Non-terrestrial networks (NTNs) contribute significantly to offering connectivity services for the upcoming sixth generation (6G). However, many services demand real-time uplink data transmission. Owing to the dynamic topology, instability of NTN channel status, resource constraints, and tremendous mobile users on ground networks; resilient and effective resource management and channel accessibility pose substantial challenges. This work investigates the uplink sum-rate maximization and latency minimization problems with a channel allocation scheme for user equipments (UEs) to high-altitude platforms (HAPs) of a particular NTN when UEs simultaneously transmit data to HAPs and HAPs forward it to low-earth orbit satellite (LEO-sat) via multiple subcarriers. We formulate a fairness resource allocation optimization problem reflecting the system utility and load of the HAP-LEO-sat link for aerial-to-space uplinks. To tackle the high-dimensional non-convex optimization problems and short visibility of LEO-sat in the learning, we advocate a novel deep reward-based multi-agent reinforcement learning (DeepRMARL) with prioritized experience replay (PER) and priority sampling approach (PSA). Furthermore, we present a long-short-term memory (LSTM) model for future environment channel prediction due to time-varying channels between UEs and HAPs. The results explicitly unveil the feasibility of our method in optimizing total uplink rate, achieving a higher fairness index, minimizing latency by approx. 12.2%, and predicting future environment states with approx. 92.8% accuracy.

Non-terrestrial networks (NTNs) are vital in providing broad coverage and unimpeded connectivity for the forthcoming sixth-generation (6G) technology. Traditional communication methods centered on the transmission process have approached the Shannon limit; substantial data volumes and large-scale fading will lead to a tangible communication bottleneck, such as increased latency concerns. Semantic communication (SemCom), which focuses on context-related meaning, presents a new paradigm for serving forthcoming NTNs. However, the problem of efficient resource allocation throughout semantic processing and transmission remains unresolved. In this study, a resource allocation and semantic information selection method for low-earth orbit (LEO) satellite-based image transmission is investigated, and a novel semantic-level performance metric termed quality of semantic (QoSem), reflecting the inherent quality of semantic information being transmitted, is defined to measure the performance of SemCom-based NTNs. Owing to NTN’s harsh channels and constrained resources, LEO satellite can only transmit critical semantic information to each high-altitude platform (HAP) so as to meet the QoSem requirement by selecting an appropriate resource to allocate. For the purpose of maximizing the overall QoSem, the selection of semantic information and allocation of computational, transmit power, and bandwidth resources are jointly optimized. To deal with the tricky resource allocation optimization problem with non-linear programming and coupled variables, a novel deep reinforcement learning (DRL)-based intelligent resource allocation algorithm, namely the mixed-action decision intelligent semantic-aware soft-actor critic (MADIS-SAC), is proposed. The evaluation results highlight that, when compared to other baselines, our method extensively enhances QoSem under dynamic situations with severely constrained resources.

Sixth-generation (6G) networks target intelligent, context-aware communications, with Non-Terrestrial Networks (NTNs) essential for ubiquitous global coverage. However, limited bandwidth, severe path loss, massive data traffic, and dynamic, harsh channel conditions challenge both reliability and efficiency. This paper proposes a two-hop NTN downlink framework that synergistically integrates Semantic Communication (SemCom) and Bit-based Communication (BitCom) for efficient image delivery from a Low Earth Orbit (LEO) satellite to ground users via High-Altitude Platforms (HAPs). A deep learning-enabled architecture is developed, featuring a semantic-aware encoder at the LEO satellite and a hybrid semantic–bit processing unit at the HAP. Exploiting full channel state information (CSI), semantic relevance, and computational capability, the HAP dynamically categorizes users into Semantic-Aware Users (SAUs) and Bit-Based Users (BBUs), forwarding semantic features to SAUs or reconstructing bit-level data for BBUs. To holistically evaluate performance, we introduce a novel Universal Performance Metric for Hybrid Semantic and Bit-based Communication (UPM-HSBC), termed Hybrid Communication Effectiveness (HCE), which jointly captures semantic relevance, spectral efficiency, and channel dynamics. Building on this, we formulate a long-term HCE maximization problem and design a Hierarchical Multi-Agent Twin Delayed Deep Deterministic Policy Gradient for Semantic–Bit Resource Allocation (HMATD3-SBRA) algorithm. The framework jointly optimizes transmit power, bandwidth, computational resources, and mode selection across LEO satellite–HAP–ground links, while explicitly capturing inter-agent dependencies. Extensive simulations reveal that our approach achieves up to 18% higher spectral efficiency and 22% reduction in latency, while effectively reducing mode-switching overhead by 15%. It further sustains 10–25% HCE gains across diverse signal-to-noise ratio (SNR) regimes and resource constraints, validating the framework’s effectiveness for semantic-aware NTN downlinks in 6G.

The integrated usage of non-terrestrial network (NTN) nodes like high altitude platform (HAP), and low earth orbit (LEO) satellite plays a significant supplemental role in the upcoming sixth generation (6G) network to provide connectivity services for massive mobile users. Various network functions can be performed intelligently using potential artificial intelligence (AI) tools. Resilient and effective network channel accessibility and resource management, however, pose significant challenges due to the mobility, scarce resources, and complexity of integrated terrestrial networks (TNs) and NTNs, and massive mobile users on ground networks. Considering that many services demand real-time uplink (UL) data transmission, in this work, we consider the uplink of cooperated air-space segments (HAPs and LEO satellites) for large-sized data transmission where HAPs transmit data simultaneously to the LEO satellites through multiple sub-carriers using non-orthogonal multiple access (NOMA) approach, and the LEO satellite receives the superimposed signal data from HAPs. To maximize the fair allocation of resources among HAPs while reflecting the LEO satellite's load and utility, the problem is formulated as nonlinear mixed-integer programming (MINLP). Given that NTN channels are dynamic, we advocate a novel deep reward-based multiple-agent reinforcement learning for dynamic channel allocation. Furthermore, to handle the short visibility of the LEO satellites in learning, we employ a prioritized experience reply (PER) through a priority sampling approach (PSA). Numerical results reveal that our proposed approach significantly outperforms learning-based and conventional baseline models.

Non-terrestrial networks (NTNs) contribute significantly to offering connectivity services for the upcoming sixth generation (6G). However, many services demand real-time uplink data transmission. Owing to the dynamic topology, instability of NTN channel status, resource constraints, and tremendous mobile users on ground networks; resilient and effective resource management and channel accessibility pose substantial challenges. This work investigates the uplink sum-rate maximization and latency minimization problems with a channel allocation scheme for user equipments (UEs) to high-altitude platforms (HAPs) of a particular NTN when UEs simultaneously transmit data to HAPs and HAPs forward it to low-earth orbit satellite (LEO-sat) via multiple subcarriers. We formulate a fairness resource allocation optimization problem reflecting the system utility and load of the HAP-LEOsat link for aerial-to-space uplinks. To tackle the high-dimensional non-convex optimization problems and short visibility of LEO-sat in the learning, we advocate a novel deep reward-based multi-agent reinforcement learning (DeepRMARL) with prioritized experience replay (PER) and priority sampling approach (PSA). Furthermore, we present a long-short-term memory (LSTM) model for future environment channel prediction due to time-varying channels between UEs and HAPs. The results explicitly unveil the feasibility of our method in optimizing total uplink rate, achieving a higher fairness index, minimizing latency by approx. 12.2%, and predicting future environment states with approx. 92.8% accuracy.