Transient, deadline-aware RAN slicing for NR-V2X using finite-capacity time-varying slice queues
Vehicular Communications, cilt.61, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 61
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.vehcom.2026.101064
- Dergi Adı: Vehicular Communications
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, INSPEC
- Anahtar Kelimeler: Deadline scheduling, eMBB, Network slicing, NR-V2X, Transient queues, URLLC
- Atatürk Üniversitesi Adresli: Evet
Özet
Network slicing is a key mechanism for supporting heterogeneous vehicular services in 5G New Radio vehicle-to-everything (NR-V2X) systems. These services include latency-critical ultra-reliable low-latency communication (URLLC), throughput-oriented enhanced mobile broadband (eMBB), and event-driven vehicular safety messaging. In vehicular networks, however, traffic demand, channel quality, and slice resource needs can change rapidly because of mobility, fading, and bursty safety events. Therefore, slicing methods based mainly on stationary queueing assumptions or average-delay metrics may miss short congestion periods that affect URLLC tail latency, i.e., the delay experienced by the worst-served packets rather than the average packet delay. This paper proposes a transient, deadline-aware radio access network (RAN) slicing framework for NR-V2X systems. The framework models URLLC, eMBB, and a logical Safety traffic class as finite-capacity time-varying queues, where the physical resource block (PRB) allocation of each slice directly determines its effective packet-level service parallelism. The proposed model combines mobility-driven arrivals, fading-dependent service variation, deadline-aware queue handling, predictive deadline dropping, and a tail-guided PRB controller. The controller adapts slice capacity using sliding-window URLLC p99 delay, defined as the delay value below which 99% of completed URLLC packets leave the serving next-generation NodeB (gNB) queue, together with queue congestion and short-horizon backlog pressure. Safety traffic is further evaluated through a separate direct sidelink interface (PC5) and hybrid automatic repeat request (HARQ) delivery stage, so that gNB queueing effects and sidelink delivery effects can be reported separately. Simulation results show that the proposed framework stabilizes URLLC tail latency under representative NR-V2X conditions. In the representative scenario, URLLC p99 delay remains between 0.85 and 0.92 ms across the two gNBs, while the finite-horizon URLLC timely-completion ratio reaches 99.37% with 0.63% effective queue loss. The eMBB slice sustains approximately 12–15 Mb/s under nominal operation with no observed queue-level loss, and the PC5/HARQ delivery performance of the Safety traffic class is reported separately. Comparisons with static, adaptive, priority-based, and MaxWeight baselines show that transient queue modeling, predictive deadline dropping, and tail-guided PRB adaptation jointly improve URLLC tail-latency control while preserving broadband service. The results show that transient, deadline-aware RAN slicing can protect URLLC tail performance under time-varying NR-V2X conditions while maintaining broadband service.