latency-aware resource allocation

Deficit Round Robin (DRR) schedulers in delay-sensitive networks suffer from difficulty in configuring integral quantum parameters and lack theoretical guarantees on end-to-end delay. Method: This paper first proves the convexity of the end-to-end delay feasibility region for two-flow DRR, revealing structural regularities of feasible quantum sets in n-flow systems; leveraging convex analysis and constrained optimization, it establishes an analytically tractable and deployable framework for joint quantum optimization under strict end-to-end delay constraints. Contribution/Results: The proposed method achieves a 37% increase in packets served per round while guaranteeing delay bounds, advancing DRR parameter design from heuristic tuning to provably optimal configuration. It provides both theoretical foundations and practical tools for QoS assurance in network slicing.

This work addresses the limitations of existing 5G radio access network (RAN) slicing approaches, which prioritize data rate alone and fail to meet the stringent latency requirements of deterministic aperiodic traffic in smart manufacturing. To overcome this, the paper proposes a novel RAN slicing design that jointly considers both rate and latency, explicitly incorporating delay constraints into the core resource allocation framework—thereby moving beyond conventional bandwidth-only partitioning. By introducing a unified rate–latency service descriptor and optimizing wireless resource scheduling accordingly, the proposed scheme effectively guarantees deterministic aperiodic communication for critical services. Experimental results demonstrate its significant superiority over traditional bandwidth-centric slicing mechanisms in latency-sensitive scenarios.

To address the challenge of deterministic low-latency scheduling for periodic message transmission between antennas and remote processing units in Cloud-RAN—where strict protocol deadlines must be met while avoiding buffering and collision delays induced by statistical multiplexing—this paper pioneers the application of deterministic conflict-free scheduling to Cloud-RAN fronthaul networks. We propose two algorithms: (i) an analytical zero-buffer scheduling algorithm tailored for short-path or light-load scenarios, and (ii) PMLS (Periodic Message Latency Scheduling), a heuristic algorithm supporting bounded buffering. We theoretically prove that a zero-buffer feasible schedule always exists under short-path or low-load conditions. Experimental results demonstrate that PMLS achieves zero-delay deterministic schedules with high probability even under full load, significantly enhancing latency predictability and resource utilization.

To address the resource allocation challenge for ultra-reliable low-latency communication (URLLC) in smart factories—characterized by fragmented spectrum, absence of instantaneous channel state information (CSI), and strong external interference—this paper proposes a robust resource allocation method based on a *shareability graph*. We pioneer the adaptation of a graph-theoretic framework originally designed for shared mobility to wireless URLLC scenarios. Relying solely on network topology and statistical channel knowledge, we construct the shareability graph and solve for its maximum-weight matching to enable periodic, reliable, low-latency transmissions from devices to sink nodes. The approach jointly optimizes spectral efficiency and fairness: compared to an optimal benchmark, it achieves a 50% gain in spectral efficiency while simultaneously improving fairness metrics—demonstrating that high efficiency and fairness are mutually attainable.

To address the joint guarantee of packet-level reliability and timeliness for ultra-reliable low-latency communication (URLLC) in industrial multi-cell, multi-channel wireless networks, this paper proposes a CNN-driven dynamic link priority prediction framework, replacing conventional static link-dependent priority (LDP) scheduling. Our method innovatively integrates convolutional neural networks with graph coloring to enable adaptive interference coordination based on real-time network state, traffic characteristics, and channel opportunities. We further introduce a novel offline-training–online-lightweight-inference architecture, enabling millisecond-scale priority reconfiguration. Experiments across three representative industrial network configurations demonstrate SINR improvements of 113%, 94%, and 49%, respectively. The framework significantly enhances resource utilization, schedulability, and probabilistic real-time guarantees—overcoming the performance limitations of static scheduling in highly dynamic industrial environments.

This work proposes a 5G radio access network (RAN) slicing scheme tailored for low-latency-sensitive services to address the heterogeneous requirements of diverse applications in Industrial 4.0 and similar scenarios, which demand varying combinations of data rates and stringent latency constraints—including deterministic periodic, aperiodic, and non-deterministic traffic. The proposed approach uniquely integrates the optimization of slice architecture and wireless resource allocation across multiple classes of latency-sensitive traffic, enabling customized quality-of-service (QoS) guarantees over a shared physical infrastructure. Leveraging network virtualization and softwarization, the scheme employs dynamic resource partitioning, slice-specific configuration, and QoS-driven scheduling policies to effectively meet the demanding and diverse latency and throughput requirements in representative industrial environments, demonstrating strong potential for extension to other latency-critical vertical industries.

This work addresses the challenge that conventional configured grant (CG) scheduling struggles to meet the bounded latency requirements of deterministic communication under variable traffic conditions. To overcome this limitation, the paper proposes a novel CG scheduling mechanism that integrates traffic prediction with robust optimization, explicitly incorporating prediction uncertainty into resource pre-allocation decisions for the first time. The approach dynamically adapts to the heterogeneous latency constraints of mixed traffic types while ensuring bounded end-to-end delays. By jointly optimizing resource allocation under uncertainty, the method significantly improves resource utilization without compromising timing guarantees. Extensive evaluations demonstrate that the proposed scheme maintains superior performance even in highly dynamic and diverse traffic scenarios, thereby substantially enhancing the system’s capability to support deterministic services.

本文针对5G网络中V2X服务的严苛要求，采用基于PPO的强化学习方法优化RAN切片资源分配，以满足低延迟高可靠需求并提高资源利用效率。

This work addresses the challenge of schedulability analysis for heterogeneous periodic traffic in ultra-reliable low-latency communication (URLLC) systems, where proactive HARQ introduces slot-level timing effects—such as feedback delay—that complicate resource allocation. To tackle this, the paper proposes a discrete-time Markov chain (DTMC)-based modeling framework that accurately captures cross-slot dynamics, including HARQ round-trip latency, through an expanded state space. Coupled with a two-stage genetic algorithm, the approach optimizes offset scheduling to meet diverse reliability and latency requirements. This study is the first to apply DTMCs to timing modeling of periodic flows under proactive HARQ, enabling precise schedulability analysis that explicitly accounts for feedback delay. Simulations demonstrate that the proposed method significantly improves schedulability compared to reactive HARQ, K-repetition schemes, and non-guaranteed proactive HARQ, while maintaining manageable computational overhead.

This work addresses the inherent lack of deterministic communication support in native 5G networks, which struggles to meet the stringent bounded latency and high reliability requirements of Time-Sensitive Networking (TSN) in industrial scenarios. The paper proposes a deeply integrated 5G-TSN configuration grant scheduling scheme that achieves, for the first time, cross-domain joint scheduling. By leveraging key characteristics of TSN traffic—such as periodicity, packet size, and arrival timing—the authors design a deterministic resource allocation algorithm. This approach overcomes the capacity and flexibility limitations of existing solutions, significantly enhancing the network’s ability to support heterogeneous TSN flows with diverse periods while simultaneously guaranteeing end-to-end latency bounds and improving resource utilization efficiency.