The increasing complexity of Beyond 5G and 6G networks necessitates new paradigms for autonomy and assur- ance. Traditional O-RAN control loops rely heavily on RIC- based orchestration, which centralizes intelligence and exposes the system to risks such as policy conflicts, data drift, and unsafe actions under unforeseen conditions. In this work, we argue that the future of autonomous networks lies in a multi-agentic architecture, where specialized agents collaborate to perform data collection, model training, prediction, policy generation, verification, deployment, and assurance. By replacing tightly- coupled centralized RIC-based workflows with distributed agents, the framework achieves autonomy, resilience, explainability, and system-wide safety. To substantiate this vision, we design and evaluate a traffic steering use case under surge and drift conditions. Results across four KPIs: RRC connected users, IP throughput, PRB utilization, and SINR, demonstrate that a naive predictor-driven deployment improves local KPIs but destabilizes neighbors, whereas the agentic system blocks unsafe policies, preserving global network health. This study highlights multi- agent architectures as a credible foundation for trustworthy AI- driven autonomy in next-generation RANs.

Rapid advancement of radar technology has led to the emergence of cognitive radar systems, which utilize adaptive mechanisms to optimize performance in dynamic environments. This paper explores the integration of cognitive adaptability into smartphone-based Impulse Radio Ultra-Wideband (IRUWB) radar systems. By dynamically modifying the radar’s operational parameters based on real-time output analysis, we aim to address the limitations of current smartphone radar implementations, including high power consumption, static radar configurations, and the inherent mobility of smartphones. Our proposed Cognitive-Adaptive IR-UWB Radar (CAIR) system improves accuracy, power efficiency, and responsiveness, enabling effective target detection, target classification, gesture recognition, distance estimation, and vital sign monitoring in diverse scenarios. By incorporating cognitive radar principles, we present a novel approach to overcoming the challenges of varying environmental conditions and user contexts, ultimately delivering a more robust and versatile user experience. This paper outlines the CAIR architecture, algorithmic design, and adaptive control mechanisms, showcasing its potential to enhance smartphone radar sensing. When tested against the major smartphone use cases, our system improves accuracy by up to 11.5%, while achieving cognitive adaptability of up to 90%. Additionally, the Artificial Neural Network (ANN)-based cognitive model achieves an accuracy of 95% and an F1-score of 94%.

Accurate channel estimation is essential for reliable downlink synchronization (DLSync) in 5G and beyond wireless systems, especially during the initial access (IA) phase. This work focuses on synchronization signal blocks (SSBs), which play a crucial role in delivering system information from the base station (gNodeB) to user equipment (UE). We propose a deep learning-based approach using an inception-style neural network to estimate the channel across the SSB time-frequency grid. Our model outperforms traditional techniques such as demodulation reference signal (DMRS) interpolation and least squares (LS) estimation, especially under practical wireless conditions like multipath delay spread and Doppler shift. The proposed model achieves a bit error rate (BER) of 10−4at an SNR of 20 dB, significantly better than the 3 × 10−3BER of conventional methods. Moreover, we have observed 4-5 dB gain at high SNR with respect to LMMSE and LS estimators. These results demonstrate that our model offers more reliable and energy-efficient synchronization, even in challenging real-world environments.

Call muting-unexpected silences during voice calls due to extended RTP packet loss is a major challenge in high-mobility 5G environments, severely degrading Mean Opinion Score (MOS) and user experience. We propose DRLCQ, a Deep Reinforcement Learning-based framework that dynamically tunes Cell Individual Offset (CIO) in real time to reduce mute events and enhance voice quality. Integrated as an xApp within the O-RAN Near-RT RIC, DRLCQ leverages live network KPIs (e.g., SINR, jitter, packet loss) to learn optimal handover decisions. Evaluated against static and heuristic baselines, DRLCQ achieves over 20% fewer call mute incidents and up to 85% higher MOS, demonstrating a scalable and intelligent solution for AI-native RAN control.

This paper introduces the Next-Generation Mobile Hotspot (NGMHS), a breakthrough solution for optimizing 5G mobile hotspot performance through intelligent, AI-powered traffic management and user-focused controls. Leveraging extended Berkeley Packet Filter (eBPF) technology, NGMHS efficiently monitors per-client traffic, reducing processing overhead while providing precise real-time control. Central to NGMHS is the AI-based Network Service Detector (NSD+), which dynamically prioritizes real-time traffic, significantly enhancing video call bitrates and reducing latency for gaming applications. Our evaluations on the Samsung A54 and Galaxy S24 devices demonstrate marked improvements in user experience, with innovations such as privacy-focused OTP user-profiles and granular traffic insights. NGMHS represents a pioneering step forward in 5G connectivity, offering a seamless, secure, and highly customizable mobile hotspot experience. NGMHS is successfully deployed across Samsung's A, M, S, Fold, and Flip series models, where increased user engagement and consistent performance improvements have been observed post-deployment.

This paper proposes a Generative AI-based Digital Twin (GDT) platform for automated network feature performance evaluation, designed for Beyond 5G (B5G) networks. The platform addresses the inefficiencies of manual evaluation by utilizing a conditional Generative Adversarial Network (cGAN) to simulate network performance based on historical data and new AI/ML features. The Network GDT integrates a novel Digital Twin Augmenting Condition (DTAC) framework, allowing for real-time simulation and performance evaluation of network features. This system significantly reduces the time and cost associated with manual evaluations, improves decision-making, and optimizes Quality of Service (QoS) and Quality of Experience (QoE). The cGAN-based model dynamically generates synthetic data, enabling comprehensive performance insights and proactive AI solution testing under various network scenarios. Experimental results demonstrate high prediction accuracy for congestion use case, validating the robustness of the proposed system. The platform's dual-phase strategy ensures that AI-based solutions are rigorously tested in simulated environments before deployment in real networks, minimizing risks and enhancing stability. This approach provides a scalable and efficient solution for future B5G networks, paving the way for more reliable and optimized wireless communication systems.

Operating in high-frequency bands such as mmWave and Terahertz poses challenges due to frequent variations in channel quality. These fluctuations impact the radio protocol stack, increasing latency and reducing throughput. Existing transport layer protocols need help to adapt to the high variability of link quality and network capacity, leading to the under-utilization of resources. The absence of radio link information further hinders the transport layer's ability to handle dynamic channel conditions. This paper presents Machine Learning-based Cross Layer Improvement (ML-CLI) of the transport layer, a novel solution designed to address the challenges posed by dynamic link variations in high-frequency bands. ML-CLI leverages real-time wireless network quality estimation to optimize the transport layer for an enhanced quality of service (QoS). Various ML and deep learning models for link quality prediction are evaluated, with the Artificial Neural Network (ANN) model emerging as the top-performing model, achieving an accuracy of 98.1% and an F1-score of 0.98. The integration of ML-CLI into the ns3 simulator enables the assessment of its impact on the transport layer. The results demonstrate substantial goodput and packet loss ratio improvements, with ML-CLI providing faster recovery and improved congestion control. Notably, ML-CLI achieves a 45.22% improvement in goodput and up to a 38.52% reduction in packet loss ratio compared to the traditional TCP Cubic variant.

The evolution of telecommunications technology is at the cusp of a major transition from 5G to Beyond 5G networks. With this imminent shift, the demand for robust and efficient mitigation solutions has become increasingly vital. AI-based mitigation solutions for solving B5G network problems are directly pushed into the actual field or manually evaluated by the operator first. With Big Data involved in 5G and Beyond, evaluating them manually or without evaluation, pushing them into the real field might have severe consequences in the actual network. There is no intelligent and proactive platform to test the implications of ML models on the networks. In this paper, we propose a pioneering approach that involves the development of an AI-based 5G network clone to serve as a performance evaluation ground for AI-based mitigation solutions tailored for B5G networks. Our methodology outlines the initial phase of evaluating these mitigation solutions within the simulated environment of the AI-based 5G network clone, followed by their subsequent deployment in real-world network infrastructures. This strategy aims to ascertain the efficacy, reliability, and adaptability of the proposed solutions before their integration into the next-generation B5G networks.

Wireless 5G and beyond (B5G) technology offers multiple of machine learning (ML) use cases, including congestion detection, handover prediction, MAC scheduling and more. Many of these use cases involve solving complex problems that need neural networks (NN) or classical ML algorithms to achieve optimal solutions. However, implementing these NN inferences on resource-constraint base stations (BS) pose significant challenges. BSs has several limitations in terms of CPU frequency, number of cores, memory capacity, and the absence of dedicated ML hardware (HW) offloads. In this paper, the authors, propose a lightweight artificial intelligence (LWAI) method to derive computationally efficient mathematical relations (EMR) between Key Performance Indicators (KPIs) using reinforcement learning (RL). The derived EMR enables ML inference to be implemented on resource-constrained BSs. LWAI takes KPIs information as input and provides EMRs as output. The generated EMRs are then deployed on the BS. Inference is made to the BS using these relations. This approach makes ML inference realizable on embedded BSs with commercial-grade accuracy and optimal real-time prediction latency. We demonstrate the effectiveness and applicability of the LWAI framework in two real-world scenarios. The results highlight the potential for our approach to detect congestion by predicting physical resource block (PRB) and energy savings in BS. Our results show the efficacy of EMRs. LWAI framework-derived EMRs consume around 10% CPU cycles and 5% memory to execute as compared to NN models while maintaining over 90% prediction accuracy. Our approach opens up new possibilities for realizing ML inference ideas on resource-constraint embedded devices.

This research paper presents a novel framework designed to automate the provisioning of ML services, intelligently tailoring the ML package based on various factors such as service profiles, regional resource usage patterns, operator-defined KPIs, and current ML resource utilization in the network. Our proposed framework employs dynamic and automatic cell grouping techniques using similarity metric correlation algorithms across Base Stations (BS). It selectively trains a representative cell within each group using the best available Machine Learning (ML) model automatically determined. The trained model of the representative cell is subsequently applied to the remaining BS within the same group. To evaluate the effectiveness of our solution, we conducted extensive evaluations using real-world operator data from 5G networks, encompassing a wide range of network KPIs. The results demonstrate the remarkable impact of our framework, showcasing substantial resource savings in terms of ML Server Processing time, memory consumed, and server util percentage. Furthermore, our approach significantly reduces the number of ML trainings required, all while maintaining high ML prediction accuracies. On average, our solution achieves an impressive 39.94% reduction in ML server processing time, a substantial 60.46% reduction in ML server memory, a remarkable 75.11% reduction in server util percentages, and a total of 649 fewer ML trainings for 5G operator data across various network KPIs. These achievements highlight the efficacy of our framework in optimizing resource allocation without compromising the accuracy of ML predictions.

Identification of offline and online handwritten words is a challenging and complex task. In comparison to Latin and Oriental scripts, the research and study of handwriting recognition at word level in Indic scripts is at its initial phases. The two main methods of handwriting recognition are global and analytical. The present work introduces a novel analytical approach for online handwritten Gurmukhi word recognition based on a minimal set of words and recognizes an input Gurmukhi word as a sequence of characters. We employed a sequential step-by-step approach to recognize online handwritten Gurmukhi words. Considering the massive variability in online Gurmukhi handwriting, the present work employs the completely linked non-homogeneous hidden Markov model. In the present study, we considered the dependent, major-dependent, and super-dependent nature of strokes to form Gurmukhi characters in words. On test sets of online handwritten Gurmukhi datasets, the word-level accuracy rates are 85.98%, 84.80%, 82.40%, and 82.20% in four different modes. Besides the online Gurmukhi word recognition, the present work also provides Gurmukhi handwriting analysis study for varying writing styles and proposes novel techniques for zone detection and rearrangement of strokes. Our proposed algorithms have been successfully employed to online handwritten Gurmukhi word recognition in dependent and independent modes of handwriting.

In the past few decades, many Transmission Control Protocol (TCP) congestion control algorithms have been investigated to meet the growing network demands and to enhance the performance of TCP in lossy or high-bandwidth-delay-product (high-BDP) networks. However, it is still challenging to implement a dynamic congestion control algorithm for wide range of diverse mobile users, network conditions and applications. This paper explores avenues for enhancement of TCP congestion control algorithm for next generation mobile networks by dynamically learning the available bandwidth and deriving the Congestion Control Factor N. N is used to Adaptive Increase/Adaptive Decrease (AIAD) the Congestion Window (CWND) dynamically instead of using the traditional approach that is Additive Increase/ Multiplicative Decrease (AIMD) paradigm. Once there is congestion, our proposed algorithm will not allow the CWND to decrease multiplicatively or steeply. After dropping to a certain level (lower than legacy), we try to take the CWND to previous state adaptively with the help of calculated bandwidth (based on learning). This in turn helps to efficiently control the CWND for better network utilization especially in case of lossy and high-BDP conditions. As soon as it reaches the original state, it remains stable for longer time as compared to legacy until packet loss or time out. We demonstrate the effectiveness of our algorithm with the help of live air experiments (performed in Samsung R&D India, Bangalore) and NS3 based simulation experiments. Through our experiments, we show that our algorithm outperforms the legacy congestion control algorithms (like CUBIC, RENO, TCPW) and the existing CLTCP algorithm in terms of goodput, intra algorithm fairness and inter algorithm fairness maintaining the scalability and friendliness.

5G is one of the most sought after technology for supporting massive connectivity, reduced latency, higher throughput, D2D communication, Dual Connectivity, LTE-Wifi aggregation and many other services. Multi-Radio Access Technology (Multi-RATs) carrier aggregation (CA), also known as multi-flow CA allows different RATs to be aggregated and allocated to the UE. So, an optimized Multi-RAT support is very much desired in 5G Environment. This paper covers the 5G architecture facilitating an optimized plug and play model for providing dedicated Multi-RAT services through simulation results and studies.

Emerging 5G wireless networks are expected to herald significant transformation in industrial applications, with improved coverage, high data rates, and massive device capacity. However, the introduction of 5G wireless makes the network configuration, management, and planning extremely challenging. For efficient network configuration, every cell needs to be allocated a particular Physical Cell Identifier (PCID), which is unique in its vicinity. Wireless standards (e.g., 3GPP) typically specify a limited number of PCIDs. However, the number of cells in 5G Ultradense Networks (UDN) is expected to significantly outnumber these limited PCIDs. Hence, these PCIDs need to be efficiently allocated among the myriad of cells, such that two cells which are neighbors or neighbor’s neighbor are assigned with different PCIDs. This complicated network configuration problem becomes even more complex by dynamic introduction and removal of 5G small cells (e.g., micro, femto, and pico). In this paper, we introduce BiSON, a new Bioinspired Self‐Organizing Solution for automated and efficient PCID configuration in 5G UDN. Using two different extensions, namely, “always near‐optimal” and “heuristic,” we explain near‐optimal and dynamic auto‐configuration in computationally feasible time, with negligible overhead. Our extensive network simulation experiments, based on actual 5G wireless trials, demonstrate that the proposed algorithm achieves better optimality (minimum PCIDs in use) than earlier works in a reasonable computational complexity.

The online handwriting data are an integral part of data analysis and classification research, as collected handwritten data offers many challenges to group handwritten stroke classes. The present work has been done for grouping handwritten strokes from the Indic script Gurmukhi. Gurmukhi is the script of the popular and widely spoken language Punjabi. The present work includes development of the dataset of Gurmukhi words in the context of online handwriting recognition for real-life use applications, such as maps navigation. We have collected the data of 100 writers from the largest cities in the Punjab region. The writers’ variations, such as writing skill level (beginner, moderate, and expert), gender, right or left handedness, and their adaptability to digital handwriting, have been considered in dataset development. We have introduced a novel technique to form handwritten stroke classes based on a limited set of words. The presence of all alphabets including vowels of Gurmukhi script has been considered before selection of a word. The developed dataset includes 39,411 strokes from handwritten words and forms 72 classes of strokes after using a k-means clustering technique and manual verification through expert and moderate writers. We have achieved recognition results using the Hidden Markov Model as 87.10%, 85.43%, and 84.33% for middle zone strokes when using training data as 66%, 50%, and 80% of the developed dataset. The present work is a step in a direction to find groups for unknown handwriting strokes with reasonably higher levels of accuracy.