This paper presents a comprehensive survey of semantic protocols, protocol learning, and AI-enabled semantic protocols within the context of Open RAN and 6G networks. We systematically review the significant progress achieved in these domains, highlighting key methods such as transformer-based semantic encoders, reinforcement learning–driven protocol adaptation, and federated learning frameworks for distributed training. Across surveyed studies, notable achievements include bandwidth savings of 35-70%, improved robustness under noisy conditions, and enhanced interoperability in multi-vendor environments. By consolidating findings, we identify major challenges such as the lack of standardized semantic KPIs, computational overhead at the edge, interoperability issues, and emerging security vulnerabilities. Furthermore, we categorize open research opportunities into theoretical, methodological, technical, and implementation directions, providing a clear roadmap for future development. This survey ultimately positions semantic communication and AI-enabled protocols as pivotal enablers for meaning-centric, adaptive, and efficient next-generation O-RAN/6G networks.

Intrusion detection systems (IDSs) are critical for identifying malicious activity in computer networks; however, the evaluation of machine learning (ML)–based IDS remains inconsistent and fragmented. Many existing studies rely on outdated datasets, neglect computational complexity, or use limited performance metrics. Additionally, few works leverage the full potential of modern graphics processing unit (GPU) acceleration. The objective of this study is to establish a scalable, reproducible, and standardized benchmarking framework for intrusion detection. We present an end‐to‐end, GPU‐accelerated pipeline that integrates automated data preprocessing, intrinsic dataset complexity analysis, and multiobjective hyperparameter optimization (HPO) across more than 70 publicly available datasets. Our numerical findings demonstrate that stratified sampling rates of 10% are sufficient to maintain statistical signal integrity, with class probability deviations remaining below 0.01 relative to the full population. Furthermore, feature‐reduced configurations decrease the model size by a median of 60% while maintaining weighted F 1 scores within 0.01 of the baseline. Finally, experimental complexity analysis reveals that the GPU‐accelerated modeling stages achieve empirical time‐invariance ( O (1)), reducing training latency by up to two orders of magnitude compared with traditional central processing unit (CPU) workflows. These contributions offer a rigorous quantitative view of the performance‐efficiency trade‐offs essential for next‐generation IDS evaluation.

This paper introduces two comprehensive composite fading models: the α-FMixture and the α-GMixture. These models unify a broad spectrum of previously reported composite fading distributions and enable the systematic construction of hundreds of new ones. Their first-order statistics (namely, probability density function (PDF), cumulative distribution function (CDF), moment-generating function, and higher-order moments) are derived in compact, tractable forms that, for specific cases, yield singularity-free expressions improving upon existing state-of-the-art solutions. By consolidating a wide range of multiplicative shadow (MS) and LOS shadow (LOSS) fading families into a single parameterized framework, they provide a unified analytical basis for obtaining the key statistics of numerous composite fading scenarios. Serving asplug-and-playbase models, they enable direct evaluation of metrics such as outage probability (OP) and average symbol error rate (ASER) for both established and new composite fading models without rederiving channel statistics. Their unified structure and flexible parameterization also support systematic analysis of fading and shadowing in emerging scenarios, providing practical and extensible tools for next-generation wireless channel characterization.

Fluid antenna systems (FASs) offer a promising solution to enhance communication reliability by dynamically exploiting spatial diversity. Although FAS has shown potential in static scenarios, its application to vehicular communication has not yet been explored. In this context, this paper is the first to investigate channel estimation for vehicular communication systems employing FAS, specifically based on the IEEE 802.11p standard. To this end, a port selection method is proposed, leveraging the preamble-based information to identify the optimal FAS port for efficient reception. Conventional estimation and deep learning (DL)-based approaches are compared to evaluate their performance in providing accurate channel estimates under realistic vehicular conditions, including nonlinear effects. Our results present the superior performance of FAS-aided communication systems over single-antenna ones, highlighting their adaptability to dynamic environments and varying system and channel conditions. In particular, the proposed system achieves a gain of at least 8 dB compared to single-antenna transmission. To further enhance reception reliability in high-mobility scenarios, a maximum ratio activated-port combination method is explored, providing an additional gain of at least 15 dB using only two activated ports.

Automated modulation classification (AMC) is crucial in electronic warfare because it enhances situational awareness and prompts responses to hostile transmissions. This research addresses the challenges of deploying AMC in limited computational environments by proposing using response-based knowledge distillation (KD) to train compact yet accurate AMC models. The methodology involves a framework that integrates a signal-based convolutional neural network (SBCNN) and an image-based convolutional neural network (IBCNN). The SBCNN extracts features from preprocessed signal data, which is subsequently used to train the IBCNN. Experimental results indicate that the SBCNN-based approach, when trained with teacher distillation, achieves superior performance compared to its standalone counterpart. Our findings show that KD has significant potential to enhance AMC performance in real-time applications by balancing computational demands with classification accuracy.

Handling class imbalance is a fundamental challenge in supervised learning, particularly in real‐world scenarios where minority classes are critical yet underrepresented. This paper presents a novel dynamic‐balancing pipeline that enhances automated machine learning (AutoML) performance on imbalanced tabular datasets. The proposed approach integrates both traditional and generative resampling techniques with adaptive, class‐specific thresholds, enabling automated and dataset‐sensitive balancing strategies. To assess its generalizability, the pipeline is applied uniformly across binary, multiclass, and multilabel classification tasks. Each configuration is evaluated within an AutoML framework using performance and efficiency metrics, with outcomes validated through statistical testing and effect size analysis. The study also incorporates dataset complexity measures—including feature‐label dependency and class overlap—to investigate how structural characteristics affect balancing efficacy. By combining principled resampling, exhaustive grid search, and rigorous evaluation, the pipeline enables more robust and efficient AutoML workflows. This work contributes a flexible and reproducible framework for addressing class imbalance, particularly in multilabel contexts, and establishes a foundation for scalable, complexity‐aware resampling in automated model development.

In conventional federated learning (FL), multiple edge devices holding local data jointly train a machine learning model by communicating learning updates with a centralized aggregator without exchanging their data samples. Owing to the communication and computation bottleneck at the centralized aggregator and inaccurate learning model caused by the non-independent and identically distributed (IID) data, we here consider a two-tier FL network, in which Internet of Things (IoT) nodes are the core clients that hold data, the model aggregators at the middle tier are the low altitude aerial platforms (UAVs), and the model aggregator at the top-most layer is the high-altitude aerial platform (UAV with relatively high altitude). Under the assumption that each IoT node has parallel computing ability, we study the energy-efficient computation and communication resource allocation in such a network within some time budget. Upon formulating the problem as an optimization problem, we solve the computation and communication resource allocation problems as the separate subproblems within a time frame, and then propose an iterative algorithm to solve the entire problem jointly. More specifically, we solve both the energy-efficient computation and communication resource allocation subproblems using the dual decomposition technique, and then apply a bisection search-based recursive technique to solve the entire energy efficiency problem jointly. Moreover, we propose offline and online client scheduling schemes that not only select the optimal edge nodes for association but also assign workload to each client based on the data quality and workload constraint. With real data, extensive simulations are conducted to verify the effectiveness of the proposed resource allocation scheme. The results further reveal that the learning performance not only is dependent on the computation and communication energy consumption of the FL process but also the model divergence weight owing to the non-IID data at client IoT nodes.

Spectrum scarcity has been driving cellular operators to utilize unlicensed spectrum in conjunction with licensed bands to deliver mobile data to its Long-Term Evolution (LTE) users, offloading the fully allocated LTE bands. However, the use of LTE in unlicensed spectrum creates numerous challenges as the fair coexistence with other technologies. A myriad of experimental works tackles the problems involved in the coexistence of different radio access technologies (RAT) in unlicensed spectrum, however, they do not cover all aspects of the problem and fail to provide the framework adopted in the experiments for reproducible research. Therefore, in this demo we present a highly configurable packetized-LTE PHY open-source framework for coexistence experiments. The framework allows the evaluation and comparison of different coexistence techniques.

Narrowband interference (NBI) is a known problem in wireless communications. However, analyses found in the NBI literature often model the signal of interest as a generic transport channel with all subcarriers conveying information of the same kind, e.g. user plane data. By taking the analysis down to physical channels' level, this paper aims to obtain a deeper understanding of the impact exerted by NBI on both the user and control planes of Long Term Evolution (LTE) systems. First, field measurements are carried out to characterize the narrowband systems operating in the recently standardized LTE Band 31 (450-470 MHz). On the basis of these findings, requirements for NBI suppression in the LTE downlink are established. Some prominent time-frequency distributions (TFDs) are then evaluated via computer simulations with respect to their ability to fulfill such requirements. Among the TFDs considered, wavelets offer the best compromise in terms of complexity and signal distortion regardless the type of physical channel. For those cases where information about the NBI center frequencies is not available a priori, we propose a blind timedomain canceller based on the Wigner-Ville distribution.

In this paper we present a new approach to code block segmentation used on the 3GPP Standard LTE-Advanced channel coding physical layer. Code block segmentation is a generic procedure commonly applied before turbo encoding whose sole function is to fragment a large transport block into smaller code blocks, reducing memory requirements of the turbo code interleaver. The main result of this paper is a speedup of 80 times over the original procedure defined by the 3GPP Std. when implemented in a DSP architecture.

In our previous work, we presented a brief analysis of the performance of the code block segmentation procedure adopted by the 3GPP LTE Advanced Standard as part of its physical layer channel coding scheme. Here, an in-depth analysis of its performance is offered together with a new approach to the LTE-Advanced code block segmentation. Code block segmentation is a generic procedure applied before turbo encoding whose function is to fragment a large transport block into smaller code blocks, reducing memory requirements. Analysis showed that only 39% of all transport blocks need segmentation. Results based on two different architectures, one focused on a DSP and the other on an FPGA, provided a speedup of 80 times over the original procedure with a total resources count of 166 slices, maximum frequency of 392 MHz and with very low latency.

A very efficient algorithm and hardware architecture for code block segmentation used on LTE-Advanced channel coding physical layer (PHY) is presented in this paper. Code block segmentation is a generic procedure which is commonly applied before turbo encoding. Its main function is to fragment a large transport block into smaller code blocks. This approach reduces memory requirements of the turbo code interleaver, without compromising its coding gain, since turbo encoder improves its performance as the size of the code block increases. The current work presents not only an optimized procedure with reduced computational complexity, but also an architecture with very low resource count, regarding ASIC or FPGA implementations, performing at a maximum frequency of 351 MHz on a FPGA architecture.