Comparison of JADE and SPADE for academic timetable generation using multi-agent systems

Abstract

Academic timetabling is a scheduling problem that involves efficiently allocating resources while satisfying multiple constraints. Traditionally, these problems have been addressed through centralized, heuristic, and metaheuristic approaches. Although these methods can produce feasible solutions, they often exhibit limitations in scalability and adaptability. In this context, multi-agent systems (MAS) have emerged as an effective alternative for solving scheduling problems, thanks to their distributed architecture, agent negotiation capabilities, and adaptability to dynamic environments. Various platforms are available for developing MAS, differing in programming languages, concurrency models, support for standards such as FIPA, usability, technical performance, and suitability for different deployment scenarios. This article presents a comparative study of two multi-agent platforms, JADE and SPADE, as applied to a scheduling problem. A case study at Universidad Arturo Prat involved developing a system based on the Contract Net Protocol to coordinate the assignment of instructors, courses, and classrooms. The platforms were evaluated using metrics that capture both the quality of the resulting timetables and the computational resource consumption. The results indicate that JADE offers greater stability and efficiency under high-load conditions, while SPADE displays advantages in less complex scenarios.