cross-layer coordination

This paper addresses the absence of a standardized taxonomy for hierarchical multi-agent systems (HMAS) and the lack of clarity in trade-offs among coordination mechanisms. To this end, we propose the first multidimensional classification framework integrating structural, temporal, and communication dimensions—comprising five axes: control hierarchy, information flow, task allocation, temporal layering, and communication topology. The framework systematically unifies classical coordination paradigms (e.g., Contract Net), hierarchical reinforcement learning, and large language model–based agents, thereby exposing novel challenges in interpretability, scalability, and secure integration. Empirical validation in industrial domains—including power grid management and oilfield operations—demonstrates that the framework effectively guides HMAS design, enhancing global efficiency while preserving local autonomy. It exhibits strong cross-domain adaptability and practical utility for real-world system engineering.

This study addresses the severe fragmentation of large language model (LLM) communication protocols in multi-agent systems, which significantly hinders interoperability. The work proposes the first structured taxonomy specifically tailored to LLM agent communication protocols, developed through an empirical-conceptual bidirectional iterative approach. By systematically analyzing nine prominent open-source protocols, the authors derive a five-dimensional classification framework encompassing communication parties, payload structure, interaction state, discovery mechanisms, and pattern flexibility. The analysis identifies recurring architectural patterns—including hybrid payloads, persistent conversation states, and runtime protocol negotiation—revealing commonalities and evolutionary trends across existing protocols. The study further forecasts a trajectory toward federated, layered protocol stacks and highlights critical research gaps, particularly in privacy preservation and policy enforcement, thereby offering theoretical guidance for future protocol design and selection.

Existing agent programming models struggle to coordinate internal decision-making with external interaction behaviors and lack effective abstractions for protocol adherence. This work proposes Kiko, a protocol-based agent programming model that introduces a decider abstraction, enabling agents to select actions and generate protocol-compliant messages exclusively from the set of legally permissible moves, thereby strictly adhering to interaction protocols in decentralized environments. Kiko fully encapsulates low-level communication details, supports flexible decision-making strategies, and integrates formal operational semantics to guarantee correct execution of arbitrary protocols. By doing so, it allows developers to focus on business logic while ensuring correctness and reliability in protocol enforcement within multi-agent systems.

This work addresses the challenges of low coordination efficiency, lack of auditability, and insufficient policy compliance in multi-agent systems for complex tasks by proposing a unified orchestration framework that integrates planning, policy execution, state management, and observability mechanisms. The framework introduces two novel protocols—the Model Context Protocol and Agent2Agent Protocol—to enable standardized, extensible communication among agents and between agents and external tools. By supporting policy governance and end-to-end traceability, the proposed architecture provides a practical, transparent, and accountable blueprint for deploying large-scale multi-agent systems within enterprise AI ecosystems.

This study addresses the challenge of coordinating heterogeneous agents and selecting appropriate communication protocols in large language model–driven multi-agent systems. Drawing from software engineering practice, it empirically evaluates the coordination capabilities of the Model Context Protocol (MCP) and Agent2Agent (A2A) across critical scenarios including agent discovery, multi-turn dialogue, and asynchronous communication. The work reveals that MCP offers lightweight efficiency but requires explicit state management, making it suitable for low-complexity tasks, whereas A2A natively supports stateful coordination—better suited for strongly state-dependent contexts—at the cost of higher implementation complexity. These findings provide actionable empirical guidance for multi-agent protocol design and clarify key trade-offs between the two approaches regarding responsibility distribution, interoperability, and access control.

Existing agent interoperability protocols primarily focus on task coordination and lack support for governance-constrained collective decision-making in multi-agent communities. Drawing on organizational theory and corporate governance standards, this work proposes a six-dimensional governance framework encompassing membership management, deliberation, voting, dissent reservation, human escalation, and audit replay. The study systematically evaluates five prominent protocols—MCP, A2A, ACP, and others—and reveals, for the first time, that the core deficiency lies not in insufficient protocol-level features but in the absence of a dedicated architectural layer for governance. It further distinguishes between scalability gaps and structural gaps. The analysis demonstrates that current protocols universally lack voting and dissent reservation mechanisms, provide only partial support for deliberation, and offer no complete set of governance primitives.

Conventional network protocol stacks (e.g., TCP/IP) are designed for data transmission and lack native support for agent-level semantic understanding and dynamic context sharing, hindering effective distributed multi-agent collaboration. Method: This paper proposes a novel internet protocol architecture tailored for agent collaboration, extending the traditional seven-layer model with two new layers: an eighth Agent Communication Layer and a ninth Semantic Negotiation Layer. It formally defines agent-to-agent communication structures and dynamic context negotiation mechanisms. Leveraging the MCP tool-calling protocol, the architecture integrates message envelopes, speech-act theory, interaction patterns, and formal schemas to construct a semantically aware communication stack. Contribution/Results: Experimental evaluation demonstrates significant improvements in collaborative efficiency and scalability for multi-agent systems executing complex tasks. The architecture provides foundational protocol support for distributed intelligent collaboration—particularly valuable in large-language-model-constrained environments.

This work addresses the limitations of existing AI agent protocols—such as MCP and A2A—which support only single-agent control and struggle to coordinate multi-agent collaboration under shared state, often resorting to unstructured chat or silent overwrites. To overcome this, the paper introduces MPAC, the first standardized coordination protocol for multi-agent collaboration. MPAC structures coordination through a five-layer architecture encompassing conversation, intent, action, conflict, and governance, treating conflict as a first-class object. It enables intent pre-declaration, structured conflict representation, and pluggable human arbitration. The protocol specifies 21 message types, three state machines, Lamport-clock-based causal watermarking, optimistic concurrency control, and multiple security configurations, with reference implementations in Python and TypeScript. Evaluated on a three-agent code review benchmark, MPAC reduces coordination overhead by 95% and achieves a 4.8× end-to-end speedup over a human-sequential baseline, without increasing per-agent decision latency.

Current agent communication protocols generally lack mechanisms for semantic alignment, clarification, and verification, shifting semantic responsibility onto prompts or application logic and thereby causing poor interoperability and high maintenance costs. This work proposes, for the first time, a human-inspired three-layer communication framework—comprising communication, syntactic, and semantic layers—and systematically analyzes 18 mainstream protocols to expose their structural deficiencies in semantic coordination. Through layered modeling, technical debt identification, and scenario mapping, the study not only derives a practical protocol selection guide but also advances agent communication beyond mere message passing toward a new paradigm of shared understanding, laying the foundation for building semantically robust, secure, and interoperable agent ecosystems.

Existing agent coordination benchmarks struggle to evaluate the efficacy of industrial scheduling paradigms under hierarchical and dynamically coupled constraints. This work proposes DESBench—a Distributed Event-driven Scheduling Benchmark—that establishes the first systematic evaluation framework tailored for industrial coordination paradigms. Built upon a discrete-event simulation environment, DESBench integrates multi-agent systems, hierarchical decision modeling, and multidimensional performance metrics to comparatively analyze four coordination mechanisms—centralized, hierarchical, heterogeneous, and holographic—under conditions of partial observability, multiple timescales, and dynamic constraints. Experimental results reveal fundamental trade-offs: centralized approaches exhibit robustness and efficiency yet poor scalability; hierarchical methods improve efficiency but suffer from inter-level misalignment; heterogeneous schemes offer flexibility at the cost of high communication overhead; and holographic mechanisms satisfy constraints effectively but lack global robustness, thereby elucidating how coordination design fundamentally shapes system behavior.

This work addresses the challenges of coordination in large-scale autonomous agent systems—particularly in collaboration, value exchange, security, and governance—by proposing a graph-centric coordination layer that unifies heterogeneous entities such as humans, agents, tools, and organizations. The architecture supports event-driven multi-party collaboration and AI-driven economic activity, uniquely embedding policy enforcement, provenance tracking, and auditability as first-class primitives within the protocol design. Leveraging a graph-first approach, economic primitives (including metering, receipts, and settlement), and cross-protocol bridging mechanisms, the system enables incremental deployment while maintaining compatibility with existing infrastructures. This foundation balances composability with strong accountability, substantially reducing integration and governance overhead, and thereby providing an open, pluralistic, and governable substrate for large-scale human–machine societies.

This work addresses the tight coupling among organizational structure, coordination mechanisms, and collaboration algorithms in multi-agent large language model systems, which hinders independent configuration and evaluation. To resolve this, the paper introduces IMACS, a novel framework that formalizes classical organizational theories—specifically Belbin team roles, Mintzberg’s coordination mechanisms, and the RACI responsibility model—into configurable components, thereby achieving orthogonal decoupling across three layers: team composition, agent alignment, and collaboration algorithms. The framework unifies six collaboration protocols under a common interface and incorporates a context-aware multi-armed bandit-based adaptive routing mechanism for dynamic, task-driven protocol selection. Experimental results demonstrate that adaptive routing significantly outperforms fixed protocols, with optimal configurations varying across model families, thus validating both the necessity of decoupled design and the efficacy of online learning.