A Cloud-Based Distributed System for Scalable Multi-Agent Simulation of Pig Health Monitoring Models

Pig farming is a cornerstone of global food security but faces severe threats from rapidly spreading infectious diseases such as African Swine Fever (ASF). Conventional single-machine GAMA simulations suffer from severe computational bottlenecks when modeling multi-agent systems for pig health monitoring. This paper proposes a novel cloud-based distributed simulation architecture that integrates the GAMA platform (in headless mode) with a Redis Pub/Sub messaging mechanism, enabling efficient parallel execution across multiple virtual nodes. The system incorporates a refined SEIR epidemiological model for disease transmission dynamics together with precision feeding and greenhouse-gas emission monitoring. Experiments conducted on OpenStack and AWS clouds demonstrate that the proposed architecture supports up to 32 concurrent simulations (each modeling one pigpen with 20 pigs, totaling 640 pigs per run). This study presents a proof-of-concept for scalable simulation, while the architecture is designed to scale to hundreds or thousands of pens by adding more worker nodes. Compared with traditional sequential desktop execution, the system reduces completion time by 45–68% (OpenStack) and up to 55% (AWS) while achieving high scalability. CPU utilization remains below 98% on AWS even at high concurrency, and RAM consumption scales efficiently with workload. Detailed performance metrics—including execution time, CPU utilization, and RAM consumption—were quantified for 15-day and 30-day scenarios. Results confirm that OpenStack delivers superior speed under moderate loads, whereas AWS provides greater stability at higher concurrency. The Redis Pub/Sub-driven master-worker design effectively overcomes the scalability limitations of existing GAMA-based systems for pig health monitoring.