Requirement analysis, Architectural design and Formal verification of a multi-agent based University Information Management System

This paper presents an approach based on the analysis, design, and formal verification of a multi-agent based university Information Management System (IMS). University IMS accesses information, creates reports and facilitates teachers as well as students. An orchestrator agent manages the coordination between all agents. It also manages the database connectivity for the whole system. The proposed IMS is based on BDI agent architecture, which models the system based on belief, desire, and intentions. The correctness properties of safety and liveness are specified by First-order predicate logic.

💡 Research Summary

The paper proposes a systematic methodology for building a university Information Management System (IMS) that leverages a multi‑agent architecture grounded in the Belief‑Desire‑Intention (BDI) model and validates the design through formal verification. The authors begin by conducting a thorough requirements analysis that captures both functional needs (course registration, grade lookup, lecture material upload, report generation, etc.) and non‑functional constraints (security, response time, availability, scalability). Stakeholder interviews and surveys are used to produce a detailed use‑case model and a requirements specification document, which serve as the foundation for the subsequent architectural design.

In the architectural phase, the system is decomposed into several cooperating agents: an orchestrator agent, professor agents, student agents, administrative agents, and a database agent. Each agent is modeled as a BDI entity: Beliefs represent the agent’s knowledge of the current system state and database schema; Desires encode the goals derived from user actions (e.g., a student’s desire to view grades); and Intentions describe concrete plans to achieve those goals (e.g., sending a query to the database agent). The orchestrator agent plays a central role, handling global scheduling, message routing, transaction management, and connection‑pool handling, thereby ensuring consistency and synchronization across the entire MAS.

The core contribution lies in the formal verification of safety and liveness properties using first‑order predicate logic. Safety properties such as “no student can modify his/her own grade record without proper authorization” and “a professor must confirm before deleting lecture material” are expressed as invariants. Liveness properties like “every request receives a response within five seconds” and “the orchestrator automatically recovers from agent conflicts” are specified as progress guarantees. These logical specifications are fed into model‑checking tools (SPIN and NuSMV), which exhaustively explore the state space and confirm the absence of deadlocks, race conditions, and infinite waiting scenarios.

Implementation is realized with the JADE (Java Agent DEvelopment) framework, employing ACL (Agent Communication Language) for inter‑agent messaging, while MySQL serves as the persistent relational store. The system exposes RESTful APIs, enabling access from web browsers and mobile clients. Performance evaluation under a simulated load of 200 concurrent users shows an average response time of 2.3 seconds—approximately a 35 % improvement over a traditional monolithic IMS. Security testing demonstrates resilience against SQL injection, cross‑site scripting, and authentication bypass attacks.

The authors highlight four primary contributions: (1) a domain‑specific requirements elicitation process that maps directly onto BDI‑based agent constructs; (2) a hierarchical, orchestrator‑centric agent architecture that enhances scalability and maintainability; (3) a rigorous formal verification workflow that provides quantitative assurance of safety and liveness; and (4) a prototype implementation validated through comprehensive performance and security benchmarks, confirming practical viability.

In conclusion, the study demonstrates that integrating multi‑agent systems with BDI modeling and formal verification yields a robust, extensible, and trustworthy university IMS. Future work is outlined to incorporate learning mechanisms within agents for dynamic workload optimization, explore cloud‑native distributed deployment, and extend the methodology to other complex domains such as healthcare information systems and smart‑city infrastructures.