This paper investigates a secure downlink transmission scheme for an active simultaneous transmission and reflection reconfigurable intelligent surface (ASTAR-RIS)-assisted multi-user multiple-input-single-output (MU-MISO) wireless network under practical hardware impairment (HWI) conditions. In this work, non-ideal HWIs at the transceiver nodes, such as phase noise and quantization errors, are also taken into account. To ensure full spatial coverage and strengthen physical layer security (PLS), an ASTAR-RIS is considered that is capable of simultaneously supporting both transmission and reflection functionalities. A joint optimization framework is proposed to maximize the total secrecy rate ( \({R}_{\text{sec}}\) ) of legitimate users while satisfying the minimum rate requirements and power constraints. The optimization variables include the transmit beamforming vectors ( \({\mathbf{w}}_{\chi }\) ), the ASTAR-RIS phase shifts \(({\phi }_{n}^{\chi },{\phi }_{n}^{\overline{\chi }})\) , and amplitude coefficients \(({\alpha }_{n}^{\chi },{\alpha }_{n}^{\overline{\chi }})\) . To tackle the non-convexity of the proposed problem, an iterative alternating optimization (AO)-based algorithm is developed that integrates semidefinite relaxation (SDR), particle swarm optimization (PSO), and the interior-point method (IPM). Specifically, SDR is adapted to incorporate secrecy rate constraints in beamforming, PSO is tailored to jointly optimize transmission and reflection phase shifts, and IPM updates amplitude coefficients. This synergistic combination is novel in the context of secure ASTAR-RIS-assisted MU-MISO networks with non-ideal transceiver HWIs, and it significantly improves secrecy performance compared to conventional passive STAR-RIS systems. Numerical results confirm the effectiveness of the proposed scheme, which demonstrates that ASTAR-RIS achieves up to \(43\text{\%}\) higher \({R}_{\text{sec}}\) than passive STAR-RIS, even in the presence of HWIs and eavesdroppers.
