Future Technology

Mixture encoder and virtual RAM-based polar decoder architecture for high-speed 5G communication systems

TR. Parthasarathy, N.R. Krishnamoorthy — Sat, 11 Apr 2026 00:00:00 +0000

This research presents an optimized architecture for Fifth Generation (5G) communication systems that includes a Mixture Encoder to support multiple combinations of Digital Signal Processing (DSP) operations required for 5G baseband processing. This allows flexible encoding with lower computational overhead, in contrast to traditional polar encoders that rely on fixed arithmetic structures and sizable lookup tables. The lookup table complexity is greatly reduced, resulting in lower memory consumption and faster access, and it also maps ranges to compact intervals. A Built-In Self-Test (BIST) module is integrated before the Mixture Encoder to ensure dependable data feeding and fault tolerance. Furthermore, a virtual channel method developed with Virtual RAM technology eliminates explicit channel processing by allowing direct memory access, bypassing redundant channel operations, and allowing conditional decoding termination before execution. This virtualized method enables early-stage error correction while increasing processing speed, reducing switching activity, and optimizing memory usage. At the receiver, a Successive Cancellation (SC) polar decoder is used to achieve low-latency, energy-efficient decoding. Removing unnecessary operations and enabling sequential recursive decoding reduces arithmetic complexity. According to the FPGA synthesis results, the combination of a mixture encoder, virtual memory access, and SC decoding results in lower power consumption, nanosecond-scale delay, improved decoding precision, and scalable hardware utilization, making the proposed architecture ideal for DSP-intensive communication systems and 5G networks.

Blockchain-driven secure message dissemination in 5G-enabled SDN-IoV using graph-based byzantine consensus and Merkle tree-BLS authentication

Ravindra Janardan Lawande, Sudhir Bapurao Lande, Manisha Lande — Tue, 14 Apr 2026 00:00:00 +0000

Internet of Vehicle (IoV) uses heterogeneous access technologies to link automobiles and their surroundings. Effective methods are essential for safeguarding data confidentiality and privacy during communication among the roadside unit (RSU), the control room, and vehicles. Many vehicle-to-infrastructure authentication-based approaches have been developed to secure the IoV environment. However, efficiency and security are challenged by instability, decentralization, and transaction-tracking features. To resolve this, a secure, lightweight, and scalable communication protocol was developed for a 5G-enabled SDN-IoV environment. Efficient block verification is achieved through the Joint-Graph Delegated Practical Byzantine Fault Tolerance (JtGr-DPBFT) mechanism, in which validators create subgraphs to reduce communication overhead. JtGr-DPBFT is combined with an Improved Gossip Algorithm (IGA) to minimize message redundancy and optimize bandwidth utilization. Moreover, a lightweight hierarchical authentication mechanism, assisted by a Merkle Tree with Boneh-Lynn-Shacham (HAMT-BLS) signatures, enables compact block verification and minimizes computational and communication costs. The proposed model achieves tamper-proof, efficient, and scalable block verification by incorporating hierarchical authentication with consensus optimization. This approach is simulated in the NS3 tool, and performance is evaluated in terms of propagation delay, transaction confirmation latency, throughput, communication cost, and network delay. Thus, secure and tamper-proof communication is developed to ensure integrity, trust, and dependability in the SDN-enabled IoV environment.

Split-CNN for intrusion detection: enhancing feature diversity and training efficiency through channel separation

Harish G N, Annapurna H S — Fri, 17 Apr 2026 00:00:00 +0000

Cyber threats are becoming more sophisticated, and advanced intrusion detection systems (IDS) are needed to detect complex attack patterns on the network. Traditional IDS approaches tend to rely on signature-based methods or manually engineered statistical features, which struggle to detect evolving cyber threats and large-scale network traffic. The paper presents an intrusion detection framework that leverages a deep learning architecture, the Split Convolutional Neural Network (Split-CNN), which enhances feature diversity and training efficiency. Another module, Split Convolution (SplitConv), is proposed in the given model and isolates input feature channels into a few semantic groups, then performs separate convolution processes. This mechanism is interrelated with the decrease in inter-channel redundancy and the increase in discrimination feature learning. To facilitate cross-dataset learning, a feature alignment framework is proposed that can be unified to integrate three standard intrusion detection datasets: NSL-KDD, UNSW-NB15, and CIC-DDoS2019. The preprocessing pipeline includes categorical encoding, feature standardization, and dataset harmonization to construct a single dataset containing 168 features that constitute the four semantic channels. It has been demonstrated that the Split-CNN model is superior compared to the baseline CNN models in both classification and detection accuracy. These findings imply that the proposed approach can provide an effective, scalable deep learning system for contemporary network intrusion detection systems.

Distributed coalition-based resource orchestration for heterogeneous IoT devices in metropolitan smart cities

Si Liu, Midhun Chakkaravarthy — Fri, 17 Apr 2026 00:00:00 +0000

The rapid proliferation of IoT devices in metropolitan environments poses critical challenges for heterogeneous device management under minimal centralized control. This paper presents DCRO, a Distributed Coalition-based Resource Orchestration framework enabling IoT devices to self-organize into dynamic coalitions for cooperative resource management. Unlike traditional hierarchical approaches that suffer from scalability bottlenecks, DCRO integrates three core components: a Self-Organizing Device Clustering Algorithm (SODCA) that adapts to topology changes without global coordination; a Game-Theoretic Coalition Formation Mechanism (GT-CFM) that drives fair resource allocation through Shapley value-based negotiation; and a Lightweight Hierarchical Consensus Protocol (LHCP) coupled with a Merkle-DAG security architecture that ensures tamper-resistant coordination without blockchain overhead. Experiments across three metropolitan testbeds demonstrate 26.2% latency reduction and 31.4% energy savings over centralized baselines, only 11.3% throughput degradation under continuous fault injection, and stable coalition convergence at 5,000 devices within 15 iterations.

Design and fabrication of an efficient solar powered desalination system for remote communities

Wed, 22 Apr 2026 00:00:00 +0000

Solar desalination is a viable solution to freshwater scarcity in remote and off-grid communities that do not have centralized infrastructure. In this work, the design, performance, and evaluation of an autonomous photovoltaic (PV)-battery-reverse osmosis (RO)-water tank desalination system is presented using simulation. The system is modeled in the MATLAB software at an hourly resolution for one complete year (8760h) with realistic climatic data from the POWER (Power System Energy) database at the National Aeronautics and Space Administration. Physically consistent models are applied to PV generation, battery energy-power behavior, RO specific energy consumption, and freshwater-storage dynamics. Baseline results indicate a community demand of 10 m³/day can be met with 98.63% daily reliability, which results in 3643.73 m³/year of desalinated water and 0.319% unmet demand with a realized SEC of 4.74 kWh/m³. However, only 56.43% of available PV-bus energy is utilized, indicating a considerable PV-curtailment. The outcome of phase 2 indicates that a decrease in the levelized cost of water (LCOW) through a relaxation of reliability to a 95% daily reliability, without compromising on acceptable service levels, is possible. Demand sensitivity, design space exploration, and dispatch policy comparison reveal the great potential of curtailment-aware operation in conjunction with sufficient water storage to enhance system efficiency and economic performance for remote desalination applications.

Real-time obstacle avoidance in mobile robots using deep reinforcement learning

Fri, 24 Apr 2026 00:00:00 +0000

Real-time obstacle avoidance is a challenge in mobile robotics, as it is an ongoing process and remains difficult to achieve in crowded, dynamic environments, where conventional planning algorithms, such as local planners, often offer limited adaptability. This paper presents a Proximal Policy Optimization-based deep reinforcement learning approach for real-time obstacle avoidance for mobile robots. The proposed system is end-to-end policy learning based on inputs from LiDAR and other auxiliary sensors, and is trained in a Gazebo-ROS environment using domain randomization to enhance robustness to sim-to-real transfer. The framework was implemented on a TurtleBot3 Burger platform and tested both in simulation and in an indoor physical environment with varying numbers of obstacles. In simulation, the proposed policy achieved a success rate of 94.2%, a 68.6% reduction in collision rate compared to the Dynamic Window Approach baseline policy, a path efficiency of 16.5%, and a 14.6% reduction in average time to goal. In real experiments, the policy has maintained success rates above 88, even under high-density conditions. The optimized onboard inference pipeline achieved less than 20 ms latency and over 50 Hz throughput on embedded hardware. These results indicate that the proposed framework is a successful and computationally feasible solution to real-time robotic navigation in dynamic environments.

Experimental and sustainability-driven engineering assessment of recycled concrete aggregates: mechanical performance, durability, environmental and economic implications

Sun, 26 Apr 2026 00:00:00 +0000

With the increased demand for concrete in modern infrastructure, coupled with the depletion of natural aggregates and the rising volume of construction and demolition waste, there has been a growing need for sustainable construction materials. This study evaluates the mechanical performance, durability, environmental impacts, and economic feasibility of concrete incorporating recycled concrete aggregates (RCA) as a partial or total replacement for natural coarse aggregates. Five concrete mixtures were prepared with RCA replacement levels of 0%, 25%, 50%, 75%, and 100%, aiming for a compressive strength of 30 MPa. Experimental investigations were conducted on compressive and flexural strength, workability, water absorption, and rapid chloride permeability, while environmental and economic performance were assessed through life cycle assessment and cost analysis. Results showed that concrete with 50% RCA achieved compressive strengths of 29.3 MPa and 4.1 MPa and flexural strength of 4.1 MPa at 28 days, which can be considered acceptable structural performance. Workability decreased with increasing RCA content due to increased porosity and water absorption, but improved significantly with the addition of a superplasticizer and aggregate pre-treatment, increasing the slump from 83mm to 120mm. Full RCA replacement resulted in a reduction of CO₂ emissions by 32%, embodied energy by 33%, and concrete production cost by USD 12.3/m³. Overall, RCA shows a good potential for sustainable and circular infrastructure development.

Federated reinforcement learning for energy-aware load balancing in edge-fog-cloud IoT continuum

Si Liu, Midhun Chakkaravarthy — Mon, 27 Apr 2026 00:00:00 +0000

Energy efficiency remains a major challenge in deploying IoT systems, especially in scenarios requiring large numbers of devices while balancing computational requirements and operational lifetimes. This paper proposes a federated reinforcement learning framework for adaptive load balancing in the edge-fog-cloud continuum that optimizes energy efficiency and supports diverse quality of service requirements. The proposed framework addresses the limitations of traditional centralized machine learning approaches that require collecting sensitive operational information and transmitting it to cloud servers for centralized analysis. This increases the risk of privacy violations and introduces communication overheads that limit the responsiveness of IoT systems. The proposed framework employs a federated reinforcement learning approach, enabling edge nodes to collaboratively learn an optimal load-balancing policy without transmitting operational information. The proposed framework uses a context-aware reward function that optimizes multiple objectives based on temporal patterns, device energy levels, and application criticality. This enables the proposed framework to adapt its optimization objectives and balance energy efficiency and performance maximization. The proposed framework introduces a new action-space pruning mechanism that accelerates the optimization process by leveraging domain knowledge of possible load-balancing patterns. The proposed framework uses a distributed experience replay buffer to reduce trial-and-error in reinforcement learning. The proposed framework demonstrates its effectiveness in optimizing energy efficiency through a series of experiments in a real-world IoT environment and a centralized machine learning approach. The proposed framework demonstrates that distributed machine learning approaches can outperform centralized ones for optimizing energy efficiency in IoT systems.

Federated leveraging AI-assisted MPPT for real-time photovoltaic performance optimization using machine learning

Srinivas S, Shamala N, Yogesh T — Wed, 06 May 2026 00:00:00 +0000

Maximum power point tracking is a necessary power optimization technique for maximizing the energy output of photovoltaic PV systems that operate in variable environmental conditions. While classical MPPT algorithms, such as Perturb and Observe and Incremental Conductance (IncCond), are well implemented, research is now underway on artificial intelligence techniques to improve tracking performance. This paper reports on a control-oriented benchmarking study of classical and AI-assisted MPPT strategies within a unified discrete-time simulation framework implemented in MATLAB. An artificial neural network is proposed in an AI-assisted MPPT architecture to estimate the optimal PV operating voltage, while a conventional proportional-integral voltage controller enforces it and maintains closed-loop stability. The performance of P&O, IncCond, and the AI-assisted MPPT is studied under uniform irradiance steps, fast irradiance fluctuations, and partial shading conditions. Under a 1000->600 W/m² irradiance step, IncCond delivers a tracking efficiency of 99.89% with very low power ripple (0.0014 W), which is significantly better than P&O (98.13%) and the AI-assisted approach (67.92%). In partial shading, P&O and IncCond have efficiencies of 96.21% and 92.99%, respectively, while the AI-assisted MPPT has an efficiency of 67.84%. These results show that the IncCond is a strong and reliable baseline, and that the AI-assisted MPPT offers valuable insight into hybrid control design and requires careful, control-aware integration.

A quantitative benchmarking framework for reinforcement learning-based low-dose CT image denoising

Amit Bhupal Pattar, Thimmaraju S N — Sat, 09 May 2026 00:00:00 +0000

Low-dose computed tomography (LDCT) reduces radiation exposure but increases noise and structural degradation, which may affect diagnostic reliability. This paper presents a quantitative benchmarking framework to assess reinforcement learning (RL)-based LDCT denoising under standardized, reproducible experimental conditions. The proposed pipeline combines dataset splitting with controlled fragments, percentile preprocessing, classical and deep learning baselines, an RL denoising environment modeled as a Markov Decision Process, and multi-metric statistical validation. Experimental results on multi-level LDCT data show that the proposed RL_stageB model provides the highest overall reconstruction fidelity, which can achieve a mean PSNR 22.732 ± 0.947 dB and SSIM 0.929 ± 0.056, which is higher than strong classical baselines such as bilateral filtering 22.618 dB and Gaussian filtering 22.727 dB, while reducing edge distortion, Edge-L1=0.242. Statistically significant improvements (p<1e-300) are found in most comparisons using paired Wilcoxon signed-rank testing. The robustness analysis demonstrates that it maintains stable performance under both noise conditions and a small range of seed variance, with RMSE: 0.0696-0.0713. These findings present RL as an adaptive sequential denoising approach and provide a benchmarking framework for future LDCT restoration studies from a reliability perspective.

Enhancing operational and infrastructure integration in shared mobility

Madhuri Vikas Mane, Deepak Kumar, Kamal Agarwal — Sun, 10 May 2026 00:00:00 +0000

The carpooling system is an automated system that eases travelers’ misery and helps them find cars quickly. One application that will be turbocharged is carpooling, where solo drivers to work can ask other passengers in our application for a ride. It provides the car user with an easy-to-use platform between the car owner and the car user. The existing carpooling schemes aim to reduce carbon footprint and environmental impact by matching passengers and drivers along the way. However, they are generally severely deficient in features, including inefficient static route planning, low-quality ride-matching algorithms, and a lack of high-quality user trust mechanisms. Our proposed system can resolve these problems by operating algorithmically (e.g., using A-Star, which dynamically adjusts to optimize routes based on the real environment) and by leveraging advanced machine learning models, such as clustering and recommendation systems, to improve the precision of ride matching. We also enhance the trust feature by providing comprehensive profiles of drivers with verified contact details, vehicle condition reports, user ratings, and reviews, thereby offering an efficient, dependable, and secure carpooling experience.

Apache Hadoop for large-scale data processing using machine learning techniques

Nidaa Ghalib Ali , Mohanaed Ajmi Falih, Ali Ajmi Falih — Thu, 14 May 2026 00:00:00 +0000

As big data volumes increase and data variety becomes greater, there is a need for more advanced technology. The paper discusses Volume, Variety, and Velocity, which are known as the 3Vs of Big Data, along with Valence and Veracity. As organizations battle with these complexities, Apache Spark perhaps emerges as a technology that can overcome the limitations of Hadoop MapReduce to enable real-time analytics. The focus of this paper is on Big Data. The study evaluates the effectiveness of the K-Nearest Neighbors (KNN) algorithm on structured data. Decision Tree regression is evaluated on unstructured data, and logistic regression on semi-structured data in this study. The algorithms performed well on structured data; however, all the models failed to predict unstructured data. Moreover, an examination of the framework’s performance proves the computational efficiency of Apache Hadoop and Apache Spark. Furthermore, in terms of processing speed across all data types and algorithms, Spark outperformed Hadoop. As a result, it requires advanced analytical tools. Apache Spark is a modern, high-performance data processing framework that enables organizations to manage Big Data in real time.

From algorithms to intelligence: exploring the fusion of AI and computer science for emerging technological solutions

Fri, 15 May 2026 00:00:00 +0000

Artificial intelligence and computer science will converge to produce the next set of emerging technologies. Artificial intelligence enables machines to learn, make intelligent decisions, and adapt, but it is built on the foundations of computer science, including algorithms, data structures, optimization, complexity theory, and computing platforms. This theoretical review examines the role of these foundations in modern artificial intelligence systems and their contributions to the development of scalable, explainable, efficient, and deployable technologies. The paper discusses the technical foundations of machine learning, deep learning, graph-based intelligence, neuro-symbolic systems, explainable artificial intelligence, and edge intelligence in terms of algorithmic reasoning, data structures, learning optimization, computational efficiency, and computational infrastructure. It also discusses the role of integrating artificial intelligence and computer science across major application areas, including smart health, cybersecurity, robotics, natural language processing, Internet of Things systems, edge computing, and sustainable digital infrastructure. To improve the paper's conceptual framework, it proposes an Algorithm-to-Intelligence Integration Framework that connects computer science foundations, artificial intelligence paradigms, system requirements, application domains, and future technologies. The survey finds that intelligent systems should, in the future, combine adaptive learning with robust computational design to achieve responsible, secure, sustainable, and deployable technological advancement.

A data-driven multivariable framework for operational regime identification, product transition detection, and anomaly detection in industrial pumping systems using SCADA data

Wed, 20 May 2026 00:00:00 +0000

This study analyzes a centrifugal pumping system in an industrial facility using fifteen months of operational data collected from a Supervisory Control and Data Acquisition (SCADA) system. Applying a flow greater than zero criterion, 15,049 records corresponding to active operation were retained; after quality control and removal of incomplete and feature-inconsistent observations, 14,501 records were used for the multivariable analysis. Instead of analyzing variables independently, the study characterizes system behavior through the relationships among hydraulic, electrical, and fluid-related variables. Principal Component Analysis (PCA) is applied first, and the first two components explain 69.8% of the total variance. Based on this reduced representation, K-means clustering identifies two operational regimes, corresponding to dominant and low-load conditions. A Gaussian Mixture Model (GMM) applied to fluid density reveals two product regimes with mean values of 716.84 kg/m³ and 830.35 kg/m³. In addition, anomaly detection based on the Mahalanobis distance identifies 73 anomalous observations (0.5% of the dataset), associated with reduced discharge pressure, lower pressure differential, and decreased power consumption, indicating degraded operating conditions. The proposed framework provides a physically interpretable representation of system behavior, enabling the identification of operational regimes, product-related variations, and anomalous conditions within a unified analytical approach. This supports its application in industrial monitoring environments aligned with Industry 4.0 (I4.0) principles.

A federated learning approach for predicting AI capability synergy effects in manufacturing mergers and acquisitions: engineering collaboration and economic value creation

Tengfei Fan, Minghao Huang — Sat, 30 May 2026 00:00:00 +0000

Manufacturing mergers and acquisitions increasingly target AI capabilities, yet predicting synergy effects remains constrained by cross-enterprise data privacy barriers that render centralized approaches impractical. This study proposes a novel horizontal federated learning framework for AI capability synergy effect prediction in manufacturing M&A, integrating federated histogram-aggregated gradient boosting trees with FedAvg-optimized deep neural networks through a two-stage decoupled training strategy, alongside a systematically constructed engineering collaboration indicator system encompassing R&D compatibility, production system interoperability, and AI talent overlap. Empirical validation across 286 authentic manufacturing M&A cases from 23 enterprises demonstrates that the proposed framework achieves superior predictive performance to centralized machine learning and traditional econometric baselines while preserving complete data confidentiality, with engineering collaboration indicators contributing more substantially to synergy prediction than financial variables, and AI-intensive acquirers generating pronounced post-merger economic value premiums following a time-lagged pattern. These findings establish a methodological bridge between privacy-preserving machine learning and strategic management research, providing manufacturing executives with a comprehensive decision support toolkit for target screening, due diligence, and post-merger integration planning.

Multiplier leadership optimization algorithm (MLOA) for high-resolution images in approximate DWT-based compression systems

R. Anitha, Sri Adibhatla Sridevi — Mon, 01 Jun 2026 00:00:00 +0000

Image compression is a basic need for efficient storage and transmission of high-resolution visual information of modern imaging and sensing systems. Algorithmic-level approximation within biorthogonal discrete wavelet transforms (DWT)-based compression has become an effective means in reducing the computation cost while keeping the image perceptual quality. A convolution-based wavelet framework introduced at the multiplier level to control the approximation leads to lower power usage and silicon area in hardware implementations in a systematic manner. In this work, the Multiplier Leadership Optimization Algorithm (MLOA) is used to select exact or approximate multiplier configurations under PSNR and SSIM constraints for energy-efficient hardware implementation. Wallace tree, Dadda tree, Vedic, and Baugh-Wooley multiplier architectures are embedded into the wavelet transform for efficient computation. Simulating with image datasets such as Castle, Baboon, Cameraman, Woman, and Boat shows that the evaluated configurations maintain PSNR values above 30 dB, while SSIM is used as the primary feasibility constraint for structurally sensitive images. The FPGA synthesis results show that Dadda-based MLOA configuration achieves the lowest normalized power and delay among the evaluated multiplier architectures, while the Multiplier Leadership Optimization Algorithm-based Leader-Column Approximate Kogge-Stone Adder (MLOA-LC-AKSA) configuration achieves 145 LUTs, 3.6 ns delay, 52 mW power, 187.2 pJ power-delay product, and 277 MHz maximum frequency. Furthermore, the parallel execution of row-wise and column-wise wavelet convolutions yields a throughput improvement of up to 41%. These results validate the algorithmic-level approximation, HDL-based hardware feasibility, and the suggested framework's parallel-processing capacity as a scalable, energy-efficient, hardware-oriented high-resolution picture compression solution.

Imbalance-resistant multiclass attack classification in real-time IoT water networks using SMOTE-enhanced random forests

Anita Anand, Shivangi Surati — Thu, 04 Jun 2026 00:00:00 +0000

The implementation of smart water distribution systems that rely on the Internet of Things (IoT) has substantially increased the need for intrusion detection systems capable of distinguishing among various categories of attackers. Such granularity is essential for timely and appropriate incident response. The nature of telemetry streams in operational settings is imbalanced: normal traffic is prevalent, whereas the rare but important classes of attacks are represented by a small number of attacks. In such circumstances, the traditional type of classifier can achieve high overall accuracy but fails to identify minority threats of greatest operational interest. This paper introduces a multi-class attack classification model that is robust to class imbalance and operates in real time on the IoT water network, classifying samples using the Synthetic Minority Over-sampling Technique (SMOTE) combined with a Random Forest (RF) ensemble classifier. The data used in the study is a collection of 1,048,575 telemetry records that simulate smart water infrastructure behavior by combining network indicators such as AnomalyScore, DataRate, and Protocol with physical-process indicators such as WaterFlowRate (Lpm), thereby covering cyber-physical interactions. An RF model trained on the original imbalanced dataset is compared with one trained on SMOTE-balanced data and evaluated on an unseen imbalanced test set. Even though the baseline achieves 99.3% accuracy, its recall is 0% for the rare DoS and DDoS classes. However, in comparison, the SMOTE-enhanced model obtains 99.88% accuracy and a higher recall of 92.31% for DoS and 99.66% for DDoS, and the macro- averaged F1-score rises from 0.60 to 0.93. The most discriminative features are recognized as AnomalyScore, DataRate, and WaterFlowRate (Lpm), which support interpretability and informed decision-making in sustainability-sensitive smart water infrastructure.

Future technologies for climate adaptation: AI-based modeling, ICT systems, and engineering solutions within India’s governance framework

Fri, 05 Jun 2026 00:00:00 +0000

Climate change is accelerating the frequency and intensity of climate hazards in India, heightening the urgency for robust, technology-enabled adaptation strategies. While emerging tools such as AI-based climate modeling, ICT-enabled monitoring systems, satellite analytics, digital-twin simulations, and climate-resilient engineering solutions offer transformative potential for anticipatory governance, their effective deployment depends on a responsive legal and institutional framework. This study critically examines how India’s environmental governance architecture engages with these technologies and identifies the regulatory, institutional, financial, and intellectual-property constraints that shape their adoption. Using a qualitative doctrinal and policy-analysis methodology, the study assesses statutory instruments, judicial doctrines, and policy frameworks, complemented by comparative insights from the European Union, the United States, and Australia. The findings reveal that although India’s environmental laws provide strong normative foundations, they remain mitigation-centric and lack explicit mandates for technology-driven adaptation. Institutional fragmentation, weak data governance systems, limited adaptation finance, and barriers to technology transfer further constrain technological integration. The study argues for a comprehensive climate-adaptation law that incorporates technology facilitation, harmonized data and IP frameworks, multi-level coordination, and equity-focused provisions. Such reforms are essential to build an innovation-enabling, accountable, and resilient adaptation regime suited to India’s escalating climate challenges.

Optimization of wind turbine blade designs using computational fluid dynamics and structural analysis: a review

Mon, 08 Jun 2026 00:00:00 +0000

Wind turbine blade optimization requires coordinated improvement of aerodynamic efficiency, structural reliability, fatigue life, manufacturability, and computational cost. This systematic literature review synthesizes studies on wind turbine blade design optimization using computational fluid dynamics and structural or aeroelastic analysis, with attention to design variables, modeling approaches, coupling strategies, optimization methods, validation practices, and limitations. The review protocol was registered with the OSF Registries under DOI 10.17605/OSF.IO/VAR9T. Following a PRISMA-guided process, 233 records were identified from Scopus, Web of Science, ScienceDirect, IEEE Xplore, Google Scholar, and manual reference checks. After removing 25 duplicates, 208 records were screened, 34 full texts were assessed, and 14 studies were included for qualitative synthesis. The literature clustered into three streams: CFD-based aerodynamic shape optimization, especially airfoil, blade-tip, chord, twist, and sweep refinement; aeroelastic or multidisciplinary optimization balancing annual energy production with loads, fatigue, and control constraints; and structural or composite optimization addressing mass, stiffness, stress, deflection, buckling, laminate design, and manufacturability. Many studies used hybrid workflows combining selective high-fidelity CFD with reduced-order, beam, cross-sectional, or surrogate models. Integrated aero-structural optimization appears most practical, but comparisons remain limited by inconsistent load cases, incomplete validation, limited reporting of uncertainty, and insufficient treatment of manufacturability.

Cross-domain aspect-based sentiment analysis using DeBERTaV3 and bio-CRF: a syntactic-aware span extraction approach

Udayalaxmi Aditya Teki, P. Ranjana — Tue, 09 Jun 2026 00:00:00 +0000

Aspect-Based Sentiment Analysis (ABSA) often experiences a significant performance decline in cross-domain settings due to vocabulary variation and domain-specific aspect expressions. Although transformer-based models achieve strong in-domain performance, they primarily rely on contextual embeddings and often ignore the syntactic structures that remain consistent across domains. Existing methods rarely integrate structured decoding with adaptive syntactic fusion for robust aspect boundary detection. This paper proposes a syntactic-aware cross-domain ABSA framework based on DeBERTaV3 and BIO-CRF decoding to alleviate the above problems. The proposed model introduces part-of-speech and dependency-relation embeddings, in addition to contextual embeddings, and uses an attention-based model to dynamically fuse syntactic and semantic information at multiple levels. We further apply a Conditional Random Field (CRF) layer to enforce valid BIO transitions and enhance the consistency of multi-word aspect spans under domain shift. The model was evaluated in three English review domains: Restaurant, Laptop, and Device across six zero-shot cross-domain transfer settings (D→L, D→R, L→D, L→R, R→D, and R→L). Test results show consistent advances over robust transformer-based and prompt-based baselines. The proposed method yields F1 scores for aspect extraction between 0.72 and 0.81 and achieves sentiment classification accuracies between 74.32% and 85.19%. The best performance was achieved in the L→R transfer setting. Through paired bootstrap testing (p < 0.01), Statistical analysis confirms that the proposed model achieves significant improvements over baseline methods. The results demonstrate that incorporating explicit syntactic knowledge, adaptive feature fusion, and structured decoding substantially improves cross-domain generalization in fine-grained sentiment analysis.

Impact of AI on triple bottom line performance and economic sustainability in megaprojects: a systematic review and conceptual framework

Zilu Ni, Yamunah Vaicondam, Malarvilly Ramayah — Wed, 10 Jun 2026 00:00:00 +0000

Megaprojects have significant impacts on global infrastructure development, yet they continue to face sustainability challenges, including high costs, environmental damage, and social conflict. Artificial intelligence (AI) technologies are transforming construction management, but there is little literature examining the integration of AI into construction and megaproject sustainability. This gap is addressed through a comprehensive literature review on the impact of AI on the triple bottom line (TBL) performance and economic sustainability of megaprojects and by proposing a conceptual framework supported by research propositions. A Boolean combination of three keywords in the Scopus database resulted in 348 initial articles, from which 18 key articles were selected for further analysis. All three keyword categories identified only five articles pertinent to the current research topic, highlighting a clear knowledge gap. Analysis shows that AI research has matured in economic performance areas such as cost estimation and resource optimization, with 87% of reviewed papers addressing economic aspects. Research on environmental performance, particularly carbon emissions and waste management, has progressed but remains limited. Social performance, including stakeholder management and community impact assessment, is the least explored dimension. Based on the Technology-Organization-Environment (TOE) framework and stakeholder theory, this study develops a theoretical model with three layers: AI technology inputs, TOE conditions, and TBL performance outputs, in which economic sustainability serves as a higher-level outcome. Four propositions are developed to identify how AI impacts each TBL dimension and economic sustainability. This study contributes to the theoretical groundwork and direction for future empirical studies.

A unified mathematical and computational framework for predictive analysis of complex dynamic systems

Pradeep Kumar H S, Harsha S — Fri, 12 Jun 2026 00:00:00 +0000

Interdependent operating states, rather than individual load curves, are increasingly required for short-term power system prediction. This study formulates and tests a common mathematical-computational approach for one-step-ahead prediction of demand, generation, load shedding, and imbalance derived from them within a dynamic grid environment. Operational data were archived as hourly data, prepared chronologically, transformed into a multivariate feature space, and split into training, validation, and test sets. The proposed framework is tested against naïve persistence, Ridge regression, Random Forest, XGBoost, and the VAR models separately and is based on a vector autoregressive mathematical core and an XGBoost residual-correction layer. In the results, the model's effectiveness is target-dependent. The hybrid framework consistently performed best for generation and was statistically similar to the advanced models for demand, load shedding, and derived imbalance. The mathematically derived variables, source-composition features, and short-term dynamic indicators were found to have different contribution values for each target in ablation and feature-importance analyses. Stressed load-shedding conditions were also found to have lower predictive accuracy in regime-specific testing. These results show that mathematically constrained residual learning is useful for coherent forecasting of continuous operating states, while sparse stress-related variables necessitate extensions to the model to learn them in an event-aware fashion. The study offers a replicable methodology for predictive analysis of a shorter time horizon in the context of grid operation.

Predicting silver economy demand during population aging transition via federated learning-based multi-platform behavioral data collaboration

Qing He, Mustazar Mansur, Hazrul Izuan Bin Shahiri — Wed, 17 Jun 2026 00:00:00 +0000

The accelerating global population aging has fueled a surge in financial demands in the silver economy, making it critical to forecast elderly financial demands accurately for product allocation and risk management in institutions. The study proposes the Federated Learning-based Silver Economy Prediction Framework (FL-SEPF), which is a collaborative framework for privacy-preserving prediction in the silver economy. FL-SEPF features a four-layer architecture with an adaptive weighted federated aggregation strategy that dynamically computes platform-specific weights based on data volume, data quality, and local validation loss to address Non-IID heterogeneity. Local models employ BiLSTM with attention mechanisms, and a cross-platform attention fusion module integrates multi-dimensional features for multi-task prediction covering demand type classification and intensity regression. Differential privacy and Top-K gradient sparsification ensure privacy protection and communication efficiency. Experiments on four-platform datasets, covering 185,000–203,000 elderly users, demonstrate that FL-SEPF achieves an F1-score of 0.8312±0.0047 and AUC-ROC of 0.8927±0.0038, outperforming FedAvg by 3.5% and XGBoost by 11.4% in F1-score, with only a 1.7% gap compared to centralized Transformer training. Ablation studies confirm that adaptive weighted aggregation contributes the largest performance gain (3.34% F1 drop upon removal). Under extreme Non-IID conditions, FL-SEPF shows only 5.5% F1 degradation versus 11.3% for FedAvg, and at a privacy budget ε=1.0, performance loss is limited to approximately 1.0%. SHAP analysis reveals that financial behavior features, particularly portfolio diversity index and credit utilization rate, are the dominant predictors. This study provides a systematic federated learning solution for silver economy demand prediction under privacy-compliant conditions.

Securing national connectivity infrastructure through identity resilience: implications for zero trust–aligned telecom security

Shiva Kumara, Maunik Shah — Tue, 23 Jun 2026 00:00:00 +0000

The importance of identity-centric controls for securing national connectivity infrastructure in cloud-native telecom environments is increasingly recognized. Modern telecom control planes are built on software-defined and service-based architectures. Identities are both a trust boundary and a significant attack surface. This study evaluates the effects of identity compromises on security and operational behavior in a simulated cloud-native telecom control plane. In this paper, we describe a scenario-based experimental approach to assessing three security postures: (i) perimeter-based, (ii) Zero Trust-based, and (iii) Zero Trust-based with basic identity-resilience mechanisms. Our findings demonstrate that perimeter-based security was bypassed in all evaluated attack scenarios and that it provided broad control-plane reachability. Zero Trust aligned security reduced attack success to less than 15% and limited lateral propagation. The attack success rate dropped to zero across all tested scenarios when identity resilience mechanisms were added. The average blast radius reduced from more than five services under perimeter security to near zero with identity-resilient Zero Trust. The measured request-success rate during attack and containment windows decreased from 100% under the perimeter baseline to 0% under the Zero Trust and identity-resilient configurations for unauthorized or quarantined requests. This decrease was primarily due to intentional policy-based denial rather than infrastructure failure. The results in the simulated environment show that identity resilience can enhance Zero Trust by reducing the persistence of compromised identities. The results also show the security-availability trade-offs, which must be further validated in telecom environments at production scale.

Geospatial-based AHP approach for rainwater harvesting sites in semi-arid areas: a case study of Wadi Abu Ghraibat

Hisham M. Jawad Al Sharaa, Nadia Aziz, Israa Hatem — Wed, 24 Jun 2026 00:00:00 +0000

Rainwater harvesting is becoming an important method of water management in semi-arid areas characterized by seasonal and perennial water scarcity. The investigation demonstrates a novel Analytic Hierarchy Process technique based on geospatial methods, applied to assess site suitability for rainwater harvesting in the semi-arid Wadi Abu Ghraibat area. Elevation, slope, precipitation, and drainage density were the four variables considered important in this respect. Each of these criteria was processed using both GIS and remote sensing data and weighted by the Analytic Hierarchy Process, which assigned 24% to elevation, 10% to slope, 22% to precipitation, and 44% to drainage density, with a consistency ratio of 0.06, which was acceptable. Therefore, the suitability map can be considered the area’s most suitable for implementing rainwater harvesting practices. Although it is limited to water-stressed environments, interesting findings on water resource management lessons emerge. GIS-based integrated approach combining MCDA with GIS for sustainable water management in drought-prone regions.

Green cement production technology for reducing GHG emissions from the industrial sector of Kazakhstan

Sat, 27 Jun 2026 00:00:00 +0000

Reducing clinker and cement consumption is one of the key pathways to lowering CO₂ emissions from cement-related industrial processes. This study investigates the molecular interaction mechanism of an ester-based polycarboxylate ether (PCE) fragment with Ca²⁺ and SiO₂ as a simplified representation of PCE-assisted silica-fume systems and evaluates how such material-efficiency assumptions can be incorporated into Kazakhstan-specific greenhouse gas emission scenarios. Density functional theory calculations were performed using the B3LYP-D3/6-311++G(d,p) level of theory, followed by molecular electrostatic potential, non-covalent interaction, reduced density gradient, electron localization function, and QTAIM analyses. The results indicate that carboxylate oxygen atoms in the PCE fragment act as the main coordination sites for Ca²⁺, while the SiO₂ model contributes additional oxygen-containing interaction sites. The ternary PCE–Ca²⁺–SiO₂ system shows a more connected interaction network than the isolated PCE and PCE–SiO₂ systems, supporting the plausibility of Ca²⁺-mediated adsorption and dispersion in silica-rich cementitious environments. In parallel, greenhouse gas emissions from Kazakhstan’s Industrial Processes and Product Use sector were assessed under three scenarios: without measures, with current measures, and with additional measures. The additional-measures scenario incorporates material-efficiency assumptions related to optimized use of PCE–silica fume, clinker reduction, and process improvements. The results should be interpreted as a molecularly informed scenario framework. The study contributes to the discussion of green cement production technologies and industrial decarbonization pathways in Kazakhstan.

Transfer learning in neural networks: leveraging pre-trained models for improved performance

Abdul Sttar Ismail Wdaa , Iraq Ali Hussein, Ali Azeez Ahmed — Mon, 29 Jun 2026 00:00:00 +0000

Transfer learning has become a key technique for improving the accuracy of neural networks in low-resource, low-data environments. The quantitative comparative analysis of the pre-trained models includes ResNet50, VGG16, BERT, GPT, and the baseline CNN and LSTM models. They are compared across three different application areas: computer vision, natural language processing (NLP), and medical imaging. The five benchmark datasets used were ImageNet, CIFAR-10, SST-2, IMDB, and Chest X-Ray. All experiments used the same preprocessing pipeline and evaluation metrics (accuracy, F1 score, precision, recall, and ROC-AUC). Results showed that models trained on the pre-trained data achieved consistently greater accuracy than the baselines in all domains (9-20%) and F1-score (0.09-0.16) gains. ResNet50 achieved 92% accuracy on CIFAR-10, compared to 72% for the CNN baseline, whereas BERT hit 92% on SST-2, with 80% accuracy for LSTM. VGG16 improved the accuracy of Chest X-Ray classification from 78% to 87% and reduced training time by up to 60%. There were a few instances of minor overfitting and domain mismatch, emphasizing the need for adaptive fine-tuning strategies. The results demonstrate that transfer learning significantly improves convergence speed, generalization, and computational efficiency, making it a promising approach for AI applications across domains such as healthcare, NLP, and autonomous systems.

An explainable AI-based workforce intelligence framework for integrating future skill demand and employee attrition prediction with risk-aware decision analytics

Prathap D L, Thimmaraju S N — Tue, 30 Jun 2026 00:00:00 +0000

Artificial intelligence supports workforce analytics by improving skill assessment, attrition prediction, and talent planning. However, external labor-market skill demand and internal employee attrition risk are often analyzed separately. This study presents a workforce intelligence framework based on explainable AI that combines skill-demand cluster analysis, attrition prediction, explainability (SHAP), and aggregate risk-aware decision support. Two public datasets were used: the Jobs and Skills Mapping for Career Analysis dataset and the IBM HR Analytics Employee Attrition Dataset. TF-IDF was applied to job-related text to generate clusters of job-required skills and to predict auxiliary pay grades, and Logistic Regression, Random Forest, and XGBoost were evaluated for attrition prediction. Logistic Regression was the best-performing model for identifying the risk of attrition, with a recall of 0.6170, an F1-score of 0.4328, and an ROC-AUC of 0.7954. The best recall and F1-score were obtained at a threshold of 0.40, with values of 0.7872 and 0.4901, respectively, as determined by threshold analysis. Over time, SHAP identified frequent business travel, job level, lab technician position, and total years of work as important factors in attrition. The Workforce Risk Score was a combination of normalized skill demand and normalized aggregate attrition risk, with the highest-ranked skill-demand cluster being moderate at 0.396. The framework provides actionable summary-level decision support for workforce planning.

Dynamic graph neural networks meet causal inference: estimating AI’s heterogeneous effects on supply chain resilience

Yang Liu, Jiaming Yang, Jian Chen — Sat, 04 Jul 2026 00:00:00 +0000

Estimating heterogeneous treatment effects in network-embedded environments requires methods that simultaneously account for relational change, conditional heterogeneity, and quasi-experimental shocks. This study proposes TGAT-CF, an approach that pairs a temporal graph attention encoder for evolving supplier-customer relationships with a generalized random forest for conditional treatment effects, triangulated against staggered difference-in-differences and double machine learning. Temporal graph embeddings replace scalar centrality as moderators, and the resulting design allows identification to be cross-checked across three estimators in settings where technology adoption unfolds inside relational dynamics. The framework is applied to a balanced quarterly panel of 2,847 Chinese A-share manufacturers over 20 quarters from 2020Q1 to 2024Q4, yielding 56,940 firm-quarter observations amid simultaneous AI diffusion and trade policy uncertainty. The temporal encoder reduces mean squared error by 21.4 percent relative to a static GraphSAGE baseline and by 6.6 to 9.4 percent relative to dynamic baselines (Diebold-Mariano, p < 0.05). A one-standard-deviation rise in AI stock raises supply chain resilience by 0.34 standard deviations, an effect that is 2.6 times larger under high uncertainty. Conditional effects differ by a factor of 2.9 between modular and centralized configurations, and the temporal profile follows a J-curve peaking at event time 2. Network centrality is the leading moderator, ahead of ownership structure, while operational efficiency, supplier adjustment, and information processing mediate nearly three-quarters of the total effect. The three estimates converge within a 7 percent band. AI capability, therefore, acts as a network-dependent, rather than a universal, determinant of supply chain resilience.

Assessment of solar energy potential and climatic effects on utility-scale photovoltaic power generation in The Gambia

Youssef Kassem, Hüseyin Çamur, Ernest Sagnia — Wed, 08 Jul 2026 00:00:00 +0000

The aim of the present study is to investigate the technical and economic viability of implementing photovoltaic (PV) systems in The Gambia by analyzing the availability of the solar resource, system performance, generation capacity, and economic viability. Long-term climate data from the NASA POWER database have been used to evaluate global horizontal irradiation, ambient temperature, rainfall, relative humidity, cloud cover, and wind speed at five locations in The Gambia. The analysis shows good potential for solar energy use, with annual GHI values ranging from 5.776 to 5.886 kWh/m²/day. Seasonal analysis revealed higher electricity generation during the dry season due to lower cloud cover, rainfall, and humidity. For 100% PV penetration, annual electricity generation ranges from 886.13 GWh in Soma to 913.87 GWh in Banjul. Correlation analysis shows that cloud cover (r = −0.976) and precipitation (r = −0.944) have the greatest negative impact on PV electricity production. System losses were also found to range from 21.34% to 22.80%, mainly due to variations in temperature and radiation. Moreover, the economic analyses indicated that the cost of electricity of proposed systems is within the range of 50.77-52.94 USD/MWh with a payback period of less than six years. This demonstrates the economic viability of utility-scale photovoltaic solar energy systems. The results show that the use of large-scale photovoltaic solar power systems can help provide a cost-effective alternative to fossil fuel-based electricity production by government institutions in The Gambia.