International Journal of Autonomous and Adaptive Communications Systems (12 papers in press)

Regular Issues

• Open Access

  Motion Capture and Damage Recognition Method Based on Memetic Algorithm for Edge AI in Industry 5.0 Human Computer Collaboration ( Free Full-text Access )

  by Yuhang Li 
  Abstract: Traditional motion capture systems typically rely on optical or inertial sensors, which have limitations such as difficulty adapting to lightweight deployments at the edge. This article proposes a motion capture and recognition method based on the Memetic algorithm. First, the meme algorithm is combined with computer vision technology and edge computing architecture to capture human image sequences through edge deployed cameras. And use the location processing advantage of edge computing to reduce data transmission delay. Then, the Memetic algorithm is used to preprocess the image sequence and accurately extract key skeletal points of the human body. Solved the problem of low accuracy and poor anti-interference ability in extracting bone points in public scenes. Subsequently, standardised human motion data is generated by calculating the relative positions and angles between key skeletal points. Provide technical support for innovative applications of soft computing and edge intelligence in industrial and urban ecosystems.
  Keywords: Edge Artificial Intelligence; Soft Computing; Motion Capture; Industry 5.0; Lightweight Algorithms; Computer Collaboration.
  DOI: 10.1504/IJAACS.2026.10079240
   
  
• Spectrum Allocation in 6G Communication Networks Using Enhanced Dual Deep Q-Network  
  by Monika Gupta, Ishani Mishra, Sonika Katta, Pavithra G, Swapnil S. Ninawe, Preeti Khanwalkar 
  Abstract: In 6G communication networks, spectrum allocation process is crucial for effectively managing and utilising the limited spectrum resources available. However, traditional spectrum allocation approaches encounter significant challenges, particularly in dynamic environments; often result in suboptimal spectrum utilisation and delays in the allocation process. To overcome this issue, we propose an Enhanced Dual Deep Q-Net (EDDQN) scheme for spectrum allocation. EDDQN is an advanced version of DQN that makes use of two neural networks: an online network for action selection and a target network for Q-value estimation. By using two separate neural networks, EDDQN overcomes overestimation bias in Q-value estimation, a significant drawback of the original DQN. The zebra optimisation algorithm is used to optimise the hyper-parameters of EDDQN in order to improve its performance. Through extensive simulations, we prove the superiority of proposed spectrum allocation compared with traditional spectrum allocation approaches.
  Keywords: 6G communication network; spectrum allocation; enhanced dual deep Q-Net; online network; target network; and zebra optimization algorithm.
  DOI: 10.1504/IJAACS.2026.10075893
   
  
• Reversible Data Hiding in Encrypted Image via Block Classification and Adaptive Coding  
  by Yuanyuan Fu, Shiliang He, Jiaohua Qin 
  Abstract: Reversible data hiding in encrypted images (RDHEI) enables secret information embedding while ensuring error-free data extraction and perfect image recovery. However, the fixed vacating room method used to divide encrypted images into blocks results in limited embedding space. To address this limitation, this paper proposes a block classification and adaptive coding-based RDHEI scheme. Specifically, block-level encryption and permutation are first applied to the original image. Then, adaptive block classification is conducted, and block labels are recorded in a block label location map to achieve first-stage room vacating. For blocks with insufficient available space, adaptive coding and block label map compression are further employed to realise second-stage room vacating. Experimental results demonstrate that the proposed method achieves reversibility and separability while providing a higher embedding rate than existing approaches.
  Keywords: Reversible data hiding in encrypted image; Adaptive block classification; Adaptive coding strategy.
  DOI: 10.1504/IJAACS.2026.10076297
   
  
• A Review on Advanced Nanostructured Thin Films for High-Capacity and Long-Lasting Battery Technologies  
  by A. Rani Sangeetha, K. Visalakshi 
  Abstract: Nanostructured thin films represent a promising pathway for next-generation high-performance batteries due to their enhanced ion transport, large surface area, and improved electrochemical stability. This review summarises key advances reported between 2019 and 2025 in thin-film materials used for battery electrodes, electrolytes, and separators. Emphasis is placed on advanced fabrication techniques, including atomic layer deposition and solgel synthesis, together with modern characterisation methods. The influence of nanostructured morphology on electrochemical performance is critically examined, particularly in terms of chargedischarge efficiency, interface stability, and cycling durability. In addition, hybrid and multilayer architectures, such as composite and interface-engineered systems, are discussed for their ability to improve conductivity and energy density. Major challenges related to large-scale manufacturing, cost, and long-term structural integrity are identified, and potential mitigation strategies are outlined. Overall, the review highlights how integrating materials science, nanoengineering, and AI-driven design can accelerate the commercialisation of nanostructured thin-film battery technologies.
  Keywords: Nanostructured Thin Films; High-Capacity; Battery Technologies; Electrochemical Processes; Polyanionic Compound.
  DOI: 10.1504/IJAACS.2026.10076356
   
  
• Cross-Modal Attention for Fake News Detection: Integrating Text and Image Features with Multi-Modal Fusion and Advanced Methods  
  by Yamini Devi Jonnala, J. Sirisha Devi 
  Abstract: The rise of digital media accelerates the spread of fake news, impacting public opinion, politics, and health. Traditional detection methods analyse text and images separately, failing to capture cross-modal relationships. This study proposes a multimodal framework that processes text and image to address these challenges. The text data is first processed using bidirectional long short-term memory (BiLSTM) and CapsNet, where BiLSTM captures sequential dependencies and CapsNet with self-attention enhances spatial feature extraction. Simultaneously, image data is processed using ResNet and vision transformer (ViT) to extract meaningful features, creating a single representation. The extracted text and image features are then refined through multi-head self-attention. These refined features are fed into a cross-modal attention mechanism, which integrates and fuses the enhanced representations from both modalities the fused representation passes through a fully connected layer for classification. The specified models achieves a high accuracy of 96.2% and 95.2% for text and image inputs.
  Keywords: BiLSTM; CapsNet; ResNet,ViT; Fully connected Layer; Multiple Multi-Model; Multi-head self-attention and Cross-Modal Attention Network.
  DOI: 10.1504/IJAACS.2026.10077164
   
  
• A comprehensive review on DV-hop-based localisation and flooding routing protocols in underwater acoustic sensor networks  
  by Pankaj Singh Yadav, Pabitra Mohan Khilar 
  Abstract: Underwater acoustic sensor networks (UASN) tackle various marine applications through their essential tasks even though they deal with major hurdles including acoustic propagation delays alongside limited bandwidth and additional energy expenditures. UASN dependability relies heavily on accurate node localisation and efficient data dissemination. The distance vector-hop (DV-Hop) algorithm represents a popular location identification system because it provides straightforward implementation and requires minimal hardware deployment. Underwater conditions that are unpredictable often prompt operators to select flooding-based routing because of its reliability. Most approaches develop independently from each other even though the systems work together in practice. The paper delivers a comprehensive evaluation of DV-Hop-based localisation together with flooding-based routing protocols in UASNs while identifying their individual advances and remaining gaps and potential integration prospects. Multiple state-of-the-art algorithms currently suffer from three main shortcomings as they fail to handle network mobility, ignoring acoustic signal limitations and lack performance enhancement through cross-layer integration.
  Keywords: UASNs; underwater acoustic sensor networks; DV-hop localisation; flooding-based routing; mobility-aware protocols; energy-efficient communication; acoustic channel modelling.
  DOI: 10.1504/IJAACS.2026.10078117
   
  
• Optimizing Controller Placement for SDN using Self-Play Reinforcement Learning  
  by Ouafae Benoudifa, Abderrahim A.I.T. Wakrime, Redouane Benaini 
  Abstract: The placement of the Software-Defined Networking (SDN) controller is a critical component of SDN architecture, influencing the network's performance, scalability, reliability, and overall efficiency. The MuZero agent architecture offers an innovative approach to optimize coherence and efficiency in SDN controller placement. Known for its autonomous learning through decision-making and planning, the MuZero algorithm adapts to the complex challenge of controller placement. This study applies the MuZero agent to simulate scenarios where controllers are strategically assigned to nodes in an SDN, aiming to maximize network coherence while adhering to predefined constraints. The results demonstrate the successful optimization of controller placement, showcasing the effectiveness of the MuZero approach in enhancing SDN performance.
  Keywords: Artificial Intelligence; Software-Defined Networks; Data Sharing; Controller Placement Problem; MuZero Algorithm; Reinforcement Learning; Autonomous Learning.
  DOI: 10.1504/IJAACS.2026.10078475
   
  
• Public Cultural Policy Opinion Analysis Based on Multimodal Joint Attention Mechanism  
  by Dandan Liu 
  Abstract: The research on public opinion analysis of public cultural policies is helpful for timely warning of potential contradictions and risks in policy implementation. Therefore, a public cultural policy opinion analysis method based on multimodal joint attention mechanism is proposed. By improving the K-means algorithm to cluster the data collected by form focused web crawlers, a data anomaly detection model based on multimodal joint attention mechanism is established. Abnormal data is determined and removed through anomaly scores. Using the data after removing anomalies as input, output the results of public opinion classification and recognition, establish a BERT-BDCA model, and achieve public cultural policy public opinion analysis through multiple steps such as word embedding layer and attention processing. The experimental results show that the maximum recall rate of the proposed method for public opinion data reaches 98.75%, the maximum accuracy exceeds 98%, and the maximum time is only 16.31 minutes
  Keywords: Multimodal joint attention mechanism; Public cultural policies; Public opinion analysis; Improving the K-means algorithm; BERT-BDCA model.
  DOI: 10.1504/IJAACS.2026.10079211
   
  
• Developing a New Energy-Efficient Joint User and Power Allocation Framework using a Heuristic Algorithm-based Deep Network in Futuristic 5G Millimeter Wave Networks  
  by Subramani Gandhi, Nagarajan M, Sudhagar G 
  Abstract: Millimeter wave is a crucial role in developing Fifth Generation (5G) network and power allocation framework. Moreover the existing techniques limits from high power consumption, leading to increased energy usage. In order to solve the issues, the research study introduces future 5G mmWave networks to enhance energy-efficient joint user and power allocation. This work is considered the Base Station (BS) on/off switching mechanism. At first, optimal data generation is performed by Intellectual Frilled Lizard Optimization with Random Updates (IFLO-RU). The Adaptive Residual Autoencoder with Spatiot-Temporal Attention (ARAE-STA) is employed for optimizing the power allocation across distinct devices or users in a network. Some of the multi-objective functions such as the number of switched-off BSs, energy efficiency, network throughput, and power consumption are evaluated. Finally, the developed framework's performance analyzes the other related methods to confirm the efficacy of the implemented system.
  Keywords: Energy-Efficiency; Joint User and Power Allocation; Future 5G Millimeter Wave Networks;Intellectual Frilled Lizard Optimization with Random Updates; Adaptive Residual Autoencoder with spatiot-tempora.
  DOI: 10.1504/IJAACS.2026.10079558
   
  
• Graphene-Enabled Reconfigurable 2x2 MIMO dielectric resonator antenna for multiband THz applications  
  by Ritesh Kushwaha 
  Abstract: This work proposes a 2x2 frequency-reconfigurable multiple-inputmultiple- output (MIMO) dielectric resonator antenna (DRA) utilising graphene for multi-band operation in terahertz (THz) applications. The design integrates monolayer graphene into a novel feed structure, where chemical potential tuning enables four discrete switching states: OFFOFF, OFFON, ONOFF, and ONON. The antenna maintains efficient impedance matching across the 0.451.0 THz range, enabling effective dynamic control of the resonant frequencies. The DRA achieves high inter-element isolation (>30 dB), totalactive- reflection-coefficient (TARC) below 0.7 dB, and notable frequency agility with minimal reconfiguration complexity. MIMO performance metrics include an envelope correlation coefficient (ECC) below 0.003, diversity gain (DG) close to 10 dB, and balanced mean effective gain (MEG) at both ports. The channel capacity loss (CCLremains below 0.5 bits/s/Hz throughout most of the operating band, confirming the antennas potential for high-data-rate THz links. The antenna is suitable for THz adaptive wireless communication.
  Keywords: Reconfiguration Dielectric Resonator; Graphene; MIMO; TARC,CCL; THz band.
  DOI: 10.1504/IJAACS.2026.10079623
   
  
• Deep Learning Optimized Cloverleaf MIMO Antenna for Enhanced Wireless Connectivity in Smart Agriculture  
  by Ogirala Pranitha, K.V. Prashanth, Jagadeesh Chandra Prasad Matta, Hema Chandra Rao Bitra, Ramesh Babu Juturi 
  Abstract: Smart farming systems demand reliable, high-speed wireless networks for seamless data transmission and real-time monitoring. However, existing solutions face limitations such as narrow bandwidth, high mutual coupling, low gain, and poor spectral and radiation efficiency. To overcome these challenges, this research presents a Deep Learning Optimized Cloverleaf MIMO antenna integrated with an Extended Long Short-Term Memory (E-LSTM) model, fine-tuned using the Gazelle Optimization Algorithm (GOA). The Cloverleaf antenna delivers compact, multi-band operation with high isolation, achieving a broad 214.8 GHz bandwidth, 9.1 dB peak gain, 52% radiation efficiency, and mutual coupling below 15 dB. Additionally, the system achieves a spectral efficiency of 10 bps/Hz at 15 dB signal-to-noise ratio and channel capacity loss (CCL) below 0.35 bits/s/Hz. This framework enhances predictive performance for agricultural data. Collectively, the proposed system offers an intelligent, scalable wireless infrastructure tailored to meet the advanced communication needs of smart agriculture.
  Keywords: Smart Agriculture; Deep learning; Gazelle Optimization Algorithm; MIMO Antenna; Long Short-Term Memory.
  DOI: 10.1504/IJAACS.2026.10079653
   
  
• A New Approach to Energy-Efficient Routing in WSNs Using a Modified LEACH Protocol  
  by Mahendra Dongare, Satish Jondhale, Balasaheb Agarkar 
  Abstract: In wireless sensor networks (WSNs), hierarchical clustered routing protocols are pivotal in optimising energy utilisation. The low-energy adaptive clustering hierarchy (LEACH) contributes to higher energy depletion if rotation of cluster heads is not systematically managed. To address this limitation, we introduce an enhanced routing strategy as average energy and residual energy-based modified LEACH (aerem-LEACH); designed to enhance the energy efficiency of WSNs by simultaneously considering both the average network energy and the residual energy of individual nodes during CH selection process. It determines ideal number of cluster heads, restricts nodes located near the sink from forming clusters to avoid excessive energy burden, and introduces a novel threshold mechanism for more effective CH selection. Additionally, the protocol leverages a hybrid communication model including free space propagation, multi-hop routing, and adaptive data transmission ensuring minimal energy usage. Proposed aerem-LEACH achieves a network lifetime enhancement from 9% to 57% compared to existing protocols.
  Keywords: Low-energy adaptive clustering hierarchy(LEACH); average energy residual energy based modified LEACH (aerem-LEACH); Stable Energy Efficient Network (SEEN); LEACH-Mobile (LEACH-M); LEACH-Centralized.
  DOI: 10.1504/IJAACS.2026.10079657