Tri-Hybrid Beamforming Design for integrated Sensing and Communications

TL;DR

This work tackles energy-efficient integrated sensing and communications (ISAC) for mmWave/THz with ultra-large antenna arrays by introducing a tri-hybrid beamforming framework that combines digital, analog, and dynamic metasurface antenna (DMA) beamformers. It formulates a non-convex multi-objective problem that jointly optimizes $w$, $oldsymbol f$, and $oldsymbol M$ to maximize a weighted sum of communications SNR $|oldsymbol h^H oldsymbol M oldsymbol f w|^2$ and sensing power $|oldsymbol g^H oldsymbol M oldsymbol f w|^2$, under unit-modulus analog constraints, DMA feasibility, and a total transmit power $P_t$. An efficient alternating optimization algorithm with closed-form updates and Dinkelbach’s transform is developed, achieving monotone convergence and a complexity of $oxed{ ext{O}(I N_r^2)}$. Simulation results show that tri-HBF substantially improves spatial gain and energy efficiency over fully digital and conventional hybrid architectures, with only modest degradation in beam alignment. This highlights the practicality of scalable, energy-efficient ISAC systems based on tri-hybrid beamforming for next-generation wireless networks.

Abstract

Tri-hybrid beamforming architectures have been proposed to enable energy-efficient communications systems in extra-largescale antenna arrays using low-cost programmable metasurface antennas. We study the tri-hybrid beamforming design for integrated sensing and communications (ISAC) to improve both communications and sensing performances. Specifically, we formulate a multi-objective optimization problem that balances communications signal-to-noise ratio (SNR) and the sensing power at a target direction, subject to constraints on the total power consumption and physical limitations inherent to the trihybrid beamforming architecture. We develop an efficient iterative algorithm in which the variables are updated in a closed form at each iteration, leading to a low-complexity and fast-execution design. Numerical results show that the tri-hybrid architecture improves spatial gain and energy efficiency, though with reduced beam alignment capability compared to conventional hybrid beamforming architectures.

Figures (4)

• Figure 1: Illustration of the system model.
• Figure 2: Convergence behavior of Algorithm \ref{['Alg: DT-SGPI']}.
• Figure 3: Performance tradeoff of considered architectures.
• Figure 4: Performance of different transceiver architectures with respect to the number of radiating elements per waveguide $N_{\rm u}$.