour emerges from the interaction between these individual entities, being crucial the adoption of the proper cooperation protocols. This paper studies and compares possible approaches to solve consensus problems in such distributed smart parking systems, and particularly addressing the negotiation strategies. For this purpose, the Contract Net Protocol, the Fnglish auction and the Dutch auction negotiation strategies were implemented in an agent-based smart parking system using the JADE framework. The experimental results allowed to perform a comparative analysis, considering the satisfaction levels of the actors, the scalability and the negotiation time to decide which approach better fits with the smart parking problem.