⚡ electrical engineering

A Robust Two-Stage Protocol for STAR-RIS-Aided ISAC Networks: Joint Beamforming and Mode Optimization

This paper proposes a robust two-stage protocol for STAR-RIS-aided ISAC networks that jointly optimizes beamforming, mode selection, and RIS coefficients to maximize communication sum-rate while guaranteeing sensing accuracy under DoA estimation errors and imperfect NLoS channel knowledge.

Original authors: Ziming Liu, Tao Chen, Giacinto Gelli, Vincenzo Galdi, Francesco Verde

Published 2026-06-23

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Ziming Liu, Tao Chen, Giacinto Gelli, Vincenzo Galdi, Francesco Verde

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a busy city street where a smart building needs to do two things at once: talk to people inside the building (like sending emails or streaming video) and keep an eye on people outside (like tracking pedestrians for safety). Usually, a "smart mirror" on the wall (called a RIS) can only reflect signals to one side. But this paper introduces a super-mirror called a STAR-RIS that can both reflect signals to the outside and let signals pass through to the inside simultaneously.

Here is the simple breakdown of how the authors solved the problem of making this work efficiently and reliably, even when things aren't perfect.

The Problem: The "Foggy Window" and the "Moving Target"

The authors identified two main headaches in real-world scenarios:

The Moving Target: People outside are walking around. The system tries to guess where they are (like aiming a flashlight), but the guess isn't perfect. If you aim based on a bad guess, the signal misses.
The Foggy Window: The signals bouncing off buildings and trees (the "indirect" path) are hard to measure instantly. Most previous designs assumed we knew exactly how these bounces worked, which is like assuming you know exactly how a foggy window distorts an image before you even look through it. In reality, we only know the average behavior of the fog, not the exact moment-to-moment distortion.

The Solution: A Two-Step Dance

Instead of trying to do everything at once, the authors propose a two-stage protocol (a two-step dance) that happens every time the system sends a message:

Step 1: The "Preparation" Phase (The Scout):
Think of this as a scout running ahead. The system sends out signals to:
- Talk to everyone (inside and outside).
- Listen to the people outside to get a better guess of where they are.
- Crucially: It uses a "safety net" approach. Since it knows the guess might be slightly off (due to the "fog"), it designs the signal to be robust, ensuring it works even if the guess is a little wrong.
Step 2: The "Communication" Phase (The Sprint):
Now that the scout has a better idea of where the people outside are, the system switches gears. It uses this fresh information to beam the data much more precisely. It's like a quarterback who first checks the field, then throws the ball with perfect aim based on that new info.

The "Smart Partition" Trick

The STAR-RIS is made of thousands of tiny elements (like pixels on a screen). The authors realized you don't need every pixel to do everything at the same time.

Some pixels can be set to split energy (sending some signal out, some through).
Other pixels can be set to transmit only (letting everything through).
The system dynamically decides which pixels do what, like a conductor telling different sections of an orchestra to play louder or softer to get the best sound.

The "Statistical" Safety Net

Instead of trying to measure the exact "fog" (the complex bouncing signals) every single second—which is impossible and wastes energy—the system uses statistics.

Imagine you are driving in the rain. You don't need to know the exact path of every single raindrop hitting your windshield. You just need to know that, on average, the road is slippery.
The authors' design uses this "average" knowledge to build a system that works reliably 99% of the time, without needing to know the exact position of every raindrop.

The Results: Why It Matters

The authors tested their idea against other methods (like those that assume perfect knowledge or don't split the process into two steps).

The Win: Their method achieved about 15% more data speed (throughput) than the best competing method.
The Robustness: Even when the system's guess about where people are was wrong (due to estimation errors), their design kept working well. The other methods crashed or slowed down significantly when the "guess" was bad.
The Sensing: They also proved that the system could still "see" the people outside accurately enough to satisfy safety requirements, even with all the uncertainty.

In a Nutshell

This paper presents a smarter way to use a "super-mirror" for 6G networks. Instead of guessing perfectly and hoping for the best, it uses a two-step process (scout then sprint) and statistical safety nets to handle the messiness of the real world. The result is a system that is faster, more reliable, and better at handling the fact that we can never know everything about the environment perfectly.

Technical Summary: A Robust Two-Stage Protocol for STAR-RIS-Aided ISAC Networks

Problem Statement
This paper addresses the robust design of Integrated Sensing and Communication (ISAC) systems assisted by Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RISs). While STAR-RISs offer full-space electromagnetic control—enabling concurrent service to indoor (transmission space) and outdoor (reflection space) users—practical deployment faces two critical challenges:

Sensing Uncertainty: In dynamic outdoor environments, the Direction-of-Arrival (DoA) of mobile users is not perfectly known and must be estimated. This uncertainty degrades beamforming accuracy if not explicitly modeled.
Imperfect Channel Knowledge: Instantaneous estimation of Non-Line-of-Sight (NLoS) channel components in complex outdoor environments is difficult due to scattering complexity and pilot overhead. Existing works often assume perfect instantaneous NLoS knowledge, leading to overly optimistic performance predictions.

Methodology
The authors propose a two-stage ISAC protocol operating on a slot-by-slot basis to address these challenges:

Preparation Phase: A fraction $\eta$ of the time slot is dedicated to jointly performing DoA estimation for outdoor users and downlink communication to all users. During this phase, the STAR-RIS operates in a mixed mode, partitioning elements between Energy-Splitting (ES) and Transmit-Only (TO) configurations.
Communication Phase: The remaining fraction $(1-\eta)$ of the slot utilizes the acquired angular information from the preparation phase to refine downlink transmission, with all STAR-RIS elements operating in ES mode.

Statistical Modeling and Robust Design
To handle uncertainty without requiring instantaneous NLoS Channel State Information (CSI):

DoA Modeling: Outdoor user DoAs are modeled as Gaussian random variables, where the variance is determined by the estimation accuracy (Cramér-Rao Lower Bound).
Channel Modeling: The NLoS components of outdoor links are characterized by their long-term spatial covariance statistics rather than instantaneous realizations. The LoS components are treated as dependent on the estimated DoAs.
Optimization Objective: A performance-balanced optimization problem is formulated to maximize the average communication sum-rate (throughput) while guaranteeing a required sensing accuracy (measured by the accumulated average Signal-to-Noise Ratio, ASSNR).

Algorithmic Framework
The resulting optimization problem is non-convex and mixed discrete-continuous due to the coupling of beamforming vectors, STAR-RIS coefficients, binary mode-selection vectors, and physical constraints (energy conservation, phase coupling). The authors develop a tailored Alternating Optimization (AO) framework that includes:

Fractional Programming (FP): A Lagrangian dual transform and quadratic transform are used to convert the sum-rate objective into a tractable sum-of-ratios form.
Successive Convex Approximation (SCA): Used to handle non-convex constraints, particularly the sensing constraint and the binary nature of the mode-selection vector (via penalty relaxation).
Semidefinite Relaxation (SDR): Applied to optimize the STAR-RIS transmission and reflection coefficients, followed by a rank-one recovery and Euclidean projection to enforce phase-coupling constraints ( $\cos(\theta_R - \theta_T) = 0$ ).
Convergence: The algorithm is proven to monotonically converge to a stationary point.

Key Contributions

Two-Stage Protocol: Unlike single-stage designs, this protocol explicitly coordinates sensing and communication phases, allowing the system to exploit estimated angular information for subsequent communication refinement.
Robust Statistical Design: The framework incorporates DoA estimation errors and partial NLoS knowledge (via spatial covariance) into the optimization, eliminating the need for instantaneous NLoS CSI.
Joint Optimization: The solution jointly determines the base station (BS) beamforming vectors, STAR-RIS coefficients for both stages, and the dynamic partition of the metasurface between ES and TO modes.
Efficient Algorithm: A unified iterative procedure is developed with proven convergence and characterized computational complexity.

Numerical Results
Simulations compare the proposed robust design against several benchmarks, including schemes neglecting spatial statistics (NoStat-STAR), fixed manifold designs (Fixed-STAR), and single-stage protocols (One-stage).

Throughput Gain: The proposed design achieves approximately 15% throughput gain over the most competitive benchmark (which shares the optimization framework but neglects NLoS spatial covariance statistics).
Robustness: The system maintains performance advantages even under significant DoA estimation errors and imperfect NLoS channel knowledge.
Sensing Constraints: The sensing constraints are satisfied across all considered fluctuations in target response and residual scattering disturbance, with sensing margins consistently above unity.
Scalability: Throughput increases with the number of STAR-RIS elements and BS transmit power, with the proposed method showing superior scaling compared to benchmarks.

Significance
The paper claims that the proposed robust design effectively balances the sensing-communication trade-off in STAR-RIS-aided ISAC networks. By moving away from idealized assumptions of perfect channel knowledge and leveraging a two-stage protocol with statistical modeling, the design offers a more practical and reliable approach for next-generation 6G networks, particularly in scenarios involving heterogeneous indoor/outdoor users and dynamic outdoor environments. The work highlights that explicitly accounting for estimation uncertainty and statistical channel properties yields significant performance gains over designs that ignore these factors.