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

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.