PaperlayerView exact revision on arXiv

Source-linked AI summary

Det-5G: Closing the Determinism Gap in 5G-Advanced for Industrial Closed-Loop Control

Adnan Aijaz

arXiv:2609.07386v1cs.NI

TL;DR

Closed-loop control needs predictable completion of command/feedback cycles, not just low packet latency and high reliability. Det-5G schedules each complete cycle and combines adaptive bundled transmissions with group-oriented downlink and optimized multi-user uplink scheduling. At 16 devices, NSBT has a lower 99th-percentile cycle time than dynamic scheduling and SPS/CG.

  • Problem

    Low packet latency and high reliability alone do not ensure predictable completion of causally linked command/feedback cycles under changing conditions.

    Source: §I · p. 1

  • Method

    Det-5G jointly allocates resources for complete control cycles, using adaptive bundled transmissions, group downlink delivery, and multi-user uplink optimization.

    Source: §I · p. 1

  • Results

    At 16 devices, NSBT's 99th-percentile cycle time is 3.43 ms versus 5.25 ms for dynamic scheduling and 4.75 ms for SPS/CG; against fixed K=5, average cycle time falls about 34%.

    Source: A. Joint cycle time and deadline… · p. 5, C. Multi-user cyclic scalability · p. 6

  • Takeaways & Limitations

    Reserved Det-5G changes reliability resources only when link state persistently changes, while the allocated cycle remains predetermined.

    Source: E. Mobility and bundle update overhead · p. 7

Abstract

from arXiv · show

The ultra-reliable low-latency communication (uRLLC) capability of 5G has created significant opportunities for industrial wireless connectivity, yet widespread use of cellular networks for closed-loop control remains challenging. Closed-loop control requires more than low packet latency and high reliability: cyclic command/feedback exchanges must complete within predictable time bounds despite changing channel conditions, recovery transmissions, mobility, and multi-device contention. This paper introduces Deterministic-5G (Det-5G), a unified radio resource allocation framework for industrial closed-loop control. Det-5G treats the complete bidirectional control cycle as the scheduling object and combines coordinated downlink/uplink allocation, adaptive bundled transmissions, group-oriented downlink communication, and optimized multi-user uplink scheduling over 5G air-interface. Its performance is evaluated through a combination of closed-form analysis and Monte Carlo scheduling experiments, with comparisons against conventional dynamic grant-based scheduling, semi-persistent scheduling/configured grant operation, and fixed proactive repetition. The evaluation shows that Det-5G improves predictability of cycle completion, maintains the target reliability under changing link conditions, and scales more effectively to multi-device control than conventional reactive scheduling, while adapting radio resource use instead of continuously provisioning for the worst case as in fixed repetition. These characteristics make cycle-oriented scheduling a pragmatic solution for reducing the determinism gap that limits the use of 5G for closed-loop control in different verticals, especially as it evolves through 5G-Advanced toward 6G.

I. INTRODUCTION

Industrial control needs predictable completion of causally linked command/feedback cycles, not merely low packet latency and high reliability. Det-5G addresses the gap with cycle-oriented scheduling, while prior 5G work typically retains packet- or flow-level scheduling with reactive recovery.

I. INTRODUCTION

  • The adoption gap is unpredictable command/feedback completion, which remains a barrier to production-critical control despite private 5G connectivity.Industrial guidance identifies non-deterministic timing as a production risk.Source: §I · p. 1
  • Det-5G jointly treats command and feedback as one control transaction, allocating cycle-level reliability, recovery, and acknowledgement resources.Its evaluation compares dynamic grant-based scheduling, SPS/CG, and fixed proactive repetition using analytical and Monte Carlo methods.Source: §I · p. 1
  • For multi-device control, Det-5G combines common downlink delivery, selective recovery, and optimized sharing of uplink time-frequency resources.The framework is intended for individual control loops and coordinated device groups.Source: §I · p. 1

II. RELATED WORK

  • Prior 5G research studied latency, reliability, retransmission, and scheduling, but general industrial schedulers typically retain packet- or flow-level scheduling with reactive recovery, [17].The stated gap is a unified cycle-oriented treatment combining proactive reliability, cycle-level recovery, group downlink, and scalable multi-device allocation within 5G NR.Source: §II · p. 2

III. 5G NR AND RESOURCE ALLOCATION PRELIMINARIES

NR offers scalable numerology and shorter-than-slot transmission intervals for finer time placement. Conventional scheduling mechanisms reduce recurring overhead, but do not bind both directions into a predictable command/feedback cycle.

A. NR time structure

  • NR supports scalable OFDM numerology and 2-, 4-, and 7-symbol transmission intervals, enabling non-slot-based placement at finer time granularity.The paper uses SBT for complete selected intervals and NSBT for flexible symbol-level placement.Source: A. NR time structure · p. 2

B. Conventional scheduling for cyclic control

  • Dynamic scheduling handles downlink and uplink independently, while HARQ makes cycle completion depend on random decoding outcomes.SPS/CG reduces recurring control overhead, but does not intrinsically bind the directions into one transaction; fixed repetition trades timing uncertainty for resource use or reliability.Source: B. Conventional scheduling for cyclic… · p. 2

IV. DET-5G: DESIGN AND PROTOCOL OPERATION

Det-5G schedules the radio-interface portion of a complete control cycle, jointly provisioning its downlink and uplink. Its bundled transmissions place reliability resources before the deadline and adapt their length to link conditions.

A. System model and design principles

  • Det-5G co-designs downlink and uplink, reserves resources for a self-contained cycle, adapts bundled transmissions, and uses group communication and uplink packing for multi-user operation.The modeled devices are connected before cyclic operation; connection setup is outside the recurring deadline.Source: A. System model and design principles · p. 2

B. Single-user closed-loop control

  • A joint allocation identifies resources for the command, uplink response, and cycle-level acknowledgement, with each direction sent singly or as a bundle.The block acknowledgement follows the uplink bundle rather than each individual element.Source: B. Single-user closed-loop control · p. 2
  • Reserved operation repeats the self-contained cycle periodically, whereas cycle-by-cycle allocation can track mobility, interference, or changing link conditions.Reservation reduces recurring signaling; cycle-by-cycle operation maximizes adaptability while retaining the self-contained-cycle structure.Source: B. Single-user closed-loop control · p. 3

C. Single-user bundle-length adaptation

  • The update trigger avoids unnecessary bundle changes on stable links while responding to sustained degradation or rapidly varying reception.The two directions are adapted separately because they may have different interference and link budgets.Source: C. Single-user bundle-length adaptation · p. 3
  • Bundle length trades reliability against time-frequency use, so Det-5G adapts downlink and uplink bundles separately using recent received-signal statistics.It selects the smallest bundle meeting the receiver target plus an interference margin, subject to Bmax.Source: C. Single-user bundle-length adaptation · p. 3

D. Multi-user closed-loop control

Multi-user Det-5G avoids repeating the full single-user exchange for each device by sharing downlink delivery and packing uplink feedback. Group downlink bundle length is set using the member with the worst current link quality.

D. Multi-user closed-loop control

  • A group downlink delivers the common command, while a multi-user bundled uplink packs multiple devices’ feedback into the same time-frequency region.The cycle proceeds from joint allocation to the group downlink bundle, packed uplink bundle, and block acknowledgement.Source: D. Multi-user closed-loop control · p. 3
  • The group downlink bundle uses the single-user adaptation for the member with the worst current link quality to meet a common reliability target.Devices that fail to complete a cycle can be regrouped for subsequent joint allocation, avoiding unnecessary repetition by successful devices.Source: D. Multi-user closed-loop control · pp. 3–4

E. G-downlink optimization and group recovery

Group downlink reliability is set by the weakest device, while group-cycle acknowledgements allow recovery to target only failed members. Multi-user uplink allocation packs device-specific bundles in either SBT intervals or symbol-level NSBT regions.

E. G-downlink optimization and group recovery

  • The group downlink uses the weakest device’s current link to set a common reliability target, so persistently weak members can lengthen the bundle for everyone.Dynamic grouping can cluster devices with comparable links or isolate a temporarily degraded device for a later cycle.Source: E. G-downlink optimization and group… · p. 4
  • Group-cycle acknowledgement lets the gNB reallocate only for the incomplete subset, so successful devices need not repeat their exchange after another member fails.With periodic reservation, the regular group schedule remains reserved and exceptional recovery cycles are inserted for failed members.Source: E. G-downlink optimization and group… · p. 4

F. Multi-user bundled uplink optimization

  • For SBT uplink operation, device-specific RB demands are packed into the available RB pool, and group bundle length grows only when another SBT is needed.The objective minimizes group bundle length while meeting every device’s bundle and quality requirements.Source: F. Multi-user bundled uplink… · p. 4
  • NSBT allocation uses symbol-level time-frequency regions, scheduling the longest transmission first and filling residual opportunities with other devices.This avoids forcing every transmission to occupy a full SBT interval and is particularly effective for heterogeneous links and payloads.Source: F. Multi-user bundled uplink… · p. 4

V. PERFORMANCE EVALUATION

The evaluation combines closed-form timing and reliability analysis with Monte Carlo air-interface scheduling experiments. It uses a periodic industrial motion-control profile and models both proactive bundle sizing and reactive HARQ recovery.

V. PERFORMANCE EVALUATION

  • Evaluation combines closed-form analysis with Monte Carlo scheduling; analytical figures have no sampling uncertainty, while simulation confidence intervals use the stated independent runs.Fig. 8 uses 10,000 scheduler realizations per device-count point; Fig. 10 uses 30 runs of 100,000 cycles per speed.Source: §V · p. 5
  • The main setup uses 64-byte messages in each direction over a 3.8 GHz, 40 MHz carrier, with 30 kHz SCS, a 7-symbol 0.25 ms SBT, and 106 RBs.The per-direction residual target is 10−5, giving an approximate cycle target of 2 × 10−5; the main recovery increment is 1 ms.Source: §V · p. 5
  • For analytical proactive sizing, each link state receives the smallest bundle meeting the per-direction error target under an independent repetition model.The mapping is an evaluation model; the protocol may instead obtain the bundle from measured link quality and coding gain.Source: §V · p. 5
  • Reactive HARQ baselines model downlink and uplink transmission failures as independent geometric variables to derive cycle deadline-failure probability and expected completion time.The deadline model uses the number of recovery opportunities fitting before the deadline, the no-recovery cycle time, and the recovery interval.Source: §V · p. 5

A. Joint cycle time and deadline reliability

Under a common 4 ms deadline, Det-5G’s state-dependent bundling meets the cycle-failure target with lower average cycle time than fixed K = 5. Dynamic scheduling and SPS/CG have shorter averages but exceed the target.

A. Joint cycle time and deadline reliability

  • 4.7 × 10−6 cycle failure and 2.15 ms average cycle time are reported for Det-5G, versus 3.0 × 10−6 and 3.25 ms for fixed K = 5.Relative to fixed K = 5, Det-5G lowers average cycle time by about 34% and average proactive repetition occupancy by 44% while satisfying the target.Source: A. Joint cycle time and deadline… · p. 5, Figure 6 · p. 5
  • Dynamic scheduling averages 1.54 ms with cycle failure 5.6×10−4, while SPS/CG averages 1.04 ms with cycle failure 6.9×10−5.Both remain above the cycle-failure target in the comparison.Source: A. Joint cycle time and deadline… · p. 5
  • Fixed K = 2 averages 1.75 ms but has cycle failure 3.1 × 10−3, whereas fixed K = 5 reaches the target by fixing every cycle at 3.25 ms.Fig. 6 compares expected cycle time and complete-cycle failure probability under the common 4 ms deadline.Source: A. Joint cycle time and deadline… · p. 5, Figure 6 · p. 5

B. Retransmission-induced cycle time variation

At fixed BLER, reactive HARQ introduces cycle-time variation, while Det-5G and fixed proactive schedules have zero reactive variation once a cycle allocation is selected.

B. Retransmission-induced cycle time variation

  • HARQ-based scheduling develops 1–3 ms of cycle-time variation across the evaluated BLER range, while Det-5G and fixed proactive schedules have zero reactive variation after allocation.Link adaptation may change the predetermined budget between cycles, but does not append an unplanned HARQ tail to the allocated cycle.Source: B. Retransmission-induced cycle time… · p. 5, Figure 7 · p. 6
  • The variation measure is the 99.9th–0.1th percentile spread at fixed BLER, calculated from the exact negative-binomial distribution of aggregate downlink and uplink HARQ failures.The analytical spread is not estimated from finite trials.Source: B. Retransmission-induced cycle time… · p. 5, Figure 7 · p. 6
  • Det-5G’s selected cycle allocation has no unplanned HARQ tail, although link adaptation can change the predetermined budget between cycles.The figure isolates reactive timing uncertainty at fixed BLER.Source: B. Retransmission-induced cycle time… · p. 5, Figure 7 · p. 6

C. Multi-user cyclic scalability

With frequency multiplexing enabled for all schemes, Det-5G NSBT lowers 16-device tail cycle time relative to dynamic and SPS/CG scheduling. Its advantage emerges as larger groups increase the chance that an independent transaction needs recovery.

C. Multi-user cyclic scalability

  • The comparison gives conventional dynamic and SPS/CG schedulers the same 106-RB pool and CQI-dependent RB demand as Det-5G, while allowing frequency multiplexing.The baselines retain per-UE downlink/uplink transactions and reactive HARQ; Det-5G uses one group downlink and proactively packs bundled uplink transmissions.Source: C. Multi-user cyclic scalability · p. 6
  • At 16 devices, Det-5G NSBT reaches a 99th-percentile cycle time of 3.43 ms, versus 5.25 ms for dynamic scheduling and 4.75 ms for SPS/CG.These are about 35% and 28% lower, respectively; Det-5G SBT reaches 3.75 ms.Source: C. Multi-user cyclic scalability · p. 6
  • As group size grows, the probability that at least one independent transaction requires recovery increases.The study therefore compares 99th-percentile cycle time rather than mean latency, which favors reactive schemes when no recovery is needed.Source: C. Multi-user cyclic scalability · p. 6

D. Stringent latency/reliability stress test

The stress study tests a 1 ms full-cycle deadline at a 10^-6 cycle-failure target. Only sufficiently fine transmission granularity meets that deadline in the evaluated configurations, with Det-5G providing the largest timing margin.

D. Stringent latency/reliability stress test

  • At the 10^-6 cycle-failure target, reactive baselines require five aggregate HARQ recovery opportunities, while Det-5G uses state-dependent bundles B = 3, 4, 7.The 1 ms full-cycle deadline is an intentionally stringent radio-side stress point, not a direct conversion of the cited service-level availability.Source: D. Stringent latency/reliability… · p. 6
  • With a 60 kHz 2-symbol NSBT opportunity, cycle times are 0.93 ms for dynamic scheduling, 0.86 ms for SPS/CG, and 0.61 ms for Det-5G/fixed K = 7.At 60 kHz/7 symbols, the respective times are 3.25 ms, 3.0 ms, and 2.125 ms; the strict 1 ms deadline is reached only with sufficiently fine granularity in the evaluated set.Source: D. Stringent latency/reliability… · p. 6, Figure 9 · p. 6
  • Fixed K = 7 has the same worst-state timing bound as Det-5G but repeats seven times in every state.The analytical comparison spans 5G air-interface configurations at the 10^-6 target, with the 1 ms line marking the stress deadline.Source: D. Stringent latency/reliability… · p. 6, Figure 9 · p. 6

E. Mobility and bundle update overhead

Event-triggered reserved operation changes bundles only after persistent link-state transitions, limiting update actions while keeping the allocated cycle predetermined. The conclusion presents Det-5G as a cycle-oriented approach using existing NR mechanisms, with performance depending on transmission granularity in the strictest stress case.

E. Mobility and bundle update overhead

  • Bundle updates occur in 2.81%, 7.48%, 11.10%, and 16.70% of cycles at 1, 3, 5, and 10 m/s, respectively.The study uses a two-cycle persistence rule; nonzero-speed 95% confidence half-widths are below 0.04 percentage points.Source: E. Mobility and bundle update overhead · p. 6
  • Relative to cycle-by-cycle Det-5G, event-triggered reserved operation reduces bundle update actions by 97.2%, 92.5%, 88.9%, and 83.3% at 1, 3, 5, and 10 m/s.SPS/CG can require less routine reconfiguration, while fixed worst-case repetition pays continuous resource cost.Source: E. Mobility and bundle update overhead · p. 6
  • Reserved Det-5G changes reliability resources only when link state persistently changes, while the allocated cycle remains predetermined.This is presented as a middle ground between adaptation and continuously provisioning for the worst case.Source: E. Mobility and bundle update overhead · p. 7

VI. CONCLUDING REMARKS

  • At the main 4 ms operating point, Det-5G meets the cycle-failure target with a 2.15 ms average cycle and 44% less proactive repetition occupancy than fixed K = 5.Dynamic and SPS/CG scheduling have shorter average cycle times but exceed the cycle-failure target.Source: §VI · p. 7
  • The radio target requires sufficiently fine transmission granularity; Det-5G provides the largest timing margin among evaluated schemes, while event-triggered mobility adaptation requires bundle updates in only about 7.5% of cycles at 3 m/s.Source: §VI · p. 7
  • Det-5G uses existing NR mechanisms and is presented as complementary to 5G/TSN convergence, whose synchronization and QoS coordination do not remove radio-side timing uncertainty.The conclusion connects bounding the full wireless control transaction with extending deterministic behavior across converged wired/wireless systems.Source: §VI · p. 7
Loading 2609.07386v1…