Quantum teleportation and superdense coding are two different protocols that were developed to transmit a limited number of qubits and two classical bits, respectively. Therefore, they are inefficient for transmitting messages containing a massive number of individual qubits or classical bits. Thus, it is necessary to scale up these two protocols in larger systems to enable them to transmit massive messages. To achieve this goal, it is essential to integrate each of these two protocols within architectures that contain a quantum main memory that functions as a buffer capable of storing both quantum and classical data and managing a massive amount of data efficiently during transmission operations. Q1RAM is a cutting-edge quantum main memory and the fastest quantum main memory available, as it has a time complexity of O(1) for accessing memory locations. In this paper, two architectural schematics are proposed based on integrating Q1RAM as a core component with the quantum teleportation and superdense coding protocols. The first architectural scheme is developed to scale up the quantum teleportation protocol, making it efficient for transmitting massive messages of separable individual qubits. The second architectural schema is proposed, consisting of two Q1RAM chips, two quantum processing units, two of the classical RAMs, and a quantum gateway. This proposed schema is developed to scale up the superdense coding protocol so that it can transmit massive classical messages containing large amounts of data. The experimental simulation results validate the architectural schematics for scaling up the quantum teleportation and the superdense coding protocols for large-scale quantum communications.
We study a recently published dataset of 396 real-world cloud architectures on AWS from a wide range of industries. We focus on those architectures that contain HPC, edge or serverless components and investigate how they are built, which services they use, the types of storage used, quantify architectural complexity, assess the use of machine learning services, and distill insights regarding how cloud applications in these three domains are being built in the cloud continuum. Our data and analysis can be valuable for researchers and practitioners seeking to build and optimize cloud architectures and to help guide new research.
Routing in Internet of Things (IoT) networks becomes increasingly challenging when non-cooperative nodes participate in packet forwarding, as they reduce delivery reliability, increase communication delay, and consume network resources inefficiently. Existing trust-based and game-theoretic routing approaches generally rely on static trust evaluation or fixed hierarchical structures, limiting their ability to adapt to dynamic network conditions. To address these limitations, this paper proposes a dynamic game-theoretic trust-tree framework for trust-aware routing in IoT networks. The proposed framework constructs and continuously updates a hierarchical trust tree through a two-player bargaining game, where routing decisions are made using a trust-aware utility function that jointly considers packet forwarding reliability, residual energy, communication cost, and node trustworthiness. Unlike conventional hierarchical trust models, the proposed framework dynamically reorganizes the trust hierarchy according to node behavior, enabling adaptive parent selection and more reliable packet forwarding. The routing framework is mathematically formulated and evaluated using The ONE simulator under different network scenarios. Experimental results show that the proposed method improves the packet delivery ratio by approximately 13–16
The rapid expansion of Big Data has transformed data-driven decision-making across diverse sectors, including healthcare, cybersecurity, smart cities, and the Internet of Things (IoT). However, traditional data management platforms struggle to address the increasing volume, velocity, variety, and veracity of modern data ecosystems, resulting in scalability constraints and inefficiencies. These challenges necessitate distributed, cloud-native, and energy-efficient architectures capable of supporting Artificial Intelligence (AI) and Machine Learning (ML)–driven analytics. This paper presents a systematic literature review (SLR) of emerging Big Data analytics frameworks centered on the Hadoop ecosystem and its modern extensions, including Apache Spark, Flink, Kafka, and data lakehouse technologies. Following the PRISMA 2020 guidelines, the review analyzes 138 primary studies published between 2015 and 2025, with 60 studies selected through rigorous inclusion and quality assessment criteria. The analysis evaluates architectural evolution from batch-oriented MapReduce systems to unified batch-stream processing and cloud-native data platforms, emphasizing performance optimization, resource efficiency, and sustainable computing practices. The findings reveal significant progress in integrating AI and ML pipelines with Hadoop-based infrastructures for applications such as real-time fraud detection, anomaly detection, predictive healthcare analytics, renewable energy forecasting, and intelligent urban management. Particular attention is given to explainable AI (XAI), federated data processing, and green computing strategies that enhance transparency, trust, and environmental sustainability. This review proposes a structured taxonomy of Hadoop-based Big Data systems, categorizes emerging research trends, benchmarks modern frameworks, and identifies open research challenges. The study provides theoretical, practical, and policy-level implications to guide researchers, system architects, and decision-makers in selecting scalable, low-latency, and energy-aware Big Data solutions aligned with next-generation digital transformation and sustainable development objectives. Not applicable, as this study is not a clinical trial.
Internet of Things (IoT) technologies and healthcare present revolutionary chances to improve operational efficiency, patient outcomes, and tailored medication transformation. This paper thoroughly investigates IoT in healthcare using Latent Dirichlet Allocation (LDA) to spot important trends and research gaps in current work. To achieve this, researchers have comprehensively analyzed 11,586 published papers from 2006 to 2024 which are extracted from Scopus database. Researchers have identified 2, 5, and 10 key topics to define significant areas of the research. Over time, it compares research topics to show how important areas, including wearable technology, artificial intelligence-powered analytics, blockchain for safe data management, and edge computing, have evolved. The paper additionally examines important issues, including data privacy issues, lack of interoperability, and restricted inclusiveness for underprivileged communities. Emphasizing inclusivity, ethical compliance, and pragmatic implementation tactics catered to different healthcare environments, a strategy framework is suggested to help solve these difficulties. This paper helps IoT implementation in healthcare advance by giving actionable insights, particular discoveries, and future research directions, thereby opening the path for more fair, efficient, and sustainable healthcare systems.
The adoption of Internet of Things (IoT) devices in supply chain management has seen significant growth in recent years. However, existing IoT-based supply chain systems depend on centralized server-client architectures, which introduce security vulnerabilities and create a single point of failure that can disrupt the entire network. To address these challenges, transitioning IoT systems to a decentralized platform is essential for ensuring data security, privacy, and reliability. This paper proposes a lightweight key agreement scheme for authenticating IoT devices in supply chains using blockchain technology. Blockchain enhances security by providing immutability, transparency, resilience, full traceability, and anonymity. Additionally, smart contracts foster trust and reliability among all supply chain participants. The proposed framework ensures confidentiality and mutual authentication between nodes. To validate its robustness, the scheme was simulated using AVISPA and Scyther tools, with results demonstrating strong resilience against potential attacks. Performance evaluation shows that the proposed scheme achieves a computational cost of 0.0414 ms, demonstrating feasibility for resource-constrained IoT environments.
Modern enterprise networks generate high-volume, high-velocity traffic, making timely and accurate intrusion detection increasingly challenging for traditional batch-oriented Intrusion Detection Systems (IDS). These systems often struggle to detect sophisticated and evolving cyber threats due to delayed analysis, limited scalability, and the inability to provide continuous monitoring under dynamic workloads. This paper presents a real-time Network Intrusion Detection framework built on a distributed stream-processing architecture using Zeek, Apache Kafka, and PySpark. Network packets are captured and transformed into structured logs using Zeek, streamed reliably through Kafka, and processed in real time using PySpark for feature extraction, rule-based analysis, and machine learning–based classification. A Random Forest model is employed to identify both benign and malicious traffic across multiple attack categories, while Elasticsearch and Grafana provide low-latency storage, visualization, and alerting capabilities for operational security monitoring. The proposed system enables end-to-end real-time detection with high throughput and low latency, helping bridge the gap between rapid attack execution and timely threat response. Experimental results demonstrate the feasibility of scalable prototype IDS deployment for continuous monitoring, rapid analytics, and actionable security insights, while indicating strong promise for broader deployment with further real-world validation under diverse operational conditions.
Cloud storage misconfiguration continues to be a major source of security incidents in public cloud environments, particularly in widely used object storage platforms such as Amazon Simple Storage Service (Amazon S3). Although cloud providers offer a broad set of built-in security controls, their effectiveness depends on how consistently and correctly they are implemented. In practice, manual configuration, inconsistent deployment processes, and configuration drift often create avoidable security risks. This study presents a policy-enforced Infrastructure-as-Code (IaC) framework designed to reduce common Amazon S3 misconfiguration risks during infrastructure deployment. Using Terraform, the framework integrates security controls directly into the provisioning process, including server-side encryption with AWS Key Management Service (SSE-KMS), HTTPS-only transport enforcement, public access restrictions, least-privilege identity and access management, and audit logging through AWS CloudTrail and S3 server access logging. Rather than relying exclusively on post-deployment monitoring or compliance verification, the proposed framework applies security constraints during infrastructure provisioning to reduce the likelihood of insecure configurations being introduced. The framework was implemented and evaluated within an AWS environment using representative misconfiguration scenarios involving insecure transport, missing encryption parameters, public exposure attempts, and access policy violations. The findings indicate that embedding policy-based security controls within Infrastructure-as-Code workflows can improve deployment consistency, reduce configuration errors, and support alignment with established security guidance, including NIST SP 800–53 Revision 5 and the CIS AWS Foundations Benchmark. This study contributes a structured implementation framework for applying deployment-time security enforcement to Amazon S3 cloud storage environments.
Wireless sensor networks (WSNs) are very useful in all areas such as health care, environmental sensing, surveillance, automation, etc. and are capable of performing real time data collection. One of the main limitations of WSNs is their limited life span because they have limited power source. Also, an efficient method for collecting data has not been developed yet which also limits the life cycle of WSNs. In this paper we propose a hybrid algorithm using two meta-heuristics called Bacterial Foraging Optimization (BFO) and Oppositional Fruitfly Optimization (OFF). The goal of our proposal is to improve the efficiency of data collection while minimizing the energy consumption of WSNs. We use Weighted K-means Clustering (WKMC) to cluster WSN nodes based on their remaining energy; OFF will be used to select the Cluster Head (CH); and BFO will be used to calculate the route of a moving sink. Our approach eliminates hotspots, prevents congestions, minimizes energy waste and provides better scalability. To minimize idle energy loss we also add an adaptive cluster head sleeping strategy. Simulations were performed with NS-2 over the UNSW-NB15 dataset that includes benchmarks of common network attacks and other standard configurations for WSNs. These results clearly show that our proposed OFF-BFO model performs much better than others (LEACH, PSO-based clustering, ACO-LTAWSN, AVOA) in terms of network lifetime, PDR (Packet Delivery Ratio), Average Residual Energy per Node, Hop Count, Number of Active Nodes, End-To-End Delay. Compared to LEACH, our proposed method increases the WSN’s lifetime by 68
Efficient resource management in cloud data centers remains a critical challenge due to highly dynamic workloads, heterogeneous resources, and strict Service Level Agreement (SLA) requirements. Unstable resource allocation and improper virtual machine (VM) consolidation may lead to increased SLA violations, excessive energy consumption, and frequent VM migrations. Many existing approaches rely on reactive decision-making based solely on the current system state, while others treat workload prediction and optimization as separate processes, limiting the potential synergy between them. To address these limitations, this paper proposes DL-GAA, an intelligent framework for dynamic load balancing and VM consolidation in cloud environments. The proposed approach integrates Long Short-Term Memory (LSTM)–based workload prediction with the Green Anaconda Algorithm (GAA) within a unified closed-loop architecture. In this framework, predicted future workload trends guide the optimization process, while optimization outcomes provide feedback to improve subsequent decision-making. The performance of the proposed framework is evaluated through simulation using CloudSim 5.0 based on two real-world workload traces: PlanetLab and Azure 2019. Experimental results demonstrate that under the PlanetLab workload, DL-GAA achieves average reductions of 19.5
The Internet of AI Agents has been proposed as a natural evolution of the Internet of Things and the Artificial Intelligence of Things: a decentralized ecosystem in which autonomous agents interact, cooperate, and learn collectively. Yet, connectivity alone is not sufficient to support large-scale populations of agents acting on behalf of humans, organizations, machines, and other agents. As agentic systems grow in size, autonomy, and operational relevance, the central challenge shifts from enabling communication to governing collective behavior. This editorial argues that the next frontier of networked and distributed intelligence is the emergence of a Society of Agents: a structured population of autonomous AI agents whose interactions are shaped by shared protocols, roles, norms, memory, trust mechanisms, accountability structures, and governance processes. The analogy with human societies is not intended anthropomorphically, but structurally: human societies scale intelligence through language, institutions, collective memory, roles, incentives, and causal accountability; agent societies will require computational counterparts of these mechanisms. We identify shared societal memory and causal governance as two foundational capabilities for this transition. Shared memory enables cumulative collaboration beyond ephemeral message exchange, while causal governance makes distributed agentic operations traceable, auditable, and accountable. We formulate this transition as a scientific manifesto: the next generation of networked intelligence should be designed not merely as collections of increasingly autonomous agents, but as governed computational societies grounded in shared memory, causal accountability, institutional mechanisms, and human oversight. We outline architectural layers, research challenges, and emerging examples of multi-agent orchestration for governable collective agency.
Digital forensics is vital for analyzing extensive image data from mobile devices to identify individuals and activities in investigations. Traditional methods struggle with complex real-world images, particularly distinguishing military personnel from military-themed mannequins. This study assesses multimodal Large Language Models (LLMs) - Google’s Gemini 1.5 Pro, open-source LLAVA, and GPT-4o - for detecting military personnel in 434 mobile device images, including military personnel, mannequins, and civilians. The models achieved strong recall (0.99 for Gemini, 0.98 for LLAVA, and 0.91 for GPT-4o) but only moderate precision (0.69, 0.69, and 0.67 respectively), reflecting a notable rate of mannequin-induced false positives.] Accuracy varied from 0.793 for Gemini and LLAVA to 0.770 for GPT-4o, aligning with observed differences in contextual understanding. Contextual classification also posed challenges: Gemini achieved 0.787 accuracy for country identification, followed by GPT-4o (0.385) and LLAVA (0.121). Unit name recognition remained weak across models. Misclassification of mannequins was the primary source of error, confirming that current multimodal models overemphasize uniform and equipment cues without verifying human authenticity. To enhance interpretability and reduce false positives, we integrated an agentic orchestration layer using CrewAI and LangGraph, which structured multimodal reasoning through dedicated sub-agents for provenance validation, perception, mannequin discrimination, and evidence-grounded attribution. These agentic frameworks substantially improved forensic reliability: CrewAI achieved 0.88 precision with a mannequin false-positive rate of 0.12, while LangGraph reached 0.90 precision and reduced false positives to 0.08. Country attribution accuracy rose to 0.58 and 0.62 respectively. Although recall decreased slightly due to conservative abstention logic (CrewAI 0.74, LangGraph 0.73), this trade-off yielded higher forensic confidence and reproducible, audit-ready decision traces. The results demonstrate that integrating agentic architectures transforms multimodal LLMs from opaque classifiers into transparent, evidence-driven forensic tools - enhancing both analytic precision and the evidentiary defensibility of AI-assisted investigations.
The integration of Large Language Models (LLMs) into telecommunication systems offers transformative potential for dynamic network reconfiguration, sustainable operations, and intelligent automation. This review paper explores studies from 2014 - 2025, the convergence of Large Language Models (LLMs) within framework of Retrieval-Augmented Generation (RAG) and modern network infrastructures such as software defined network (SDN), Network Function Virtualization (NFV), and 6 G. The analysis of capabilities of Large Language Models has been done in intent translation, anomaly detection, self-healing, policy optimization, and knowledge-driven automation. A comparative analysis is conducted between Large Language Models based and traditional AI approaches. The parameter used are trade-offs in flexibility, latency, interpretability, and scalability. By bridging language intelligence and network control, Large Language Models represent a pivotal step toward autonomous, adaptive, and sustainable telecommunication systems.
Flying Ad Hoc Networks (FANETs) have emerged as a critical communication paradigm for coordinating autonomous unmanned aerial vehicles (UAVs) in dynamic, large-scale, and mission-critical environments. Their high mobility, stringent Quality of Service (QoS) requirements, and frequent link disruptions introduce routing challenges that are significantly more complex than those in MANETs and VANETs. This Systematic Literature Review (SLR), conducted in accordance with PRISMA guidelines, provides a comprehensive and unified analysis of FANET mobility models and routing protocols reported across major digital libraries. The initial search identified 1,130 records; after removing 210 duplicates, 920 unique studies were screened, yielding 192 high-quality papers included in the final analysis. The review evaluates UAV mobility models that strongly influence routing stability and network performance and categorizes routing protocols into major families, including proactive, reactive, geographic, topology-based, multipath, secure, hierarchical, bio-inspired, position-based, delay/disruption-tolerant, and cluster-based approaches. Each category is critically analyzed in terms of operational mechanisms, design motivations, performance characteristics, and limitations. Key design constraints such as link volatility, energy consumption, scalability, and real-time adaptability are systematically examined. By synthesizing the selected studies, the review highlights important trends. It identifies unresolved challenges, including maintaining stable communication under high mobility, reducing routing overhead caused by frequent topology changes, achieving energy-efficient and balanced communication, designing adaptive clustering mechanisms, and enabling intelligent, real-time routing decisions. Furthermore, the study identifies a lack of comprehensive surveys that integrate all major routing dimensions—particularly cluster-based and energy-aware architectures—and proposes future research directions focused on resilient, scalable, and AI-driven routing frameworks for next-generation multi-UAV communication systems.
Traditional cybersecurity awareness programs often fail to produce sustained behavioral change due to their static and non-personalized design. This paper presents CyberSense AI, a behavior-driven adaptive cybersecurity education framework that integrates a personalized learning engine, an interactive phishing simulation module, and a real-time threat intelligence system powered by a custom-trained machine learning (ML) model based on eXtreme Gradient Boosting (XGBoost). Beyond system implementation, we formally model the adaptive learning mechanism using a knowledge-state representation and reinforcement-inspired update rule to dynamically align question difficulty with user proficiency. To empirically validate the framework, we conducted a controlled pre-test/post-test study involving 60 participants randomly assigned to a control group and an experimental group. Results demonstrate a statistically significant improvement in phishing detection accuracy for the experimental group ( p < 0.001 , Cohen’s d = 1.47 ), along with sustained two-week knowledge retention. Behavioral analytics further reveal a monotonic improvement curve across simulation sessions, a strong engagement–performance correlation ( r = 0.72 ), and progressive reduction in false-negative (FN) rates. A feature ablation and deployment latency analysis confirms that the XGBoost subsystem achieves sub-100 ms response time, validating real-time mobile suitability. Collectively, these results establish CyberSense AI as a theoretically grounded and empirically validated framework for scalable, human-centered cybersecurity training.
Deploying large language models for structured content generation presents a fundamental tension between quality requirements and computational costs. Standard retrieval-augmented generation approaches struggle with complex multi-hop domain questions due to context loss and retrieval noise, while state-of-the-art models impose prohibitive per-query costs at scale. We introduce Cost-Aware Structured Generation, a domain-agnostic 14-stage pipeline that integrates hybrid retrieval-augmented generation with Reciprocal Rank Fusion, conditional compute routing based on query difficulty, and adaptive voting strategies with domain-specific weighting. Our architecture achieves 93.5
As edge-cloud-based Mobile Crowdsourcing systems swiftly expand, the issues of effective offloading and resource allocation have become essential for reducing energy usage and delay time, and for meeting strict deadline requirements. Nevertheless, current methods, such as deep reinforcement learning and heuristic optimization, have severe drawbacks, including high convergence instability, poor deadline awareness, and high energy consumption, leading to increased deadline miss rates under heavy workloads. To address them, this study proposes a new Physics-Informed Focal Self-Attention Neural Network-Bermuda Triangle Optimization Framework (PIFSANN-BTOF). The energy-delay cost model is developed to capture the trade-off between local and edge execution. Supervised learning of spatial crowdsourcing data is pre-processed and normalized across 500 workers and 1500 tasks in the real world. The proposed PIFSANN combines physics-informed constraint learning, focal self-attention to simulate task urgency and global load, and mutual learning to improve convergence stability. Priority scheduling on the stack and Bermuda Triangle Optimization ensure efficient resource matching. Experimental performance is better, showing the best final rate of 0.92 of a maximum task completion, lowest energy consumed of 520 Joules with a reduction of 52
Starting from Bitcoin transactions, blockchain technology emerged as a ground-breaking innovation and has spread its wings in various applications like the Internet of Things, supply chains, finance and healthcare. However, the scalability limitations of the blockchain have been a significant barrier to its extensive-scale adoption and performance. To address this challenge, one evolving approach is Sharding, which simultaneously balances decentralization, security, and scalability. Sharding is a technique that splits the network into distinct units, popularly termed as shards. Each shard operates independently and is responsible for processing its portion of the network’s state and transaction workload. It allows multiple transactions to be executed in parallel, significantly boosting speed, efficiency, and reducing latency. The technique considerably improves the throughput compared to conventional blockchain, where every node processes transactions. This article critically analyses the sharding types, protocols, their strategies, variety of consensus used, and the building blocks of sharding. Further, it explores shard-based blockchain platforms and presents the recent work on sharding for different domains. This review describes the potential security attacks and their countermeasures. The study presents a small experimental case study by implementing a lightweight adaptive sharding strategy to demonstrate how sharding behaves in practise under realistic IoT transaction loads so that the readers could bridge the gap between the theoretical and practical framework under controlled simulated environment. Further, it highlights the research gaps that lays the foundation for future research and development, enabling the researchers to focus and understand the sharding solutions in the Blockchain ecosystem.
This paper investigates the impact of beat noise and turbulence on spectral amplitude coding optical code division multiple access (SAC-OCDMA) systems operating in underwater wireless optical channels. The turbulence, arising from salinity variations and air bubbles, is modeled using the Weibull distribution. Three multi-laser source configurations are examined: a centralized shared multi-laser source, a uniformly distributed central frequency multi-laser source, and a precisely controlled central frequency multi-laser source. The results demonstrate that, although both beat noise and turbulence affect the detected optical intensity, beat noise is the dominant performance-limiting factor. Furthermore, error-free transmission is achievable under weak turbulence conditions only when a precisely controlled multi-laser source is employed, with a central frequency precision of 3 GHz at a data rate of 1.25 Gbps.
Vehicular ad hoc networks are a type of networks in which adjacent vehicles communicate wirelessly and intelligently with fixed roadside infrastructure. Caching is a process of replicating and locating data for future quick access. It is a proper method for problems that cause weak data availability. Caching frequently accessed data on the client cache is an effective technique to improve performance in a mobile environment. It can reduce traffic and data access delay and disseminate a load of overloaded servers to client caches. Also, when there is no place to store new data items in the cache, a data replacement policy should be applied in the cache. A new caching and replacement method based on fuzzy logic for vehicular ad hoc networks is proposed in this paper. The goal of this method is to increase nodes cache hit rate and data access. This means that instead of locating all information on RSUs, client nodes save important data on themselves caches and share cooperatively with each other. The results of simulations with MATLAB software shows that the proposed method increases the cache hit rate 15
