TriSLA: A Preventive and Closed-Loop SLA-Aware Architecture for Multidomain Decision-Making with Explainable Artificial Intelligence in 5G Networks
TriSLA通过结合语义意图解析、多域机器学习风险推断、可解释AI特征归因及闭环运行时SLA保证，解决了5G网络中动态资源变化下的SLA保障问题。
0 citationsRead paper
Selected work
Representative Papers
An Event-Driven Tool for Context-Aware Code Smell Detection Using SmellDSL
This work addresses the high false-positive rates of traditional code smell detection tools, which often ignore development context and thus hinder effective refactoring. To overcome this limitation, the authors propose an event-driven detection approach that integrates contextual factors such as team composition, project phase, and geographical distribution. They introduce SmellDSL, a domain-specific language that unifies static code metrics with contextual rules, and implement a scalable architecture based on a service bus. The resulting system has been integrated into Eclipse and augmented with a mobile visualization interface, enabling precise identification of smell types, severity levels, and locations. This facilitates targeted refactoring by assigning appropriate developers, thereby significantly improving both detection accuracy and actionable insights for development teams.
0 citationsRead paper
Slice Agent: Identifying and Isolating Slices in Shared Open Radio Unit
This work addresses the inability of existing Open Radio Units (O-RUs) to identify and isolate uplink network slices, which hinders slice-aware service provisioning at the radio access network (RAN) layer. To overcome this limitation, the paper presents the first O-RU-side mechanism for uplink slice identification and isolation that operates without requiring scheduling information from the O-DU. The proposed approach embeds a slice agent, extends the eCPRI protocol, and introduces a time-sensitive, dedicated FPGA pipeline architecture to enable ultra-low-latency, slice-specific data encapsulation and priority handling in multipoint-to-multipoint (MP2MP) fronthaul scenarios. The design achieves per-packet processing in just two clock cycles and supports up to 3,822 slices per slot, significantly enhancing isolation accuracy and latency performance, thereby demonstrating its feasibility and efficiency for Beyond 5G/6G Open RAN deployments.
0 citationsRead paper
ORION: Intent-Aware Orchestration in Open RAN for SLA-Driven Network Management
This work addresses the lack of end-to-end automation in Open RAN for translating high-level service intents into low-level configurations, a gap exacerbated by existing orchestration approaches that rely on manual policies and struggle to meet SLA-driven autonomy requirements. To bridge this gap, the authors propose ORION, a novel framework that introduces, for the first time in O-RAN, a hierarchical multi-agent architecture powered by large language models (e.g., GPT-5, Gemini 3 Pro, Claude Opus). ORION integrates the Model Context Protocol with a Semantic Mediation Orchestrator (SMO) translation layer to automatically convert natural language intents into executable policies. Through coordinated rApp/xApp interactions between the Non-RT and Near-RT RICs, ORION enables full lifecycle management—from intent ingestion to E2 closed-loop execution. Experimental results demonstrate 100% policy generation success and a significant reduction in configuration complexity, laying a foundation for autonomous 6G networks.
0 citationsRead paper
Narrative Consolidation: Formulating a New Task for Unifying Multi-Perspective Accounts
Traditional multi-document summarization struggles with integrating multi-perspective narrative texts (e.g., legal testimonies, historical accounts) due to its overemphasis on concision, compromising temporal coherence and factual completeness. To address this, we formally introduce *Narrative Consolidation*—a novel task requiring precise chronological ordering, comprehensive content coverage, and seamless integration of complementary details. We propose the Temporal Alignment Event Graph (TAEG), a graph-based model that explicitly unifies event alignment with temporal structure modeling. TAEG incorporates graph centrality measures (e.g., PageRank) to automatically select authoritative narrative versions. Evaluated on the Four Gospels dataset, our method achieves perfect temporal consistency (Kendall’s Tau = 1.000) and improves ROUGE-L F1 by 357.2% over baselines, demonstrating that explicit temporal backbone modeling is essential for effective narrative consolidation.
0 citationsRead paper
Recent publications
Latest Papers
TriSLA: A Preventive and Closed-Loop SLA-Aware Architecture for Multidomain Decision-Making with Explainable Artificial Intelligence in 5G Networks
TriSLA通过结合语义意图解析、多域机器学习风险推断、可解释AI特征归因及闭环运行时SLA保证，解决了5G网络中动态资源变化下的SLA保障问题。
0 citationsRead paper
An Event-Driven Tool for Context-Aware Code Smell Detection Using SmellDSL
This work addresses the high false-positive rates of traditional code smell detection tools, which often ignore development context and thus hinder effective refactoring. To overcome this limitation, the authors propose an event-driven detection approach that integrates contextual factors such as team composition, project phase, and geographical distribution. They introduce SmellDSL, a domain-specific language that unifies static code metrics with contextual rules, and implement a scalable architecture based on a service bus. The resulting system has been integrated into Eclipse and augmented with a mobile visualization interface, enabling precise identification of smell types, severity levels, and locations. This facilitates targeted refactoring by assigning appropriate developers, thereby significantly improving both detection accuracy and actionable insights for development teams.
0 citationsRead paper
Slice Agent: Identifying and Isolating Slices in Shared Open Radio Unit
This work addresses the inability of existing Open Radio Units (O-RUs) to identify and isolate uplink network slices, which hinders slice-aware service provisioning at the radio access network (RAN) layer. To overcome this limitation, the paper presents the first O-RU-side mechanism for uplink slice identification and isolation that operates without requiring scheduling information from the O-DU. The proposed approach embeds a slice agent, extends the eCPRI protocol, and introduces a time-sensitive, dedicated FPGA pipeline architecture to enable ultra-low-latency, slice-specific data encapsulation and priority handling in multipoint-to-multipoint (MP2MP) fronthaul scenarios. The design achieves per-packet processing in just two clock cycles and supports up to 3,822 slices per slot, significantly enhancing isolation accuracy and latency performance, thereby demonstrating its feasibility and efficiency for Beyond 5G/6G Open RAN deployments.
0 citationsRead paper
ORION: Intent-Aware Orchestration in Open RAN for SLA-Driven Network Management
This work addresses the lack of end-to-end automation in Open RAN for translating high-level service intents into low-level configurations, a gap exacerbated by existing orchestration approaches that rely on manual policies and struggle to meet SLA-driven autonomy requirements. To bridge this gap, the authors propose ORION, a novel framework that introduces, for the first time in O-RAN, a hierarchical multi-agent architecture powered by large language models (e.g., GPT-5, Gemini 3 Pro, Claude Opus). ORION integrates the Model Context Protocol with a Semantic Mediation Orchestrator (SMO) translation layer to automatically convert natural language intents into executable policies. Through coordinated rApp/xApp interactions between the Non-RT and Near-RT RICs, ORION enables full lifecycle management—from intent ingestion to E2 closed-loop execution. Experimental results demonstrate 100% policy generation success and a significant reduction in configuration complexity, laying a foundation for autonomous 6G networks.
0 citationsRead paper
Narrative Consolidation: Formulating a New Task for Unifying Multi-Perspective Accounts
Traditional multi-document summarization struggles with integrating multi-perspective narrative texts (e.g., legal testimonies, historical accounts) due to its overemphasis on concision, compromising temporal coherence and factual completeness. To address this, we formally introduce *Narrative Consolidation*—a novel task requiring precise chronological ordering, comprehensive content coverage, and seamless integration of complementary details. We propose the Temporal Alignment Event Graph (TAEG), a graph-based model that explicitly unifies event alignment with temporal structure modeling. TAEG incorporates graph centrality measures (e.g., PageRank) to automatically select authoritative narrative versions. Evaluated on the Four Gospels dataset, our method achieves perfect temporal consistency (Kendall’s Tau = 1.000) and improves ROUGE-L F1 by 357.2% over baselines, demonstrating that explicit temporal backbone modeling is essential for effective narrative consolidation.
0 citationsRead paper