Search

The advances in communication networks and their integration with machine learning technology have paved the way for ubiquitous and prediction-based services for consumers. However, these services consider sensitive and private data for training a machine learning model. With the emergence of model inversion and poisoning attacks, sensitive and private data can be leaked, which is a hindrance for the realization of largescale automation services concerning communication networks. Zero-trust techniques allow the networks to rate the data for their participation in service provisioning tasks, but existing works do not consider model privacy for the zero-trust services. This article proposes a Self-supervised Learning based model Inversion and Poisoning (SLIP) detection framework that enables the rating of model so that network could decide whether the model is suitable for service provisioning or has been compromised. The framework leverages several Generative AI technologies such as generative adversarial networks (GANs) and diffusion models, to realize its implementation in federated learning setting. Experimental results show that the SLIP framework helps in reducing model inversion and poisoning attacks by 16.4% and 13.2% for vehicular networks, respectively.

Cellular vehicle-to-everything (C-V2X) was introduced to support autonomous driving through 5G and beyond networks. C-V2X leverages cellular network infrastructure to integrate vehicle-to-network, vehicle-to-pedestrian, vehicle-to-infrastructure, and vehicle-to-vehicle communications. It has been suggested that Open RAN can be used to achieve the latency requirements essential to realize C-V2X as it achieves real-time optimization through the use of AI in Near real-time RAN Intelligence Controller (Near-RT RIC). The Open RAN will allow the access to historical traffic data or acquisition of data from vehicles. The data will then be transferred to Near-RT RIC for detecting network anomalies while maintaining reliable communication, which is essential for realizing autonomous driving. Open RAN also supports non-real-time RAN intelligent controller (Non-RT RIC) that allows more complex ML workflows such as policy-based feature extraction and optimization to guide vehicles when real-time acquisition is not available. Open RAN provides support for edge cloud, i.e. Open Cloud that helps to interface the Near-RT RIC with Open RAN central unit's user and control plane. Together, the Open RAN and C-V2X are considered to be the key-enabling technologies for achieving low-latency in autonomous vehicular communication networks. The issue attracted over 100 high-quality submissions from all over the world, among which 12 original contributions were eventually selected for publication. The novelty and key contributions of these articles are summarized as follows.

Future personalization services are key to Industry 5.0. Also known as the fifth industrial revolution, it leverages artificial intelligence and cognitive services for cooperation between machines and humans. This collaboration aims to increase consumer value while improving industrial automation. Industry 5.0 focuses more on cooperation than competition. Personalized services are provided by linking data from various sensors to highlight distinct product characteristics based on consumer preferences. A high-tech strategy uses the Internet of Things to provide industrial automation along with personalized consumer services through symmetric innovation via democratized Big Data knowledge.

Internet of things (IoT) has emerged as a promising technology that can be widely used in various industries to realize real-time information collection, so as to improve production efficiency and reduce running costs. By combining the technology of IoT, maritime transportation systems (MTS) can prevent vessels collision, improve the efficiency of maritime transportation and reduce the loss of revenue for ports and shipbuilders. The large amount of real-time data generated in IoT-enabled MTS can be efficiently utilized to predict the future trajectories and hotspots of vessels on the sea combined with historical data. However, the maritime traffic data in MTS cannot be effectively processed in traditional big data analysis methods, and the integrity of it needs to be checked before being used to achieve the prediction of trajectories and high-density areas of vessels. In this paper, we propose a flexible data integrity checking scheme with original data recovery in IoT-enabled MTS. In the proposed scheme, the data blocks of vessels are encoded based on the technology of erasure coding. To ensure the availability of the historical data, the existence and the integrity of the data elements stored in the cloud can be checked. Moreover, the original data blocks can be recovered efficiently if the encoded data elements have been corrupted or deleted. Security analysis demonstrates that the proposed scheme can be proved to be correct and is secure against malicious attacks. Performance analysis shows that our scheme is more efficient than the previous schemes.

Multimodal medical images are widely used by clinicians and physicians to analyze and retrieve complementary information from high-resolution images in a non-invasive manner. Loss of corresponding image resolution adversely affects the overall performance of medical image interpretation. Deep learning-based single image super resolution (SISR) algorithms have revolutionized the overall diagnosis framework by continually improving the architectural components and training strategies associated with convolutional neural networks (CNN) on low-resolution images. However, existing work lacks in two ways: i) the SR output produced exhibits poor texture details, and often produce blurred edges, ii) most of the models have been developed for a single modality, hence, require modification to adapt to a new one. This work addresses (i) by proposing generative adversarial network (GAN) with deep multi-attention modules to learn high-frequency information from low-frequency data. Existing approaches based on the GAN have yielded good SR results; however, the texture details of their SR output have been experimentally confirmed to be deficient for medical images particularly. The integration of wavelet transform (WT) and GANs in our proposed SR model addresses the aforementioned limitation concerning textons. While the WT divides the LR image into multiple frequency bands, the transferred GAN uses multi-attention and upsample blocks to predict high-frequency components. Additionally, we present a learning method for training domain-specific classifiers as perceptual loss functions. Using a combination of multi-attention GAN loss and a perceptual loss function results in an efficient and reliable performance. Applying the same model for medical images from diverse modalities is challenging, our work addresses (ii) by training and performing on several modalities via transfer learning. Using two medical datasets, we validate our proposed SR network against existing state-of-the-art approaches and achieve promising results in terms of structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR).

The provision of physical healthcare services during the isolation phase is one of the major challenges associated with the current COVID-19 pandemic. Smart healthcare services face a major challenge in the form of human behavior, which is based on human activities, complex patterns, and subjective nature. Although the advancement in portable sensors and artificial intelligence has led to unobtrusive activity recognition systems, very few studies deal with behavior tracking for addressing the problem of variability and behavior dynamics. In this regard, we propose the fusion of PRocess mining and Paravector Tensor (PROMPT)-based physical health monitoring framework that not only tracks subjective human behavior, but also deals with the intensity variations associated with inertial measurement units. Our experimental analysis of a publicly available dataset shows that the proposed method achieves 14.56% better accuracy in comparison to existing works. We also propose a generalized framework for healthcare applications using wearable sensors and the PROMPT method for its triage with physical health monitoring systems in the real world.

With the advances in 5G and IoT devices, industries are vastly adopting artificial intelligence (AI) techniques for improving classification and prediction-based services. However, the use of AI also raises concerns regarding data privacy and security that can be misused or leaked. Private AI was recently coined to address the data security issue by combining AI with encryption techniques, but existing studies have shown that model inversion attacks can be used to reverse engineer the images from model parameters. In this regard, we propose a federated learning and encryption-based private (FLEP) AI framework that provides two-tier security for data and model parameters in an Industrial IoT environment. We propose a three-layer encryption method for data security and provided a hypothetical method to secure the model parameters. Experimental results show that the proposed method achieves better encryption quality at the expense of slightly increased execution time. We also highlight several open issues and challenges regarding the FLEP AI framework's realization.

Human action recognition from skeleton data has drawn a lot of attention from researchers due to the availability of thousands of real videos with many challenges. Existing works attempted to model the spatial characteristics and temporal dependencies of 3D joints using dynamic time warping, hand-crafted, and spatial co-occurrence features. However, the representation derived from the spatial stream overemphasizes the temporal information; thus, it yields limited expressive power. Some studies use skeleton sequences as frames to enhance the expressive power of representations but lose the generalization capability because the derived temporal smoothness is specific to a particular dataset. The proposed work uses joint distance maps as a base representation that encodes the spatial and temporal information to color texture images. We increase the expressive power by extracting the feature maps from pre-trained networks on ImageNet to diversify the texture representation and propose a network architecture to model the temporal dependency explicitly. We also explore various fusion strategies to generate diverse representations from the feature maps of the pre-trained networks. The experimental results show that the proposed method achieves the best recognition accuracy when using decision-level fusion with meta-learners (Random Forest). The analysis also reveals that the use of feature-level fusion yields relatively good results in terms of the trade-off, i.e., on par recognition performance with some decision-level fusion strategies while having less tunable parameters. Extensive experimental results and comparative analysis on three benchmark datasets prove that the proposed representation and network not only yield better recognition accuracy but also exhibit stronger generalization capability on multiple datasets.

The provision of communication services via portable and mobile devices, such as aerial base stations, is a crucial concept to be realized in 5G/6G networks. Conventionally, IoT/edge devices need to transmit data directly to the base station for training the model using machine learning techniques. The data transmission introduces privacy issues that might lead to security concerns and monetary losses. Recently, federated learning was proposed to partially solve privacy issues via model sharing with the base station. However, the centralized nature of federated learning only allows the devices within the vicinity of base stations to share trained models. Furthermore, the long-range communication compels the devices to increase transmission power, which raises energy efficiency concerns. In this work, we propose the distributed federated learning (DBFL) framework that overcomes the connectivity and energy efficiency issues for distant devices. The DBFL framework is compatible with mobile edge computing architecture that connects the devices in a distributed manner using clustering protocols. Experimental results show that the framework increases the classification performance by 7.4 percent in comparison to conventional federated learning while reducing the energy consumption.

The recognition of fine-grained emotions (i.e., happiness, sad, etc.) has shown its importance in a real-world implementation. The emotion recognition using physiological signals is a challenging task due to the precision of the labelled data while using facial expressions is less appropriate for the real environment. This work proposes a framework for fusing physiological signals and facial expressions modalities to improve classification performance. The feature-level fusion (FLF) and decision-level fusion (DLF) techniques are explored in this work to recognise seven fine-grained emotions. The performance of the proposed framework is evaluated using 34 subjects' data. Our result shows that the fusion of the multiple modalities could improve the overall accuracy compared to the unimodal system by 11.66% and 13.63% for facial expression and physiological signals, respectively. Our work achieved a 73.23% accuracy for seven emotions which is considerable accuracy for the spontaneous emotion corpus.

ABSTRACTStock movement prediction is regarded as one of the most difficult, meaningful, and attractive research issues in the field of financial markets. The stock price data have non-stationary, noisy, and non-linear characteristics which make the movement and its prediction a challenging task. In this paper, we propose a framework to predict the stock price movement using phase space reconstruction (PSR) and extreme learning machines (ELM). The uniqueness of the framework is reflected by its feature transformation technique which computes the information distance from the transformed features in phase space. The distance from phase space dimensions are modelled with ELM to predict the stock price movement. A decision-level fusion is performed on the ELM models trained using each category of features to improve the prediction performance. The framework has been validated on one of the challenging Borsa Istanbul (BIST 100) dataset which is a widely used dataset in stock price prediction studies. The results from the proposed framework are compared with the conventional machine learning pipeline as well as the baseline methods, i.e., random and Naïve approach to show the effectiveness in prediction performance. Experimental results reveal that the framework improves predictive performance by 4.5% in terms of F-measure values.