OptiMA: A Transaction-Based Framework with Throughput Optimization for Very Complex Multi-Agent Systems

TL;DR

OptiMA presents a transaction-based framework to address fault tolerance and throughput bottlenecks in very complex multi-agent systems. It integrates locking-based concurrency control and TxnSP-driven scheduling to enable safe, scalable execution across hundreds of agents and plugins, while formalizing the TxnSP problem and providing a dedicated software library with ES, MIP, DP, and SA solvers. The Factory Floor Benchmark demonstrates tangible throughput improvements and confirms the practical viability of transaction scheduling under contention, with safety and liveness guarantees. The work offers solid theoretical foundations and open-source artifacts to drive future research on transaction-based MAS design and scheduling in complex, resource-constrained environments.

Abstract

In recent years, the research of multi-agent systems has taken a direction to explore larger and more complex models to fulfill sophisticated tasks. We point out two possible pitfalls that might be caused by increasing complexity; susceptibilities to faults, and performance bottlenecks. To prevent the former threat, we propose a transaction-based framework to design very complex multi-agent systems (VCMAS). To address the second threat, we offer to integrate transaction scheduling into the proposed framework. We implemented both of these ideas to develop the OptiMA framework and show that it is able to facilitate the execution of VCMAS with more than a hundred agents. We also demonstrate the effect of transaction scheduling on such a system by showing improvements up to more than 16\%. Furthermore, we also performed a theoretical analysis on the transaction scheduling problem and provided practical tools that can be used for future research on it.

Figures (8)

• Figure 1: Structure of the Framework: The transaction factory creates transactions that includes agent actions. The scheduler batches the transactions and schedules each batch. The executor runs the scheduled transactions and sends their result back to the transaction factory, which then creates new transactions depending on these results. During the execution of a transactions, the plugin manager, agent manager and postmaster checks the consistency constraints and manipulates the system if an authorized action requests it.
• Figure 2: Lock acquisition and releasing mechanisms of Two-Phase Locking and its variant
• Figure 3: An example transaction schedule
• Figure 4: An Example Case for Lemma \ref{['lem:Residual']}
• Figure 5: Two equivalent subschedules

Theorems & Definitions (27)

• theorem 1
• definition 1: Conflict Parity
• theorem 2
• theorem 3
• definition 2: Subschedule
• definition 3: Makespan and Minimum Time
• definition 4: Derived Schedule
• definition 5: Derivation Plan
• definition 6: Empty Subschedule
• definition 7: Residual Subschedule