What question did this study set out to answer?

The aim is to establish a minimalistic architecture that maintains homeostasis in distributed multi-agent systems.

April 4, 2026Open Access

AAM-V1.1: Minimal Coherent Multi-Agent Architecture with EMA Filtering and Context-Aware Meta-Policy Layer

Key Points

The aim is to establish a minimalistic architecture that maintains homeostasis in distributed multi-agent systems.
Developed a deterministic architecture with no centralized control.
Implemented signal filtering using Exponential Moving Average (EMA).
Utilized adaptive constraint processing for behavioral detection.
Employed event-driven synchronization through Belov Fiber.
Introduced context-aware meta-policy regulation.
Showed robustness in noisy environments and asynchronous execution.
Maintained coherent multi-agent behavior with minimal computational overhead.
Performed effectively under delayed communication and repeated errors.

Abstract

AAM-V1.1 is a minimal deterministic architecture for achieving homeostasis in distributed multi-agent systems. The architecture replaces centralized orchestration with layered behavioral coherence, combining: - Signal filtering (Exponential Moving Average, EMA)- Behavioral detection (Adaptive Constraint Processor, ACP)- Event-driven synchronization (Belov Fiber)- Experience-based regulation (Context-aware Meta-Policy Layer) The system demonstrates robustness under:- noisy environments,- asynchronous execution,- delayed communication,- repeated error conditions. AAM-V1.1 introduces a shift from orchestration-based AI systems to regulation-based distributed cognition, enabling coherent multi-agent behavior with minimal computational overhead. The architecture is fully deterministic, explainable, and suitable for safety-critical and resource-constrained environments.

AAM-V1.1: Minimal Coherent Multi-Agent Architecture with EMA Filtering and Context-Aware Meta-Policy Layer

Key Points

Abstract

Cite This Study