Adaptive 5G Resource Allocation for Multistatic ISAC-Based UAV Detection and Tracking

Technical Summary: Adaptive 5G Resource Allocation for Multistatic ISAC-Based UAV Detection and Tracking

Problem Statement
The proliferation of small Unmanned Aerial Vehicles (UAVs) has created significant security risks near critical infrastructure and restricted airspace, necessitating robust detection and tracking capabilities. While Integrated Sensing and Communications (ISAC) offers a pathway to leverage existing 5G networks for surveillance, practical deployment faces two primary challenges:

1. Resource Competition: Sensing and communication functions must compete for finite radio resources. Static sensing allocations are inefficient under time-varying network loads, leading to either wasted resources during low traffic or degraded sensing performance during congestion.
2. Tracking Robustness: Tracking highly maneuverable UAVs with low Radar Cross Sections (RCS) is difficult, particularly when network resources are constrained. Single-modality systems (e.g., monostatic radar) often suffer from coverage gaps and localization inaccuracies in complex environments.

Methodology
The authors propose a load-aware, adaptive multistatic ISAC framework operating over the 5G New Radio (NR) interface. The system architecture integrates an active 5G gNB (serving as a controlled illuminator) with a distributed network of Software-Defined Sensor (SDS) nodes acting as passive receivers.

Key methodological components include:

• Adaptive Waveform Design: The gNB dynamically adjusts the length of Zadoff-Chu (ZC) sensing sequences ($N_{ZC}$), the sensing transmission rate, and beam allocation based on real-time network load ($\eta$) and observed Signal-to-Noise Ratio (SNR). Longer sequences provide higher coherent processing gain and better range resolution but consume more Physical Resource Blocks (PRBs).
• Shared-Resource Modeling: A mathematical model quantifies the trade-off between communication throughput and sensing overhead. The system defines four resource-allocation policies: fixed sensing-oriented, fixed communications-oriented, adaptive ISAC, and adaptive ISAC with Signal of Opportunity (SoO) assistance.
• SoO-Assisted Multistatic Sensing: When network resources are too constrained to support active 5G sensing requirements, the framework leverages external Signals of Opportunity (e.g., satellite, FM broadcast, or other cellular transmissions) via passive SDS nodes. These nodes capture target reflections without consuming additional 5G resources.
• Performance Metrics:
  • Detection/Tracking: Analyzed using the Marcum Q-function for detection probability and an $M$-of-$N$ model for tracking continuity (requiring $M$ successful detections out of $N$ opportunities).
  • Localization Accuracy: Evaluated using the Cramér-Rao Lower Bound (CRLB) to compare monostatic (gNB only) versus multistatic (gNB + SDS) geometries.

Key Contributions
The paper makes four specific contributions to the field of 5G ISAC:

1. Load-Aware Resource Model: Development of a model that quantifies sensing overhead as a function of ZC waveform length, update rate, and beam count under shared 5G resources.
2. Trade-off Characterization: Demonstration that adaptive waveform scaling allows 5G sensing to preserve tracking capabilities while gracefully degrading performance as communication load increases, avoiding the rigid inefficiencies of fixed reservations.
3. SoO-Assisted Architecture: Introduction of a distributed sensing architecture where SDS nodes exploit external SoO to maintain tracking continuity when active 5G resources are scarce, significantly reducing tracking outages under heavy congestion.
4. Multistatic CRLB Framework: A theoretical analysis showing that multistatic geometries significantly improve localization accuracy and spatial coverage uniformity compared to monostatic sensing, particularly when leveraging longer waveforms and wider bandwidths.

Results
Simulation results based on a 5G NR FR1 system (3.5 GHz, 50 MHz bandwidth) demonstrate the following:

• Throughput vs. Sensing: Adaptive policies outperform fixed-reservation baselines. While fixed sensing policies cap throughput (e.g., at 85% or 97% of offered load), the adaptive policy tracks the offered load closely.
• Congestion Resilience: Under extreme congestion (normalized load $\eta \to 1.0$), the "Adaptive + SoO" policy maintains throughput equal to the offered load while keeping tracking outage probability low. In contrast, the purely adaptive policy (without SoO) experiences rapid tracking failure as resources are diverted to communication.
• Waveform Scaling: Increasing the ZC sequence length ($N_{ZC}$) from 71 to 1291 improves the required input SNR for a target tracking outage probability ($10^{-3}$) by approximately 12.6 dB, extending reliable tracking into lower-SNR conditions.
• Localization Accuracy: CRLB analysis reveals that multistatic sensing significantly outperforms monostatic sensing. For a long waveform ($N_{ZC}=1291$), the mean Position Error Bound (PEB) for multistatic sensing drops to 0.85 m, whereas monostatic sensing remains stagnant at ~6.9 m, highlighting the geometric benefits of distributed nodes.

Significance
The paper posits that coordinated adaptive sensing combined with distributed multistatic support represents a practical path toward resilient UAV surveillance in future wireless networks. By dynamically scaling sensing resources and leveraging passive external signals, the proposed framework addresses the critical tension between maintaining high-quality communication services and ensuring persistent, accurate UAV tracking in congested RF environments. The work validates that ISAC can move beyond static resource partitioning to a dynamic, load-aware paradigm that utilizes the full potential of existing and opportunistic wireless infrastructure.