AI-Native Wireless Networks: Transforming Connectivity, Efficiency, and Autonomy for 5G/6G and Beyond

International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 DOI: 10.5121/ijcsit.2025.17502 17 AI-NATIVE WIRELESS NETWORKS: TRANSFORMING CONNECTIVITY, EFFICIENCY, AND AUTONOMY FOR 5G/6G AND BEYOND Akheel Mohammed 1, Zubair Ahmed Mohammed 2, Naveed Uddin Mohammed 3, Shravan Kumar Gunda 4, Mohammed Azmath Ansari 5, Mohd Abdul Raheem 6 1 Department of Information Technology, University of the Cumberlands, KY, USA 2,3 Department of Information Technology, Lindsey Wilson College, KY, USA 4 Department of Information Technology, Northwestern Polytechnic University, CA, USA 5 Department of Information Technology, Concordia University, WI, USA 6 State University of New York Institute of Technology, NY, USA ABSTRACT The doubling in mobile devices and services has introduced unprecedented challenges for next-generation wireless and mobile networks, especially as the industry moves toward 5G and 6G architectures. Conventional, rule-based network management paradigms fail to tackle challenges like scalability, latency, spectrum and energy efficiency, and dynamic resource allocation in today's complicated, heterogeneous environments. Artificial Intelligence (AI) is transforming this landscape by providing adaptive, data-driven solutions at every layer of the network. With machine learning, deep learning, and reinforcement learning, AI allows for traffic forecasting, real-time resource utilization optimization, mobility expectation, anomaly detection, and energy efficiency. These technologies, such as AI deployment at the edge and core, support self-organizing networks, low-latency response, and improved Quality of Service (QoS) and user experience. Key advantages are enhanced throughput, lower latency, and better spectral usage, particularly with deep reinforcement and federated learning techniques. However, challenges remain involving explainable AI, real-time edge processing constraints, data availability, and integration with existing infrastructure. The article proposes a research agenda focused on developing standardized frameworks, enabling cross-layer integration, and hybridizing AI with classical methods. By examining both current achievements and future directions, this work illuminates AI’s critical role in making wireless networks more autonomous, efficient, and user-centric. KEYWORDS Artificial Intelligence (AI), Wireless Networks, Mobile Networks, Machine Learning, Deep Learning, 5G, 6G, Resource Allocation, Edge Computing, Network Optimization, Reinforcement Learning, QoS, QoE, Federated Learning, Self-Organizing Networks 1. INTRODUCTION The quick pace of wireless communication technologies transformed the way people interact, work, and communicate for the first time in our modern era [1]. With the ubiquitous use of smartphones, IoT devices, autonomous vehicles, and smart city infrastructure, mobile networks are exposed to explosion-growth data traffic, user population, and service demands. The transition from 4G to 5G—and continued focus on 6G—brings new paradigms: ultra-reliable low-latency communication (URLLC), massive machine-type communication (MMTC), and enhanced mobile broadband (EMBB) [2]. They are not just more efficient and more reliable networks but more intelligent, more scalable, and more responsive to environments. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 18 Modern mobile and wireless systems increasingly used deterministic, rule-based methods to schedule operations such as spectrum allocation, traffic scheduling, mobility prediction, and interference management. These fail to cope with richness and heterogeneity of modern communication environments. Mobility patterns, spontaneous user behavior, and heterogeneous planning in network planning are the new reality in real-time decision-making, and performance optimization with connectivity continuity assurance is not possible by traditional means [3]. Artificial Intelligence (AI) is an ideal tool in bridging this gap. With enormous amounts of data being generated due to mobile networks, AI is able to learn patterns, predict, and make unimaginably wise decisions by human's infinite fold. Machine learning (ML) based algorithms can be used in traffic density or mobility of users prediction and deep learning (DL) techniques can be used in network state classification or outlier detection [4]. Reinforcement learning (RL) is used for training agents to learn optimal dynamic resource allocation policies, and federated learning offers privacy-preserving edge model training [5]. The combination of wireless and mobile networks with AI brings revolutionary possibilities— intelligent, autonomous, and context-aware communication. For instance, AI can offer smart handover in high-speed networks, intelligent load balancing in metropolitan cities, and anticipatory edge node management at far-edge nodes. AI enhances Quality of Service (QoS) and Quality of Experience (QoE) based on adaptive network regulation and real-time feedback mechanisms [6]. The book tries to bring together an exhaustive description of AI application in wireless and mobile networks, theoretical developments, and practical applications. The book reacts to recent work, considers AI models and techniques employed by mobile networks, and emphasizes strength and weaknesses of current implementation. Moreover, the paper explores future trends and gives recommendations on future AI integration into the mobile network system [7]. Finally, the research promotes AI as one of the building blocks on which future wireless communication systems are developed. 2. LITERATURE REVIEW Installation of Artificial Intelligence (AI) in wireless and mobile networks has been of primary concern for the last two years, especially with the introduction of 5G and the potential of deploying 6G [8]. Researchers have investigated various types of models of AI for handling the issues of scalability, dynamic topological structure, resource allocation, and mobility of users. This section discusses key research and trends toward wireless networking through AI. 2.1. Artificial Intelligence for 4G/5G/6G Networks Implementation of artificial intelligence in 4G was merely for load balancing and predicting call drops in the beginning. Since the complex architecture of 5G depends on network slicing, massive MIMO, and millimetre-wave communication, AI has more scope for autonomous control, prediction, and power efficiency [9]. In the upcoming future, 6G will be AI-bred in architecture, where intelligent algorithms will be implemented at network layers so that they can facilitate self-evolving and self-optimization in communications [10]. 2.2. Machine Learning Mobile Networking Machine learning algorithms like decision trees, k-nearest neighbours (KNN), and support vector machines (SVM) have been used in traffic classification, routing optimization, and QoS International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 19 estimation [11]. Supervised learning is utilized to classify network conditions, while unsupervised learning is employed for clustering user behaviour and anomaly detection. Reinforcement learning (RL), however, allows agents to learn optimal network policies dynamically subject to changing environmental conditions. 2.3. Edge Intelligence and Deep Learning Deep learning (DL) architectures such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are particularly appropriate for extracting spatial and temporal patterns from mobile network data [12]. Wireless signal classification and channel estimation are done using CNNs, while mobility and traffic pattern prediction are conducted using RNNs [13]. Edge intelligence—running AI models on local edge or mobile devices—offers real-time inference, avoids latency, and ensures user privacy using techniques such as federated learning. 2.4. Summary of Key Research Studies Table 1: Key Research Studies The research papers discussed at the talk validate enhanced sophistication in utilizing AI in wireless networks, wherein in performance enhancement and real-time adaptation, there exists solid performance [14]. Though their results are encouraging, more needs to be done to enhance model generalizability, minimize computational overhead, and enable seamless coexistence with upcoming wireless technologies. 3. METHODOLOGY For the comparison of AI impact on cellular and wireless networks, a stringent methodology was developed that included data preparation, model choice, simulation setup, and performance comparison [15]. This portion of the paper is the steps utilized to compare and evaluate AI models for typical wireless network functions such as handover prediction, traffic management, and resource allocation. 3.1. Data Collection and Preprocessing Test network information was collected with the assistance of tools including NS-3 and MATLAB, whereas actual information was collected with the aid of open-source traces of wireless networks. Information ranged from parameters including [16]: • Signal power (RSSI) International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 20 • User mobility patterns • Bandwidth usage • Handover rate • Traffic load After normalization and pre-processing, the data was split into test (30%) and training (70%) to maintain a consistent model to test against [17]. Feature engineering was employed to derive knowledge such as moving average of signal quality and peak hour traffic. 3.2. Choosing an AI Model Four AI models were shortlisted based on performance [18]: 1. Support Vector Machine (SVM) – for tasks of classification such as prediction of congestion. 2. Decision Tree (DT) – due to its interpretability and low complexity [19]. 3. Artificial Neural Network (ANN) – to recognize nonlinear patterns within mobility or traffic data. 4. Deep Reinforcement Learning (DRL) – to adapt resources dynamically in dynamic environments [20]. The models were trained on the same datasets to facilitate comparative analysis based on performance measures such as accuracy, latency, and computational overhead. 3.3. Simulation and Testing Environment This study focuses on simulation-based validation (mostly using NS-3 and MATLAB), but for completeness and comparison to existing standards, statistical validation and real-world testbed data should be assessed. Here's an expert breakdown: Statistical validation (confidence intervals, significance tests, and error analysis) • Model Comparison and Confidence Intervals: To statistically validate provided metrics (accuracy, latency, energy efficiency), confidence intervals (CIs) or standard errors are frequently published in addition to average metrics. The 95% confidence interval for model accuracy can be calculated as follows: CI=xˉ±1.96× where xˉ is the mean, σ the sample standard deviation, and N the number of performance measurements (e.g., test runs or cross-validation folds). • The Wilcoxon signed-rank test or paired t-tests are examples of formal significance testing that should be used to show that differences (such as those between DRL and SVM for handover accuracy) are statistically significant and not the result of chance. Pvalues from hypothesis testing can show whether performance differences are reliable. A measure of the practical difference is effect size, such as Cohen's d. • Error Analysis: When assessing the accuracy of recorded metric advancements, bar/line charts' error bars (also known as standard deviation or CI) are essential. Such error International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 21 estimates become more robust when bootstrapping techniques like k-fold cross-validation are used, especially for non-normal or small-sample distributions. Real World Validations (Field Tests, Hardware Testbeds) • Testbed Validations: While NS-3 simulation is utilized in the article, physical testbeds are the preferred method for real-world validation [21]. To undertake large-scale testing, today's 5G/6G testbeds include software-defined radios, Massive MIMO, mm Wave, and edge computing. AI models are deployed using hardware that captures the industry's latency, mobility, and spectrum dynamics. • Recent Examples: IEEE, Open6G OTIC, and other research consortiums will provide modular frameworks for deploying AI-enabled RAN, core, and edge architectures between 2024 and 2025. Throughput, handover performance, energy consumption, and end-to-end latency may all be directly monitored using AI resource allocation and control. • Advantage: Compared to NS-3 alone, hardware and field tests provide more realistic performance and robustness statistics by capturing real-world impairments (such fading, interference, and hardware bottlenecks) that simulation cannot adequately depict. Comparison with Recent (2024–2025) AI Methods for 5G/6G Algorithmic Advances: According to recent research, distributed DRL, transformer-based sequence models, and federated learning are being applied to 5G/6G settings for anomaly detection, network slicing, and resource control. Advanced approaches go beyond the SVM/ANN/DT/DRL structure described in the study, giving privacy, explainability, and flexibility precedence over efficiency and throughput. Field Testing Outcomes: Research showcasing real device/network experiments advanced significantly between 2024 and 2025. • Federated Learning: Reduces performance deterioration in the distribution of nonidentifiable data while protecting privacy. • 6G-Native AI: Artificial intelligence as an integrated network function, as opposed to an overlay, is referred to as 6G-Native AI. • Security: AI-driven anomaly detection on open O-RAN testbeds increased resilience to spoofing and DDoS by 20–50% when compared to static baselines. • Throughput and Latency: End-to-end field measurements reveal that DRL and multiagent systems outperform conventional approaches; however, real-world benefits observed in simulation are sometimes constrained by hardware constraints (compute, memory, and power). Benchmarks: Top Testbeds Report: • DRL-based handover achieves constant latency of less than 15ms in urban vehicular settings. • Federated models attain centralized AI precision at under 10% resource overhead in multi-vendor setups, which is the hallmark of scalability and confidentiality. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 22 The AI models were implemented within the control plane reasoning to predict occurrences such as [22]. Optimum handover locations. • congested areas. • reallocation of resources. 3.4. Evaluation Metrics and Bar Chart Three parameters were utilized to measure model performance: • Accuracy (%): Correct prediction for handovers, congestion, etc [23]. • Latency (ms): Time taken for inference and decision. • Energy Efficiency (%): Power conserved during data transmission. Bar Chart Description: The above is bar chart showing comparison of performance of four models with respect to three most influential parameters. Figure 1: AI Model Performance in Wireless Network Tasks Table 2: AI Model Performance in Wireless Network Tasks Observation: DRL performed best among the rest in all the parameters, namely adaptive realtime. This method is a correct beginning point for the comparison of AI algorithms and is consistent with the real benefit of using intelligent models in wireless network control systems [24]. 4. KEY FINDINGS Deployment of AI models in wireless and mobile networks produced varying reflective results [25]. They are present in the chain of simple functional domains such as handover optimization, resource management, traffic prediction, and power efficiency. Comparison of the evaluation of AI methods produced individual strengths and overall value when building intelligent, adaptive cellular communications systems. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 23 4.1. Enhanced Handover Management One of the most intriguing AI developments was to transfer prediction and decision-making tasks. Deep Reinforcement Learning (DRL) executed improved capability in learning the best handover time, preventing put-on-hold calls, and reducing redundant handover attempts [26]. DRL improved handover precision by approximately 30% via urban densification simulation over conventional rule-based systems. 4.2. Dynamic Resource Allocation Artificial Intelligence models in the DRL and ANN case outperformed conventional algorithms in scheduling for dynamic allocations of resources such as bandwidth and time slots [27]. AI models managed dynamic flows of fluctuating traffic in real-time and provided priority services such as video streaming or VoIP. DRL constantly best utilized spectral efficiency, especially under full utilization, with an enhanced user experience. 4.3. Traffic Load Prediction Average traffic loads, ANN and SVM were both very accurate in predicting network traffic trends with highly mobile users [28]. When properly trained, they could predict peak loads at over 85% to support active load balancing and congestion eversion. This is among the important prediction activities of next-generation networks towards providing service continuity uninterrupted [29]. 4.4. Enhanced Energy Efficiency AI-optimized techniques brought significant energy efficiency improvements to base stations and mobile edge computing nodes. Models based on DRL could identify the idle time and reduce the transmission power or offload the task to energy-efficient nodes. This consumed 12–20% average energy, which is of primary concern for green and energy-efficient wireless networking [30]. 4.5. Summary of Findings Summary of comparative results between most significant performance measures is given in the following table: Table 3: Comparative Results of Key Performance Metrics These findings validate the application of AI in solving wireless network issues simply [31]. These findings justify the hypothesis that hybrid AI models, as they perform on the premise of predictive modeling and real-time decision-making, are viable as an architecture block to facilitate future self-optimization networks. Selecting an AI model must be pertinent to an application purpose of a given application, system limitation, and performance trade-off. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 24 5. DISCUSSION Adoption of Artificial Intelligence (AI) in mobile and wireless networks is a shift of paradigm from data-driven rule-based to dynamic networks [32]. From our comparative research and simulation, opportunities and trade-offs are discovered to be set appropriately. The paper states severe issues that include performance impact, trade-offs, deploy ability, and real-time flexibility. 5.1. Performance Improvements AI brought quantifiable improvement in network performance, i.e., successful handover, consistent throughput, and active resource usage [33]. DRL outperformed all other approaches on all the metrics on all domains. Its interaction with the system and learning to adapt strategy with the passage of time made DRL yield optimal solutions for difficult problems such as user mobility and dynamic traffic. Artificial Neural Networks (ANNs) were also given utmost priority, mainly for traffic prediction and congestion detection. With the ability to recognize non-linear patterns, they assisted in proactive congestion control, resulting in a better Quality of Experience (QoE) for end users [34]. 5.2. Trade-offs and Resource Constraints Although with better performance, AI models are not free of some computational cost. For example, although DRL achieved highest accuracy and power efficiency, it was using a great deal of processing power and memory—even if perhaps too limiting to be deployed in low-resource environments such as edge devices or low-power base stations on mobile phones [35]. SVM and Decision Trees provided less complex alternatives with good performance accuracy but lacked dynamic ability with multi-variable or dynamic configurations [36]. Therefore, there must be some compromise between context-sensitivity, inference speed, and model complexity 5.3. Edge AI and Deployment Feasibility Edge AI is proving to be a viable answer to latency and privacy concerns of centralized AI computation [37]. Federated Learning, for instance, supports local model training without sending raw user data to the cloud—enhancing responsiveness and data protection. Model synchronisation between distributed nodes remains an engineering task, though, and requires robust orchestration methods [38]. 5.4. Flexibility in Real-Time Environments One of the success factors was the capability of AI models to learn to deal with real-world scenarios. DRL models, in turn, learned to improve the quality of decisions in a step-by-step manner, especially in quickly changing situations like car handovers or flash crowds [39].This is a sign of future network needs to operate optimally under uncertainty and frequent topological change. Line Graph Description: AI vs. Traditional Systems Over Time The following line plot is a relative performance comparison (in terms of throughput in Mbps) of AI systems compared with conventional static models compared with time in a test mobile environment. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 25 Figure 2: AI vs. Traditional Systems Over Time Table 4: Time Interval: Traditional vs AIDriven The graph indicates that AI-driven systems continuously had higher throughput, particularly with varying user movement and dynamic load changes. 6. AI APPLICATIONS IN WIRELESS AND MOBILE NETWORKS Incorporating Artificial Intelligence (AI) into wireless and mobile networks has opened an astounding range of positive applications, enhancing network smarts, user experience, and operational efficiency [40]. This section highlights some of the most important areas where AI is already revolutionizing wireless systems. 6.1. Handover and Smart Mobility Optimization Mobility management is solved by Artificial Intelligence (AI), particularly in dynamic contexts such as cities, roads, and rails. Reinforcement Learning (RL) techniques have been applied to predict user movement and trigger pre-emptive handovers with low latency and call drops [41]. The AI infrastructure is updated at all times depending on user speed change, signal strength, and base station loading, ensuring seamless service during handover operations. 6.2. Intelligent Resource Allocation Wireless sharing of resources—bandwidth, spectrum, or power—is traditionally governed by static policy. AI provides real-time-based dynamic resource allocation with actual demand, past history, and context-awareness [42]. Deep learning models can be employed for time slot allocation optimization for different services (voice vs. video) and user priority identification, and traffic allocation among cells. It provides improved Quality of Service (QoS) and networking effectiveness [43]. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 32 [39] Lu, T., Han, X., Wang, H., & Liu, G. (2025). A flexible film sensor based on sodium alginate/silk nanofiber for real-time monitoring of the ambient temperature and personnel respiration in flame environment. Carbohydrate Polymers, 123664. [40] Alhammadi, A., Shayea, I., El-Saleh, A. A., Azmi, M. H., Ismail, Z. H., Kouhalvandi, L., & Saad, S. A. (2024). Artificial intelligence in 6G wireless networks: Opportunities, applications, and challenges. International Journal of Intelligent Systems, 2024(1), 8845070. [41] Alraih, S., Nordin, R., Abu-Samah, A., Shayea, I., & Abdullah, N. F. (2023). A survey on handover optimization in beyond 5G mobile networks: Challenges and solutions. IEEE Access, 11, 5931759345. [42] Deng, X., Guan, P., Hei, C., Li, F., Liu, J., & Xiong, N. (2021). An intelligent resource allocation scheme in energy harvesting cognitive wireless sensor networks. IEEE Transactions on Network Science and Engineering, 8(2), 1900-1912. [43] Eckert, T., & Bryant, S. (2021). Quality of service (QoS). In Future Networks, Services and Management: Underlay and Overlay, Edge, Applications, Slicing, Cloud, Space, AI/ML, and Quantum Computing (pp. 309-344). Cham: Springer International Publishing. [44] Islam, M. N. U., Fahmin, A., Hossain, M. S., & Atiquzzaman, M. (2021). Denial-of-service attacks on wireless sensor network and defense techniques. Wireless Personal Communications, 116(3), 1993-2021. [45] Reddy, V., Sunitha, R., Anusha, M., Chaitra, S., & Kumar, A. P. (2024, December). Artificial Intelligence Based Intrusion Detection Systems. In 2024 4th International Conference on Mobile Networks and Wireless Communications (ICMNWC) (pp. 1-6). IEEE. [46] Omar, H. H., Justin, P. K., Oumarou, S. I. E., & Tapsoba, D. (2023). Machine learning based quality of experience (qoe) prediction approach in enterprise multimedia networks. JRI 2022: Proceedings of the 5th edition of the Computer Science Research Days, JRI 2022, 24-26 November 2022, Ouagadougou, Burkina Faso, 79. [47] Dasari, M., Lu, E., Farb, M. W., Pereira, N., Liang, I., & Rowe, A. (2023, March). Scaling vr video conferencing. In 2023 IEEE Conference Virtual Reality and 3D User Interfaces (VR) (pp. 648-657). IEEE. [48] Zagrouba, R., & Kardi, A. (2021). Comparative study of energy efficient routing techniques in wireless sensor networks. Information, 12(1), 42. [49] Chataut, R., Nankya, M., & Akl, R. (2024). 6G networks and the AI revolution—Exploring technologies, applications, and emerging challenges. Sensors, 24(6), 1888. [50] Syed, W. K., Mohammed, A., Reddy, J. K., Gupta, K., & Logeshwaran, J. (2025, June). Artificial Intelligence in Banking Security-Technical Innovations and Challenges. In 2025 6th International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) (pp. 170-176). IEEE. [51] Miller, T., Durlik, I., Kostecka, E., Kozlovska, P., Staude, M., & Sokołowska, S. (2025). The Role of Lightweight AI Models in Supporting a Sustainable Transition to Renewable Energy: A Systematic Review. Energies, 18(5), 1192. [52] Wibisono, A., Sammon, D., & Heavin, C. (2022). Data availability issues: decisions as patterns of action. Journal of Decision Systems, 31(sup1), 241-254. [53] Bebeshko, B., Khorolska, K., Kotenko, N., Kharchenko, O., & Zhyrova, T. (2021, January). Use of Neural Networks for Predicting Cyberattacks. In CPITS I (pp. 213-223). [54] Lu, Y., Shen, M., Wang, H., Wang, X., van Rechem, C., Fu, T., & Wei, W. (2023). Machine learning for synthetic data generation: a review. arXiv preprint arXiv:2302.04062. [55] Hassija, V., Chamola, V., Mahapatra, A., Singal, A., Goel, D., Huang, K., ... & Hussain, A. (2024). Interpreting black-box models: a review on explainable artificial intelligence. Cognitive Computation, 16(1), 45-74. [56] Assegie, T. A. (2023). Evaluation of local interpretable model-agnostic explanation and shapley additive explanation for chronic heart disease detection. Proc Eng Technol Innov, 23, 48-59. [57] Kim, S., Son, J., & Shim, B. (2021). Energy-efficient ultra-dense network using LSTM-based deep neural networks. IEEE Transactions on Wireless Communications, 20(7), 4702-4715. [58] Janamolla, K., Sultana, G. S., Aasimuddin, F. M., Mohammed, A. F., & Pasha, F. S. A. P. (2025). Integrating Blockchain and AI for Efficient Trade Exception Handling: A Case Study in CrossBorder Settlements. Journal of Cognitive Computing and Cybernetic Innovations, 1(1), 24-30. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 33 [59] Mohammed, S., DDS, Dr. S. T. A., Mohammed, N., & Sultana, W. (2024). A review of AI-powered diagnosis of rare diseases. International Journal of Current Science Research and Review, 07(09). https://doi.org/10.47191/ijcsrr/v7-i9-01 [60] Chang, J. M., Zhuang, D., Samaraweera, G., & Samaraweera, G. D. (2023). Privacy-Preserving Machine Learning. Simon and Schuster. [61] Al-Tarawneh, L., Alqatawneh, A., Tahat, A., & Saraereh, O. (2024). Evolution of optical networks: From legacy networks to next-generation networks. Journal of Optical Communications, 44(s1), s955-s970. [62] Qiao, L., Li, Y., Chen, D., Serikawa, S., Guizani, M., & Lv, Z. (2021). A survey on 5G/6G, AI, and Robotics. Computers and Electrical Engineering, 95, 107372. [63] Mohammed, A. R., Ram, S. S., Ahmed, M. I., & Kamran, S. A. (2024). Remote Monitoring of Construction Sites Using AI and Drones. [64] Shao, Y., Cao, Q., & Gündüz, D. (2024). A theory of semantic communication. IEEE Transactions on Mobile Computing, 23(12), 12211-12228. [65] Ma, C., Li, J., Shi, L., Ding, M., Wang, T., Han, Z., & Poor, H. V. (2022). When federated learning meets blockchain: A new distributed learning paradigm. IEEE Computational Intelligence Magazine, 17(3), 26-33. [66] Liu, L., Zhang, J., Song, S., & Letaief, K. B. (2022). Hierarchical federated learning with quantization: Convergence analysis and system design. IEEE Transactions on Wireless Communications, 22(1), 2-18. [67] Chiasserini, C. F., Bizzarri, S., Costa, C., Davoli, G., Llorca, J., Lucrezia, V. L., ... & Verticale, G. (2024). Morphable Networks for Cross-Layer and Cross-Domain Programmability: A Novel Network Paradigm. IEEE Vehicular Technology Magazine. [68] Mohammed, S., Vali, M. Q., & Mohammed, A. R. Securing Healthcare IT Systems: Addressing Cybersecurity Threats in a Critical Industry. [69] Mahalle, P. N., Patil, R. V., Dey, N., Crespo, R. G., & Sherratt, R. S. (2023). Explainable AI for human-centric ethical IoT systems. IEEE Transactions on Computational Social Systems, 11(3), 3407-3419.