Sepideh Javadi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sepideh Javadi has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Mohammad Robat Mili
Pasargad Institute for Advanced Innovative Solutions (PIAIS), Tehran, Iran
9 shared publications - Naofal Al-Dhahir
The University of Texas at Dallas
8 shared publications - Sajad Faramarzi
School of Electrical Engineering, Iran University of Science and Technology, Tehran, Iran
5 shared publications - Eduard Axel Jorswieck
Institute for Communications Technology, TU Braunschweig, Braunschweig, Germany
4 shared publications - Farshad Zeinali
GIC & 6GIC, Institute for Communication Systems (ICS), University of Surrey, Guildford, U. K
3 shared publications - Hosein Zarini
Department of Computer Engineering, Sharif University of Technology, Tehran, Iran
3 shared publications
Latest Publications
- Open Access
- 11 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 89172-89184
Sum Rate Maximization for Beyond Diagonal-RIS-Aided Non-Terrestrial Networks Using RSMA
Non-terrestrial networks (NTNs) are expected to revolutionize global high-speed connectivity and enable low-latency communication for a wide range of applications. Despite their advantages, NTNs face critical challenges such as high power consumption, limited onboard processing resources, and severe spectral congestion. To address these issues, recent studies have introduced energy- and spectrum-efficient technologies such as Beyond Diagonal Reconfigurable Intelligent Surface (BD-RIS) and Rate-Splitting Multiple Access (RSMA), which have shown great potential in enhancing overall network performance. This paper investigates the problem of sum-rate maximization in BD-RIS-assisted NTNs integrated with RSMA, focusing on the joint optimization of multiple design variables. Specifically, the objective is to simultaneously design the beamforming vectors, adapt the RSMA transmission rates, and configure the BD-RIS phase-shift matrix. A non-convex optimization problem is formulated to capture this coupled design, which is inherently challenging due to its high dimensionality and nonlinear characteristics. To efficiently tackle this complexity, we propose a learning-based solution leveraging a Meta Twin Delayed Deep Deterministic Policy Gradient (Meta-TD3) algorithm. This advanced reinforcement learning framework enables adaptive and dynamic optimization under diverse channel conditions. Numerical simulations validate the effectiveness of the proposed approach, demonstrating that the BD-RIS-assisted scheme significantly outperforms conventional systems without BD-RIS in terms of achievable sum rate.
- 11 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 89172-89184
Open AccessSum Rate Maximization for Beyond Diagonal-RIS-Aided Non-Terrestrial Networks Using RSMA
Non-terrestrial networks (NTNs) are expected to revolutionize global high-speed connectivity and enable low-latency communication for a wide range of applications. Despite their advantages, NTNs face critical challenges such as high power consumption, limited onboard processing resources, and severe spectral congestion. To address these issues, recent studies have introduced energy- and spectrum-efficient technologies such as Beyond Diagonal Reconfigurable Intelligent Surface (BD-RIS) and Rate-Splitting Multiple Access (RSMA), which have shown great potential in enhancing overall network performance. This paper investigates the problem of sum-rate maximization in BD-RIS-assisted NTNs integrated with RSMA, focusing on the joint optimization of multiple design variables. Specifically, the objective is to simultaneously design the beamforming vectors, adapt the RSMA transmission rates, and configure the BD-RIS phase-shift matrix. A non-convex optimization problem is formulated to capture this coupled design, which is inherently challenging due to its high dimensionality and nonlinear characteristics. To efficiently tackle this complexity, we propose a learning-based solution leveraging a Meta Twin Delayed Deep Deterministic Policy Gradient (Meta-TD3) algorithm. This advanced reinforcement learning framework enables adaptive and dynamic optimization under diverse channel conditions. Numerical simulations validate the effectiveness of the proposed approach, demonstrating that the BD-RIS-assisted scheme significantly outperforms conventional systems without BD-RIS in terms of achievable sum rate.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-6
URLLC Service for Stacked Intelligent Metasurface-Assisted RSMA Systems
A stacked intelligent metasurface (SIM) enables high-speed electromagnetic wave processing by leveraging multiple layers, giving them more computational flexibility and efficiency compared to the simpler and single-layer reconfigurable intelligent surfaces (RIS). By performing wave-domain signal processing across multiple layers and employing low-complexity transmit RF chains, SIM significantly reduces precoding delay and power consumption, making it a compelling candidate for the sixth-generation (6 G) wireless systems. This study proposes a SIM-assisted downlink multi-user multiple-input single-output (MU-MISO) network using rate splitting multiple access (RSMA), which provides ultra-reliable low-latency communication (URLLC) services to all users. Specifically, an optimization problem is developed to maximize the sum rate, while satisfying both users' quality-of-service (QoS) requirements and the available power at the base station (BS). This is achieved through a hybrid optimization framework that combines the successive convex approximation (SCA) technique for optimizing the transmit power and common-rate allocation and a deep reinforcement learning (DRL)-based proximal policy optimization (PPO) algorithm for configuring the SIM phase shifts. Simulation results validate the effectiveness of the proposed hybrid approach, demonstrating superior sum-rate performance compared to other benchmarks.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-6
URLLC Service for Stacked Intelligent Metasurface-Assisted RSMA Systems
A stacked intelligent metasurface (SIM) enables high-speed electromagnetic wave processing by leveraging multiple layers, giving them more computational flexibility and efficiency compared to the simpler and single-layer reconfigurable intelligent surfaces (RIS). By performing wave-domain signal processing across multiple layers and employing low-complexity transmit RF chains, SIM significantly reduces precoding delay and power consumption, making it a compelling candidate for the sixth-generation (6 G) wireless systems. This study proposes a SIM-assisted downlink multi-user multiple-input single-output (MU-MISO) network using rate splitting multiple access (RSMA), which provides ultra-reliable low-latency communication (URLLC) services to all users. Specifically, an optimization problem is developed to maximize the sum rate, while satisfying both users' quality-of-service (QoS) requirements and the available power at the base station (BS). This is achieved through a hybrid optimization framework that combines the successive convex approximation (SCA) technique for optimizing the transmit power and common-rate allocation and a deep reinforcement learning (DRL)-based proximal policy optimization (PPO) algorithm for configuring the SIM phase shifts. Simulation results validate the effectiveness of the proposed hybrid approach, demonstrating superior sum-rate performance compared to other benchmarks.
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15 (21622337) , 1558-1562
URLLC Service for Beyond Diagonal RIS-Enabled ISAC Systems: A Meta-PPO Algorithm
This letter investigates rate-splitting multiple access (RSMA)-based short-packet downlink transmission in an integrated sensing and communication (ISAC) system under channel uncertainty. Specifically, a base station (BS) provides ultra-reliable low-latency communication (URLLC) services to multiple users with the aid of a beyond-diagonal reconfigurable intelligent surface (BD-RIS), while simultaneously detecting a target in the presence of clutter. The objective is to maximize the sum rate by jointly optimizing transmit beamforming, common rate allocation, and BD-RIS reflection beamforming. To address the optimization problem, we propose a Meta proximal policy optimization (Meta-PPO) algorithm, which is a fusion of Meta-learning techniques and the PPO algorithm. Numerical results demonstrate that the proposed BD-RIS-based model outperforms the conventional RIS scheme, while the proposed Meta-PPO algorithm achieves faster convergence and superior performance compared with other deep reinforcement learning (DRL)-based methods.
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15 (21622337) , 1558-1562
URLLC Service for Beyond Diagonal RIS-Enabled ISAC Systems: A Meta-PPO Algorithm
This letter investigates rate-splitting multiple access (RSMA)-based short-packet downlink transmission in an integrated sensing and communication (ISAC) system under channel uncertainty. Specifically, a base station (BS) provides ultra-reliable low-latency communication (URLLC) services to multiple users with the aid of a beyond-diagonal reconfigurable intelligent surface (BD-RIS), while simultaneously detecting a target in the presence of clutter. The objective is to maximize the sum rate by jointly optimizing transmit beamforming, common rate allocation, and BD-RIS reflection beamforming. To address the optimization problem, we propose a Meta proximal policy optimization (Meta-PPO) algorithm, which is a fusion of Meta-learning techniques and the PPO algorithm. Numerical results demonstrate that the proposed BD-RIS-based model outperforms the conventional RIS scheme, while the proposed Meta-PPO algorithm achieves faster convergence and superior performance compared with other deep reinforcement learning (DRL)-based methods.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (4) , 6951-6956
ISAC Multi-Active STAR-RIS-Aided NOMA System With Element Selection: A DRL Approach
In this paper, to obtain high efficiency and performance in a wireless communication system, a multi-active Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surface (STAR-RIS)-aided Integrated Sensing and Communication (ISAC) framework utilizing Non-Orthogonal Multiple Access (NOMA) technology is introduced. Unlike conventional STAR-RIS systems that employ either fully passive or fully active elements, the element selection approach allows each element to operate in either an active or passive mode to reduce power consumption. We aim to maximize the energy efficiency (EE) of the system by collaboratively optimizing the base station (BS) transmit beamforming vectors, the STAR-RIS coefficient matrices, and the element selection matrix. The formulated optimization problem is non-convex, which makes it challenging to solve using conventional methods. To tackle this challenge, we propose a deep reinforcement learning (DRL) approach to optimize the active-passive mode vector, phase shifts, and BS beamforming. Consequently, a twin delayed deep deterministic policy gradient (TD3) algorithm is applied to address the complex optimization problem. Simulation results reveal that our proposed method achieves a higher EE than current benchmarks.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (4) , 6951-6956
ISAC Multi-Active STAR-RIS-Aided NOMA System With Element Selection: A DRL Approach
In this paper, to obtain high efficiency and performance in a wireless communication system, a multi-active Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surface (STAR-RIS)-aided Integrated Sensing and Communication (ISAC) framework utilizing Non-Orthogonal Multiple Access (NOMA) technology is introduced. Unlike conventional STAR-RIS systems that employ either fully passive or fully active elements, the element selection approach allows each element to operate in either an active or passive mode to reduce power consumption. We aim to maximize the energy efficiency (EE) of the system by collaboratively optimizing the base station (BS) transmit beamforming vectors, the STAR-RIS coefficient matrices, and the element selection matrix. The formulated optimization problem is non-convex, which makes it challenging to solve using conventional methods. To tackle this challenge, we propose a deep reinforcement learning (DRL) approach to optimize the active-passive mode vector, phase shifts, and BS beamforming. Consequently, a twin delayed deep deterministic policy gradient (TD3) algorithm is applied to address the complex optimization problem. Simulation results reveal that our proposed method achieves a higher EE than current benchmarks.
- 18 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 29 (10) , 2441-2445
Efficient Resource Allocation for RSMA-Enabled OIRS-Assisted VLC Systems With Imperfect CSI
Efficient resource allocation is essential in visible light communication (VLC) systems. This study explores the use of optical intelligent reflecting surfaces (OIRS) to enhance the data rate and overall power efficiency of VLC systems under imperfect channel state information (IPCSI), leveraging rate-splitting multiple access (RSMA) for efficient interference management. To this end, we introduce a multi-objective optimization framework to explore the fundamental trade-off between maximizing effective rate and minimizing effective power. In contrast to conventional energy efficiency (EE) approaches, the proposed framework offers flexible system design by allowing the relative importance of rate and power objectives to be adjusted through the reward function. Additionally, it encourages fair resource allocation by penalizing imbalanced rate and power distributions among receivers. We propose a robust resource allocation framework based on the critical actor asynchronous meta-advantage (MA3C) reinforcement learning (RL) algorithm. Simulation results demonstrate substantial robustness to IPCSI via MA3C. Additionally, the integration of RSMA significantly boosts the total data rate and reduces total power, outperforming other benchmarks.
- 18 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 29 (10) , 2441-2445
Efficient Resource Allocation for RSMA-Enabled OIRS-Assisted VLC Systems With Imperfect CSI
Efficient resource allocation is essential in visible light communication (VLC) systems. This study explores the use of optical intelligent reflecting surfaces (OIRS) to enhance the data rate and overall power efficiency of VLC systems under imperfect channel state information (IPCSI), leveraging rate-splitting multiple access (RSMA) for efficient interference management. To this end, we introduce a multi-objective optimization framework to explore the fundamental trade-off between maximizing effective rate and minimizing effective power. In contrast to conventional energy efficiency (EE) approaches, the proposed framework offers flexible system design by allowing the relative importance of rate and power objectives to be adjusted through the reward function. Additionally, it encourages fair resource allocation by penalizing imbalanced rate and power distributions among receivers. We propose a robust resource allocation framework based on the critical actor asynchronous meta-advantage (MA3C) reinforcement learning (RL) algorithm. Simulation results demonstrate substantial robustness to IPCSI via MA3C. Additionally, the integration of RSMA significantly boosts the total data rate and reduces total power, outperforming other benchmarks.
- 2 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (12) , 19819-19824
Stacked Intelligent Metasurfaces and STAR-RIS-Enabled Terahertz ISAC System
In this letter, a novel terahertz (THz) band integrated sensing and communication (ISAC) framework is proposed, which synergistically integrates stacked intelligent meta-surfaces (SIM) with simultaneously transmitting and reflecting reconfigurable intelligent surfaces (STAR-RIS). In this system, SIM enables high-precision wave-based beamforming with reduced complexity, while STAR-RIS provides full-space coverage and facilitates partitioning sensing and communication spaces. Together, they address the key challenges of multi-target sensing and multi-user communication in the THz band. The Cramér-Rao bound (CRB) is derived for two-dimensional directionof-arrival (DoA) estimation, and an optimization problem is formulated to maximize energy efficiency (EE) under this CRB constraint. To address this coupled non-convex optimization problem, we propose the meta soft actor-critic (Meta-SAC) algorithm, a meta-learning-based solution that jointly optimizes power allocation, SIM phase shifts, and STAR-RIS transmissionreflection (T-R) coefficients. To efficiently solve this problem, the generative adversarial networks (GANs) are integrated with the Meta-SAC algorithm. Simulation results demonstrate significant improvements in EE and sensing accuracy, showcasing the transformative potential of SIM and STAR-RIS integration for next-generation ISAC networks.
- 2 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (12) , 19819-19824
Stacked Intelligent Metasurfaces and STAR-RIS-Enabled Terahertz ISAC System
In this letter, a novel terahertz (THz) band integrated sensing and communication (ISAC) framework is proposed, which synergistically integrates stacked intelligent meta-surfaces (SIM) with simultaneously transmitting and reflecting reconfigurable intelligent surfaces (STAR-RIS). In this system, SIM enables high-precision wave-based beamforming with reduced complexity, while STAR-RIS provides full-space coverage and facilitates partitioning sensing and communication spaces. Together, they address the key challenges of multi-target sensing and multi-user communication in the THz band. The Cramér-Rao bound (CRB) is derived for two-dimensional directionof-arrival (DoA) estimation, and an optimization problem is formulated to maximize energy efficiency (EE) under this CRB constraint. To address this coupled non-convex optimization problem, we propose the meta soft actor-critic (Meta-SAC) algorithm, a meta-learning-based solution that jointly optimizes power allocation, SIM phase shifts, and STAR-RIS transmissionreflection (T-R) coefficients. To efficiently solve this problem, the generative adversarial networks (GANs) are integrated with the Meta-SAC algorithm. Simulation results demonstrate significant improvements in EE and sensing accuracy, showcasing the transformative potential of SIM and STAR-RIS integration for next-generation ISAC networks.
- Open Access
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 6, 3806-3815
Resource Allocation in STAR-RIS-Aided SWIPT With RSMA via Meta-Learning
Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) is a cutting-edge concept for the sixth-generation (6G) wireless networks. In this paper, we propose a novel system that incorporates STAR-RIS with simultaneous wireless information and power transfer (SWIPT) using rate splitting multiple access (RSMA). The proposed system facilitates communication from a multi-antenna base station (BS) to single-antenna users in a downlink transmission. The BS concurrently sends energy and information signals to multiple energy harvesting receivers (EHRs) and information data receivers (IDRs) with the support of a STAR-RIS. Furthermore, an optimization is introduced to strike a balance between users’ sum rate (SR) and the total harvested energy. To achieve this, an optimization problem is formulated to optimize the energy/information beamforming vectors at the BS, the phase shifts at the STAR-RIS, and the common message rate. Subsequently, we employ a meta deep deterministic policy gradient (Meta-DDPG) approach to solve the complex problem. Simulation results demonstrate that the proposed Meta-DDPG algorithm performance compared to other base lines, yielding higher data rates and improved harvested energy. This advancement in algorithmic performance substantiates the effectiveness of the proposed STAR-RIS system with SWIPT and RSMA, positioning it as a promising solution for the resource-efficient, high-performance demands of 6G networks.
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 6, 3806-3815
Open AccessResource Allocation in STAR-RIS-Aided SWIPT With RSMA via Meta-Learning
Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) is a cutting-edge concept for the sixth-generation (6G) wireless networks. In this paper, we propose a novel system that incorporates STAR-RIS with simultaneous wireless information and power transfer (SWIPT) using rate splitting multiple access (RSMA). The proposed system facilitates communication from a multi-antenna base station (BS) to single-antenna users in a downlink transmission. The BS concurrently sends energy and information signals to multiple energy harvesting receivers (EHRs) and information data receivers (IDRs) with the support of a STAR-RIS. Furthermore, an optimization is introduced to strike a balance between users’ sum rate (SR) and the total harvested energy. To achieve this, an optimization problem is formulated to optimize the energy/information beamforming vectors at the BS, the phase shifts at the STAR-RIS, and the common message rate. Subsequently, we employ a meta deep deterministic policy gradient (Meta-DDPG) approach to solve the complex problem. Simulation results demonstrate that the proposed Meta-DDPG algorithm performance compared to other base lines, yielding higher data rates and improved harvested energy. This advancement in algorithmic performance substantiates the effectiveness of the proposed STAR-RIS system with SWIPT and RSMA, positioning it as a promising solution for the resource-efficient, high-performance demands of 6G networks.
- 23 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 24 (4) , 3209-3224
Energy Efficient Design of Active STAR-RIS-Aided SWIPT Systems
In this paper, we consider the downlink transmission of a multi-antenna base station (BS) supported by an active simultaneously transmitting and reconfigurable intelligent surface (STAR-RIS) to serve single-antenna users via simultaneous wireless information and power transfer (SWIPT). In this context, we formulate an energy efficiency maximisation problem that jointly optimises the gain, element selection and phase shift matrices of the active STAR-RIS, the transmit beamforming of the BS and the power splitting ratio of the users. With respect to the highly coupled and non-convex form of this problem, an alternating optimisation solution approach is proposed, using tools from convex optimisation and reinforcement learning. Specifically, semi-definite relaxation (SDR), difference of convex functions (DC), and fractional programming techniques are employed to transform the non-convex optimisation problem into a convex form for optimising the BS beamforming vector and the power splitting ratio of the SWIPT. Then, by integrating meta-learning with the modified deep deterministic policy gradient (DDPG) and soft actor-critical (SAC) methods, a combinatorial reinforcement learning network is developed to optimise the element selection, gain and phase shift matrices of the active STAR-RIS. Our simulations show the effectiveness of the proposed resource allocation scheme. Furthermore, our proposed active STAR-RIS-based SWIPT system outperforms its passive counterpart by 57% on average.
- 23 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 24 (4) , 3209-3224
Energy Efficient Design of Active STAR-RIS-Aided SWIPT Systems
In this paper, we consider the downlink transmission of a multi-antenna base station (BS) supported by an active simultaneously transmitting and reconfigurable intelligent surface (STAR-RIS) to serve single-antenna users via simultaneous wireless information and power transfer (SWIPT). In this context, we formulate an energy efficiency maximisation problem that jointly optimises the gain, element selection and phase shift matrices of the active STAR-RIS, the transmit beamforming of the BS and the power splitting ratio of the users. With respect to the highly coupled and non-convex form of this problem, an alternating optimisation solution approach is proposed, using tools from convex optimisation and reinforcement learning. Specifically, semi-definite relaxation (SDR), difference of convex functions (DC), and fractional programming techniques are employed to transform the non-convex optimisation problem into a convex form for optimising the BS beamforming vector and the power splitting ratio of the SWIPT. Then, by integrating meta-learning with the modified deep deterministic policy gradient (DDPG) and soft actor-critical (SAC) methods, a combinatorial reinforcement learning network is developed to optimise the element selection, gain and phase shift matrices of the active STAR-RIS. Our simulations show the effectiveness of the proposed resource allocation scheme. Furthermore, our proposed active STAR-RIS-based SWIPT system outperforms its passive counterpart by 57% on average.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 14 (3) , 781-785
Meta-Learning for Resource Allocation in Uplink Multi-Active STAR-RIS-Aided NOMA System
Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) is a novel technology which enables the full-space coverage. In this letter, a multi-active STAR-RIS-aided system using non-orthogonal multiple access in an uplink transmission is considered, where the second-order reflections among multiple active STAR-RISs assist the transmission from the single-antenna users to the multi-antenna base station. Specifically, the total sum rate maximization problem is solved by jointly optimizing the active beamforming, power allocation, transmission and reflection beamforming at the active STAR-RISs, and user-active STAR-RIS assignment. To solve the non-convex optimization problem, a novel deep reinforcement learning algorithm is proposed which integrates Meta-learning and deep deterministic policy gradient (DDPG), denoted by Meta-DDPG. Numerical results reveal that our proposed Meta-DDPG algorithm outperforms the DDPG algorithm with 19 improvement, while second-order reflections among multi-active STAR-RISs provide 74:1% enhancement in the total data rate.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 14 (3) , 781-785
Meta-Learning for Resource Allocation in Uplink Multi-Active STAR-RIS-Aided NOMA System
Simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) is a novel technology which enables the full-space coverage. In this letter, a multi-active STAR-RIS-aided system using non-orthogonal multiple access in an uplink transmission is considered, where the second-order reflections among multiple active STAR-RISs assist the transmission from the single-antenna users to the multi-antenna base station. Specifically, the total sum rate maximization problem is solved by jointly optimizing the active beamforming, power allocation, transmission and reflection beamforming at the active STAR-RISs, and user-active STAR-RIS assignment. To solve the non-convex optimization problem, a novel deep reinforcement learning algorithm is proposed which integrates Meta-learning and deep deterministic policy gradient (DDPG), denoted by Meta-DDPG. Numerical results reveal that our proposed Meta-DDPG algorithm outperforms the DDPG algorithm with 19 improvement, while second-order reflections among multi-active STAR-RISs provide 74:1% enhancement in the total data rate.