CRB Minimization for RIS-Aided mmWave Integrated Sensing and Communications

Wanting Lyu, Songjie Yang, Yue Xiu, Ya Li, Hongjun He, Chau Yuen, Zhongpei ZhangIEEE Internet of Things Journal 2024 · DOI

TL;DR

This paper tackles the challenge of simultaneously performing sensing and communication in millimeter-wave (mmWave) networks using Reconfigurable Intelligent Surfaces (RIS). They propose a novel optimization framework to minimize the sensing accuracy bound (Cramér-Rao Bound) while ensuring communication quality, achieving near-optimal sensing performance with significantly reduced hardware complexity compared to full digital systems.

What Problem Does This Paper Address?

The integration of sensing and communication (ISAC) in next-generation wireless networks (B5G/6G) promises enhanced capabilities, from autonomous driving to smart environments. Millimeter-wave (mmWave) frequencies offer vast bandwidth for both, but suffer from severe path loss and blockage. Reconfigurable Intelligent Surfaces (RIS) are emerging as a solution to mitigate these issues by intelligently reflecting signals. However, optimizing ISAC systems, especially in the mmWave band with RIS, presents significant challenges:

  • Multi-hop Attenuation: Signals traversing multiple hops (e.g., BS to RIS to target) experience substantial signal loss.
  • Sensing Accuracy vs. Communication Quality: There's an inherent trade-off between optimizing for precise sensing (e.g., target localization) and ensuring reliable communication for users.
  • Hardware Complexity: mmWave systems, especially with hybrid beamforming, require careful design to balance performance and cost.
  • Joint Optimization: Optimizing beamforming for both sensing and communication, along with RIS phase shifts, is a complex, non-convex problem.

This paper specifically addresses the challenge of minimizing the sensing accuracy bound (Cramér-Rao Bound - CRB) in a RIS-aided mmWave ISAC system while guaranteeing communication data rates.

Methods

The authors propose a sophisticated optimization framework to tackle the complex, non-convex problem of minimizing the CRB for target angle estimation while ensuring user data rates. Key methodological contributions include:

  • Semi-Self Sensing RIS: A novel approach where sensors are integrated directly into the RIS to capture radar echoes, alleviating multi-hop attenuation.
  • CRB Minimization Framework: The core problem is formulated as minimizing the trace of the CRB matrix for target angle estimation, subject to communication data rate constraints.
  • Fisher Information Matrix (FIM) Transformation: The CRB minimization problem is reformulated using the FIM, making it more tractable.
  • Block Coordinate Descent (BCD) with Penalty-CCCP: To handle the coupled non-convex problem, a BCD approach is employed, decomposing it into sub-problems. A penalty-CCCP method is used to address non-convexity in hybrid beamforming optimization.
  • Successive Convex Approximation (SCA) and Second-Order Cone (SOC) Constraints: These techniques are applied to solve the non-convex hybrid beamforming optimization.
  • Manifold Optimization (MO): For optimizing unit-modulus constrained analog beamforming and RIS phase shifts, MO is utilized, leveraging complex circle manifold properties.

Datasets

Not discussed in the paper. The paper focuses on a simulated mmWave ISAC network with defined system parameters.

Key Insights and Ideas

  • RIS as a Dual-Purpose Enabler: The research demonstrates that RIS can effectively serve a dual role: enhancing communication links and acting as a sensing platform by integrating sensors.
  • CRB as a Practical Sensing Metric: Using the Cramér-Rao Bound as the optimization objective provides a theoretically grounded way to improve sensing accuracy, directly linking it to the variance of parameter estimation.
  • Hybrid Beamforming's Sweet Spot: The proposed hybrid beamforming approach, combined with RIS, offers a compelling balance between performance and hardware cost, achieving near-full digital performance with significantly fewer RF chains.
  • Trade-offs are Quantifiable: The paper clearly illustrates the trade-off between sensing accuracy (CRB) and communication demands (data rate threshold), providing valuable insights for system design.
  • Algorithm Effectiveness: The proposed multi-stage optimization algorithm (BCD, Penalty-CCCP, SCA, MO) effectively converges to a solution that significantly improves sensing performance.

Key Results

  • Significant CRB Reduction: The proposed algorithm achieves substantial reductions in the CRB for target angle estimation compared to baseline methods, especially those without optimized RIS phase shifts.
  • Near-Full Digital Performance: The hybrid beamforming approach with 6 RF chains achieves approximately 96% of the sensing accuracy (as measured by CRB) of a full digital beamforming system (with 128 RF chains), highlighting its efficiency.
  • Convergence: The proposed algorithms (SCA and BCD) demonstrate fast convergence to optimal or near-optimal solutions.
  • Impact of System Parameters: The results show that increasing transmit power, the number of RIS reflecting elements, and the number of RIS sensing elements all contribute to improved sensing accuracy.
  • Trade-off Visualization: Figure 6 clearly shows that as the data rate threshold increases, the sensing performance (CRB) deteriorates, quantifying this critical trade-off.

Limitations and Future Work

  • Assumptions: The paper assumes a point-like target and a simplified channel model (Swerling-I for RCS fluctuations). The target's direction is assumed to be fixed.
  • Near-Field and Extended Targets: The current work focuses on far-field point targets. Future work could explore near-field scenarios and extended targets, which present more complex modeling challenges.
  • Clutter and Multi-RIS: Incorporating clutter in the environment and considering multi-RIS systems could lead to more realistic and robust ISAC designs.
  • User and Target Location Estimation Error: The impact of estimation errors in user and target locations on the system's performance is not explicitly addressed and could be a future research direction.
  • Hardware Constraints: While hybrid beamforming addresses hardware cost, further research could delve into more specific hardware impairments and their impact.

Summary updated

BibTeX
@article{lyu2024crb,
  author = {Wanting Lyu and Songjie Yang and Yue Xiu and Ya Li and Hongjun He and Chau Yuen and Zhongpei Zhang},
  title = {{CRB Minimization for RIS-Aided mmWave Integrated Sensing and Communications}},
  journal = {IEEE Internet of Things Journal},
  year = {2024},
  doi = {10.1109/jiot.2024.3361939},
}
Figure