middleware development

This study addresses the lack of systematic comparative analysis among open-source message-oriented middleware systems, which hinders informed selection by developers. Through a comprehensive literature review and feature engineering, the authors conduct a structured evaluation of ten mainstream systems across 42 functional dimensions—encompassing 134 fine-grained attributes—including critical aspects such as transaction support, active messaging, and multi-tenancy. The work presents the first publicly available, extensively annotated dataset of message middleware features, offering fine-grained insights into their capabilities. This resource not only highlights the pivotal role these systems play in supporting cloud-native applications but also establishes a verifiable benchmark and actionable guidance for future optimization and community-driven development.

This work addresses the interoperability and runtime coordination challenges among heterogeneous digital twins in federated ecosystems, which arise from divergent modeling approaches and technology stacks. To overcome these issues, the paper proposes a federated node manager architecture that enables modular and deployable integration of digital twins through capability-controlled exposure, protocol and data schema adaptation, and real-time exchange of states and events. Built upon service-oriented and event-driven design principles, the proposed system demonstrates the feasibility of efficient collaboration among diverse digital twins in an intelligent transportation emergency response scenario. This implementation successfully translates conceptual federated architectures into a practical, operational technical pathway, marking the first realization of such an integrated solution in real-world settings.

Heterogeneous AI agents—diverse in architecture, vendor, and knowledge representation—lack a decentralized, scalable interoperability protocol. Method: This paper introduces MOD-X, a Modular, Open, and Decentralized agent eXchange framework. MOD-X features a semantic publish-subscribe message bus, a hierarchical protocol design enabling dynamic capability discovery and workflow orchestration, and a security verification mechanism integrating lightweight blockchain with smart contracts. It employs protocol-translation middleware, distributed state management, and seamless integration of multi-paradigm agents—including rule-based systems, neural networks, symbolic reasoners, and legacy system wrappers—without centralized coordination. Contribution/Results: Experimental evaluation demonstrates MOD-X’s superior scalability, security, and interoperability flexibility, enabling efficient cross-agent collaboration in decentralized settings. The framework advances the practical deployment of interoperable AI agent ecosystems.

This work proposes CODECO, a framework designed to address the challenges of traditional centralized Kubernetes in federated edge environments, where heterogeneous infrastructure, device mobility, and multi-provider collaboration are prevalent. CODECO enables edge autonomy while preserving global consistency through co-orchestration of data, computation, and networking. It integrates a semantic application model, a partitioned federation mechanism, AI-driven scheduling decisions, and a hybrid governance model. Built upon an extended Kubernetes architecture, CODECO supports context-aware microservice deployment and adaptive management. The framework’s efficacy in orchestrating applications across complex federated edge-cloud scenarios is validated through a reproducible experimental platform, demonstrating its capability to efficiently manage dynamic and heterogeneous edge environments.

Existing large-scale systems face challenges including poor cross-domain collaboration, weak dynamic adaptability, and inefficient human–machine interaction. To address these, this paper proposes a large language model (LLM)-enhanced hierarchical heterogeneous holon architecture for adaptive, human-centered systems of systems (SoS). Methodologically, it introduces four specialized holon types—supervisory, planning, task, and resource—organized within a three-layer structure (reasoning, communication, and capability), with LLMs deeply embedded in the reasoning layer for autonomous decision-making and real-time reconfiguration. The architecture is validated in a 3D urban traffic simulation, demonstrating scalability and sub-100-ms response latency. Furthermore, the work defines quantifiable metrics for efficiency and scalability, enabling both simulation-based optimization and practical deployment.

To address the challenge of decentralized multi-agent coordination in cloud-edge heterogeneous environments lacking centralized server support, this paper proposes a decentralized multi-agent collaboration framework. The framework introduces a novel logistics-object modeling approach and an abstract agent interface, enabling dynamic service-flow orchestration, topology-aware distributed publish-subscribe communication, and many-to-many service election. It integrates plug-and-play node discovery, flexible task reconfiguration, and fault-adaptive agent replacement. By synergistically combining multi-agent systems (MAS), fault-tolerant scheduling, and dynamic topology management, the framework achieves autonomous, real-time, and scalable decision-making coordination without a central coordinator. Experimental results demonstrate significant improvements in system resilience, real-time performance, and robustness for mission-critical autonomous scenarios, confirming its capability for highly available deployment.

Current 3GPP and O-RAN standards lack cross-device coordination abstractions for time-sensitive medical workflows, hindering their ability to meet the stringent requirements of emergency care scenarios. This work proposes the Collective Adaptive Intelligence Plane (CAIP), which enables workflow-level context binding, deadline-aware scheduling, semantic stream association, and data locality-based privacy preservation through lightweight coordination mechanisms compatible with existing RRC, QoS/SDAP, and O-RAN E2 interfaces—without introducing new protocol layers. CAIP represents the first integration of workflow coordination abstractions into the 6G healthcare networking standards framework, supports incremental deployment on 5G Advanced, and provides a standardized evolutionary path toward 6G. Its feasibility has been validated in an ICU collaboration scenario.

This work addresses the complexity, routing ambiguity, and unreliable shutdown commonly introduced by ad hoc glue code in existing modular distributed systems. To overcome these issues, the paper proposes CNS—a lightweight, local-first hybrid event bus that seamlessly bridges local and distributed publish-subscribe contexts through a unified event model and consistent routing semantics. CNS employs an asynchronous fire-and-forget primary path while supporting request-response extensions on the same topic. It integrates typed event keys, family-based serialization and validation, and NATS-backed distributed transport. Prototype evaluation demonstrates low-latency performance: approximately 30 microseconds for local delivery, 1.26–1.37 milliseconds for purely distributed communication, and 1.64–1.89 milliseconds for hybrid bridging, with validation overhead remaining manageable—making CNS suitable for structured inter-process communication and efficient messaging among resource-constrained nodes.

Cloud-native microservices lack lightweight, decentralized adaptive capabilities in dynamic environments. Method: This paper proposes an event-driven, rule-engine-based local autonomy framework that decouples monitoring, decision-making, and execution. It introduces a service-instrumentation workflow grounded in a simplified MAPE-K model—emphasizing Monitor and Execute phases—supports declarative adaptation actions, unifies metric collection, and employs a lightweight rule engine. Contribution/Results: The framework innovatively enables decentralized self-healing, self-protection, and self-optimization, demonstrating that localized decisions can collectively emerge into system-wide adaptability. Evaluated on an enhanced TeaStore benchmark, it achieves database auto-recovery, DDoS auto-mitigation, and traffic auto-optimization with minimal code modifications, significantly improving system resilience and architectural independence.

Current LLM-driven agents lack an engineering foundation for deployment in complex distributed systems, facing significant challenges in composability, interoperability, governance, and trustworthiness. This work proposes an engineering paradigm—Agentic Service-Oriented Computing (ASOC)—that introduces six core principles and a five-dimensional research agenda. By integrating service computing, agent orchestration, lifecycle management, and trustworthy design under constraints of security, compliance, and observability, ASOC enables credible governance of agent ecosystems. The framework provides both theoretical grounding and actionable engineering pathways to support reliable, accountable, and certifiable deployment of agent-driven service systems in enterprise and societal contexts.

Existing SFC simulation tools struggle to support flexible deployment and rapid validation of complex services in cloud-network environments. This paper proposes Mini-SFC, a lightweight, modular SFC simulation framework featuring a novel hybrid architecture that integrates numerical simulation with containerized virtual simulation. It enables runtime dynamic topology reconfiguration and RESTful online control. The framework provides standardized solver interfaces, minimalist module design, and open, extensible APIs—significantly lowering the barrier for algorithm validation. Experimental results demonstrate that Mini-SFC reduces SFC policy validation time by over 60% and enhances the generalizability of management algorithms across heterogeneous environments. The project is open-sourced and has been adopted by multiple universities for SFC optimization research.