Search

This comprehensive review paper examines the challenges faced by blockchain technology in terms of scalability and proposes potential solutions and future research directions. Scalability poses a significant hurdle for Bitcoin and Ethereum, manifesting as low throughput, extended transaction delays, and excessive energy consumption, thereby compromising efficiency. The current state of blockchain scalability is analyzed, encompassing the limitations of existing solutions such as Sharding and off-chain scaling. Various proposed remedies, including layer 2 scaling solutions, consensus mechanisms, and alternative approaches, are investigated. The paper also explores the impact of scalability on diverse blockchain applications and identifies potential future research directions by integrating data science techniques with blockchain technology. Notably, nearly 110 primary research papers from reputable scientific databases like Scopus, IEEE Explore, Science Direct, and Web of Science were reviewed, demonstrating scalability in blockchain comprising several elements. Transaction throughput and network latency emerge as the most prominent concerns. Consequently, this review offers future research avenues to address scalability challenges by leveraging data science techniques like distributed computing and parallel processing to divide and process vast datasets across multiple machines. The synergy between data science and blockchain holds promise as an optimal solution. Overall, this up-to-date understanding of blockchain scalability is invaluable to researchers, practitioners, and policy makers engaged in this domain. [ABSTRACT FROM AUTHOR]

Distributed computing systems have been widely used as the amount of data grows exponentially in the era of information explosion. Job completion time (JCT) is a major metric for assessing their effectiveness. How to reduce the JCT for these systems through reasonable scheduling has become a hot issue in both industry and academia. Data skew is a common phenomenon that can compromise the performance of such distributed computing systems. This paper proposes SMART, which can effectively reduce the JCT through handling the data skew during the reducing phase. SMART predicts the size of reduce tasks based on part of the completed map tasks and then enforces largest-first scheduling in the reducing phase according to the predicted reduce task size. SMART makes minimal modifications to the original Hadoop with only 20 additional lines of code and is readily deployable. The robustness and the effectiveness of SMART have been evaluated with a real-world cluster against a large number of datasets. Experiments show that SMART reduces JCT by up to 6.47%, 9.26%, and 13.66% for Terasort, WordCount and InvertedIndex respectively with the Purdue MapReduce benchmarks suite (PUMA) dataset. [ABSTRACT FROM AUTHOR]

Wireless sensor networks which are based on distributed information processing technology are taking an increasingly key role in industrial digital twin scenarios. There are many important issues in the access of networks. One of the most important issues is how to improve network access control and the effectiveness of load balancing. Based on the industrial digital twin technology, this article first introduces several typical network access and network loads and performs tree-structured processing on the outliers generated during the chain formation process to reduce the length of the data transmission path, optimize the main chain head and subchain chain head selection strategy and chaining rules, and perform nonchain operations on common nodes and chain heads near sink to reduce data inverse transfer. The experimental results show that this paper uses the digital twin calculation formula to accurately and objectively determine the remaining cluster head and the distance head and the base station, so that when the node distance is limited, the network energy consumption can be balanced as much as possible, and the network load is promoted. [ABSTRACT FROM AUTHOR]

The article presents the discussion on paradigm of virtualization evolving from a virtualized local area network (LAN) and private networks to the solidification of network function virtualization (NFV). Topics include coupled with the pervasive utilization of artificial intelligence (AI) at all network levels; and the Digital Twin (DT) paradigm recently deemed as a promising tool for network design, optimization, management.

Cloud computing is a prominent and evolving distributed computing paradigm that provides users with on-demand services through a network of diverse autonomous systems with flexible computational structures. The significance of task scheduling becomes evident, serving as a vital component to elevating cloud computing's overall performance. Streamlining cost-effective execution and optimizing resource utilization is a key objective, given the NP-hard nature of the task scheduling problem. Although numerous meta-heuristic techniques have been explored to address task allocation challenges, ample opportunities remain for the development of optimal strategies. This paper presents a state-of-the-art task assignment model that revolves around OptiAssign particle swarm optimization (PSO), with a strong emphasis on the crucial role played by efficient dependency handling and multi-level task scheduling. The primary aim of this model is to optimize the utilization of virtual machine capacities, simultaneously minimizing execution time, makespan, wait time, and overall execution costs within a variety of distributed computing systems. This novel algorithm showcases outstanding performance when compared to traditional approaches in task scheduling, highlighting the importance of skillful dependency management and the implementation of multi-level task scheduling strategies. The results of this study further affirm the effectiveness of the model in addressing the inherent complexities of scenarios involving intricate task dependencies and diverse scheduling priorities. [ABSTRACT FROM AUTHOR]

In recent years, Internet of Things security incidents occur frequently, which is often accompanied by malicious events. Therefore, anomaly detection is an important part of Internet of Things security defense. In this paper, we create a process whitelist based on the K-Core decomposition method for detecting anomalous processes in IoT devices. The method first constructs an IoT process network according to the relationships between processes and IoT devices. Subsequently, it creates a whitelist and detect anomalous processes. Our work innovatively transforms process data into a network framework, employing K-Core analysis to identify core processes that signify high popularity. Then, a threshold-based filtering mechanism is applied to formulate the process whitelist. Experimental results show that the unsupervised method proposed in this paper can accurately detect anomalous processes on real-world datasets. Therefore, we believe our algorithm can be widely applied to anomaly process detection, ultimately enhancing the overall security of the IoT. [ABSTRACT FROM AUTHOR]

However, as host-switch graph cannot represent such systems, the authors investigate interconnection networks with multi-port hosts by introducing a multi-ported host-switch graph. The past years have seen a flurry of activity in the area of distributed computing and its related fields. The accepted papers for this special issue address the following aspects in computing and networking: Parallel and distributed computing Security and privacy Interconnection networks Artificial intelligence Self-stabilizing algorithms Parallel and distributed computing The first contribution, entitled "Efficient Parallel Implementations to Compute the Diameter of a Graph" by Takafuji et al., presents an efficient implementations of the Blocked Floyd-Warshall algorithm, which uses no barrier synchronization and invokes only one kernel call by using Single Kernel Soft Synchronization (SKSS) techniques. [Extracted from the article]

This special section focuses on the state-of-the-art technologies on parallel and distributed computing techniques for artificial intelligence (AI), machine learning (ML), and deep learning (DL). AI, ML, and DL can enable computers the ability to learn from a large amount of data and use the learned model to optimize a complex problem or discover rules in a complicated system. AI, ML and DL can be applied to push forward the boundaries for many domains and significantly influence our daily life. [ABSTRACT FROM AUTHOR]

Consensus-based data aggregation in d-regular bipartite graphs poses a challenging task for the scientific community since some of these algorithms diverge in this critical graph topology. Nevertheless, one can see a lack of scientific studies dealing with this topic in the literature. Motivated by our recent research concerned with this issue, we provide a comparative study of frequently applied consensus algorithms for distributed averaging in d-regular bipartite graphs in this paper. More specifically, we examine the performance of these algorithms with bounded execution in this topology in order to identify which algorithm can achieve the consensus despite no reconfiguration and find the best-performing algorithm in these graphs. In the experimental part, we apply the number of iterations required for consensus to evaluate the performance of the algorithms in randomly generated regular bipartite graphs with various connectivities and for three configurations of the applied stopping criterion, allowing us to identify the optimal distributed consensus algorithm for this graph topology. Moreover, the obtained experimental results presented in this paper are compared to other scientific manuscripts where the analyzed algorithms are examined in non-regular non-bipartite topologies. [ABSTRACT FROM AUTHOR]

We consider standard T-interval dynamic networks, under the synchronous timing model and the broadcast CONGEST model. In a T-interval dynamic network, the set of nodes is always fixed and there are no node failures. The edges in the network are always undirected, but the set of edges in the topology may change arbitrarily from round to round, as determined by some adversary and subject to the following constraint: For every T consecutive rounds, the topologies in those rounds must contain a common connected spanning subgraph. Let H r to be the maximum (in terms of number of edges) such subgraph for round r through r + T - 1 . We define the backbone diameterd of a T-interval dynamic network to be the maximum diameter of all such H r 's, for r ≥ 1 . We use n to denote the number of nodes in the network. Within such a context, we consider a range of fundamental distributed computing problems including Count/Max/Median/Sum/LeaderElect/Consensus/ConfirmedFlood. Existing algorithms for these problems all have time complexity of Ω (n) rounds, even for T = ∞ and even when d is as small as O(1). This paper presents a novel approach/framework, based on the idea of massively parallel aggregation. Following this approach, we develop a novel deterministic Count algorithm with O (d 3 log 2 n) complexity, for T-interval dynamic networks with T ≥ c · d 2 log 2 n . Here c is a (sufficiently large) constant independent of d, n, and T. To our knowledge, our algorithm is the very first such algorithm whose complexity does not contain a Θ (n) term. This paper further develops novel algorithms for solving Max/Median/Sum/LeaderElect/Consensus/ConfirmedFlood, while incurring O (d 3 polylog (n)) complexity. Again, for all these problems, our algorithms are the first ones whose time complexity does not contain a Θ (n) term. [ABSTRACT FROM AUTHOR]

The precise evaluation of micro-crack sizes and locations is crucial for the safe operation of structures. Traditional detection techniques, however, suffer from low spatial resolution, making it difficult to accurately locate micrometer-scale cracks. A method and experimental study were proposed in this paper for identifying and locating micro-cracks using optical fiber strain sensing based on OFDR to address this issue. The feasibility of this method for micro-crack detection was verified by the combination of a polyimide-coated sensing optical fiber (PISOF) and tight sheath sensing optical fiber (TSSOF). A calculation method for micro-crack widths based on distributed optical fiber strain curves was established, and the test results of different optical fibers were compared. Through multiple verification experiments, it was found that the strain peak curves of both fiber types could accurately locate micro-cracks with a precision of 1 mm. Additionally, the crack widths could be obtained by processing the distributed strain curves using a computational model, enabling the accurate capture of micro-crack characteristics at the 10 μm level. A strong linear relationship was observed between the optical fiber stretching length and the crack width. Notably, the relative error in calculating the crack width from the strain curve of PI fiber was very small, while a linear relationship existed between the maximum strain value of the TSSOF and the crack width, allowing for the calculation of the crack width based on the maximum strain value. This further validated the feasibility of the method designed in this paper for the analysis of micro-crack characteristic parameters. [ABSTRACT FROM AUTHOR]

The concept of the meta-universe is still in its early stages, but many leading tech companies have invested heavily in research and development for this technology. The development of meta-smart cities is a significant trend. In the meta-universe environment, integrating information systems is crucial for analyzing AI big data. Establishing an integrated platform for medical information systems is key to advancing information technology. In the context of the meta-universe, creating an efficient and unified integration platform to eliminate medical information silos and reduce system integration costs has become a pressing issue in medical informatization. This paper proposes a medical information system integration method based on an integration platform and utilizing cloud computing technology as a data center. The core business layer uses the integration software "Ensemble" as the integration platform. The underlying data center employs a Hadoop storage cluster with distributed data storage and parallel computing technology, and the existing scheduling algorithm is studied and analyzed to enhance the resource scheduling algorithm for medical small file data. The effectiveness of the algorithm is simulated and verified on an experimental platform, demonstrating improved efficiency in resource scheduling. [ABSTRACT FROM AUTHOR]

Pretrained language models have achieved great success in various natural language understanding (NLU) tasks due to their capacity to capture deep contextualized information in text using pretraining on large-scale corpora. Tokenization plays a significant role in the process of lexical analysis. Tokens become the input for other natural language processing (NLP) tasks, like semantic parsing and language modeling. However, there is a lack of research on the evaluation of the impact of tokenization on the Arabic language model. Therefore, this study aims to address this gap in the literature by evaluating the performance of various tokenizers on Arabic large language models (LLMs). In this paper, we analyze the differences between WordPiece, SentencePiece, and BBPE tokenizers by pretraining three BERT models using each tokenizer while measuring the performance of each model on seven different NLP tasks using 29 different datasets. Overall, the model pretrained with text tokenized using the SentencePiece tokenizer significantly outperforms the other two models that utilize WordPiece and BBPE tokenizers. The results of this paper will assist researchers in developing better models, making better decisions in selecting the best tokenizers, improving feature engineering, and making models more efficient, thus ultimately leading to advancements in various NLP applications. [ABSTRACT FROM AUTHOR]

With the ever-increasing advancement in blockchain technology, security is a significant concern when substantial investments are involved. This paper explores known smart contract exploits used in previous and current years. The purpose of this research is to provide a point of reference for users interacting with blockchain technology or smart contract developers. The primary research gathered in this paper analyses unique smart contracts deployed on a blockchain by investigating the Solidity code involved and the transactions on the ledger linked to these contracts. A disparity was found in the techniques used in 2021 compared to 2023 after Ethereum moved from a Proof-of-Work blockchain to a Proof-of-Stake one, demonstrating that with the advancement in blockchain technology, there is also a corresponding advancement in the level of effort bad actors exert to steal funds from users. The research concludes that as users become more wary of malicious smart contracts, bad actors continue to develop more sophisticated techniques to defraud users. It is recommended that even though this paper outlines many of the currently used techniques by bad actors, users who continue to interact with smart contracts should consistently stay up to date with emerging exploitations. [ABSTRACT FROM AUTHOR]

In the dynamic realm of big data processing, conquering the challenges imposed by highdimensional datasets is imperative. This paper introduces a groundbreaking advancement in dimensionality reduction, employing Variational Auto-Encoder (VAE) within the Spark distributed framework. The deliberate selection of the "TLC" dataset, representative of New York City taxi trips with inherent high dimensionality, highlights the practicality of our approach. Our research showcases the virtuoso performance of VAE, achieving an impressive 95.12% reduction ratio and 89.26% accuracy. This highlights VAE's ability to elegantly distill essential information while discarding superfluous dimensions, achieving a harmonious balance between reduction and accuracy. Furthermore, building on the demonstrated superiority of Spark over Hadoop in prior successes, our adoption of VAE aligns with the overarching goal of enhancing big data processing. Spark's consistent advantage as a distributed framework reaffirms its reliability in handling diverse machine learning algorithms. This paper not only contributes to the advancement of machine learning in big data processing but also underscores the adaptability, versatility, and consistent performance of our approach across various methodologies and frameworks. The success of VAE in reducing dimensionality, coupled with Spark's inherent advantages, positions this research as a valuable contribution to the exploration of advanced techniques in distributed big data processing. [ABSTRACT FROM AUTHOR]

This editorial presents a summary of the Special Issue on Artificial Intelligence Solutions and Applications for Distributed Systems in Smart Spaces, presented in the "Computer Science & Engineering" section of I Electronics i (ISSN 2079-9292). The application of artificial intelligence (AI) in distributed environments has become a research area of high-added value and economic potential. The Present Issue This Special Issue consists of seven papers that address pivotal topics in the field of AI solutions for distributed systems and their applications. [Extracted from the article]

An editorial is presented on deep learning and big data analytics for processing and analyzing data in 5G and 6G applications. Topics include research activities and areas such as bioinformatics, beyond 5G and 6G communications, healthcare, internet of things (IoT), manufacturing business, and social networks; and providing services with massive connectivity, ultra-low latency, extremely high security, extremely low energy consumption, and ultra-high data-rate.

Internet of Things (IoT) devices often operate with limited resources while interacting with users and their environment, generating a wealth of data. Machine learning models interpret such sensor data, enabling accurate predictions and informed decisions. However, the sheer volume of data from billions of devices can overwhelm networks, making traditional cloud data processing inefficient for IoT applications. This paper presents a comprehensive survey of recent advances in models, architectures, hardware, and design requirements for deploying machine learning on low-resource devices at the edge and in cloud networks. Prominent IoT devices tailored to integrate edge intelligence include Raspberry Pi, NVIDIA's Jetson, Arduino Nano 33 BLE Sense, STM32 Microcontrollers, SparkFun Edge, Google Coral Dev Board, and Beaglebone AI. These devices are boosted with custom AI frameworks, such as TensorFlow Lite, OpenEI, Core ML, Caffe2, and MXNet, to empower ML and DL tasks (e.g., object detection and gesture recognition). Both traditional machine learning (e.g., random forest, logistic regression) and deep learning methods (e.g., ResNet-50, YOLOv4, LSTM) are deployed on devices, distributed edge, and distributed cloud computing. Moreover, we analyzed 1000 recent publications on "ML in IoT" from IEEE Xplore using support vector machine, random forest, and decision tree classifiers to identify emerging topics and application domains. Hot topics included big data, cloud, edge, multimedia, security, privacy, QoS, and activity recognition, while critical domains included industry, healthcare, agriculture, transportation, smart homes and cities, and assisted living. The major challenges hindering the implementation of edge machine learning include encrypting sensitive user data for security and privacy on edge devices, efficiently managing resources of edge nodes through distributed learning architectures, and balancing the energy limitations of edge devices and the energy demands of machine learning. [ABSTRACT FROM AUTHOR]

Thanks to technological advancements, our lives are getting more intertwined with the connected world as more smart devices are coming to market. As such, the clear separation of devices as things (end-devices in IoT systems) and human operated devices is getting increasingly buried. This led to the creation of the term Internet of Everything (IoE) which is defined by Cisco as the “the networked connection of people, process, data, and things” [1]. The main difference between IoT and IoE is inclusion of people in the ecosystem, which greatly increases the number of connected parties. This increase in connected parties creates a strain on our existing infrastructure which is relying on cloud computing for performing most operations. Even though this resource provides heaps of computational power, the weak link in this scenario is the network, where all the connected devices can easily overload the available bandwidth, leading to slow response speeds and low general availability. The answer to this problem lies with technology that already exists and is not yet fully exploited as a distributed computing powerhouse, IoT. This paper aims to summarise the concept of computing at the edge, common architectural patterns, existing solutions, while also discussing real-world applications. [ABSTRACT FROM AUTHOR]

In the realm of distributed computing frameworks, such as Apache Spark and MapReduce Hadoop, the efficacy of these frameworks varies across diverse applications and algorithms contingent upon distinctive evaluation metrics and critical parameters. This research paper diligently scrutinizes the extant body of research that compares these two frameworks concerning said evaluation metrics and parameters. Subsequently, it conducts empirical investigations to authenticate the performance of these frameworks in the context of an iterative Gradient Boosting Tree Regression (GBTR) algorithm. Remarkably, the comparative analyses in previous studies encompass a spectrum of iterative machine learning regression and classification techniques, batch processing, SQL, and Graph processing algorithms. Furthermore, numerous investigations have explored the application of machine learning algorithms encompassing logistic regression, Page Rank, K-Means, KNN, and the HiBench suite. This paper presents the comparison between the two distributed computing platforms on iterative GBTR for classification task on the HIGGS dataset from the physics domain and for the regression task on the Covid-19 dataset from the healthcare domain. The empirical findings corroborate that Apache Spark exhibits superior execution speed in iterative tasks when the available physical memory significantly exceeds the dataset size. Conversely, Hadoop outperforms Spark when dealing with substantial datasets or constrained physical memory resources. [ABSTRACT FROM AUTHOR]

Brain tumors come in various types, each with distinct characteristics and treatment approaches, making manual detection a time-consuming and potentially ambiguous process. Brain tumor detection is a valuable tool for gaining a deeper understanding of tumors and improving treatment outcomes. Machine learning models have become key players in automating brain tumor detection. Gradient descent methods are the mainstream algorithms for solving machine learning models. In this paper, we propose a novel distributed proximal stochastic gradient descent approach to solve the L1-Smooth Support Vector Machine (SVM) classifier for brain tumor detection. Firstly, the smooth hinge loss is introduced to be used as the loss function of SVM. It avoids the issue of nondifferentiability at the zero point encountered by the traditional hinge loss function during gradient descent optimization. Secondly, the L1 regularization method is employed to sparsify features and enhance the robustness of the model. Finally, adaptive proximal stochastic gradient descent (PGD) with momentum, and distributed adaptive PGD withmomentum(DPGD) are proposed and applied to the L1-Smooth SVM. Distributed computing is crucial in large-scale data analysis, with its value manifested in extending algorithms to distributed clusters, thus enabling more efficient processing ofmassive amounts of data. The DPGD algorithm leverages Spark, enabling full utilization of the computer's multi-core resources. Due to its sparsity induced by L1 regularization on parameters, it exhibits significantly accelerated convergence speed. From the perspective of loss reduction, DPGD converges faster than PGD. The experimental results show that adaptive PGD withmomentumand its variants have achieved cutting-edge accuracy and efficiency in brain tumor detection. Frompre-trained models, both the PGD andDPGD outperform other models, boasting an accuracy of 95.21%. [ABSTRACT FROM AUTHOR]

As a typical application of machine learning in the mobile field of security and privacy protection, the main characteristics of arts and crafts materials are to advocate the practicality and functionality of design, emphasizing the unity of practicality and aesthetics, and practicality is the first. The so-called practicality refers to whether the designed product meets the specific functional requirements or aesthetic requirements, and whether it is convenient, comfortable and safe to use. Based on computer vision, this paper studies the material and aesthetic characteristics of arts and crafts works, and the research shows that this method has achieved the best performance. More specifically, the maximum expected value of this method is 0.69, which is much higher than the corresponding values of 0.54 and 0.50 in references [1] and [2] respectively. The experimental results show that our method is specially designed for the distributed learning of process aesthetics, and it is still a very effective tool for the aesthetic classification task of process aesthetic feature images. Based on computer vision, this paper studies the creative style and aesthetic characteristics of arts and crafts materials, and makes a comprehensive investigation and beneficial exploration on their development, aesthetic characteristics and creative style. [ABSTRACT FROM AUTHOR]

Recently, there has been a surge in interest surrounding the field of distributed edge computing resource scheduling. Notably, applications like intelligent traffic systems and Internet of Things (IoT) intelligent monitoring necessitate the effective scheduling and migration of distributed resources. In addressing this challenge, distributed resource scheduling must weigh the costs associated with resource scheduling, aiming to identify an optimal strategy amid various feasible solutions. Different application scenarios introduce diverse optimization objectives, including considerations such as cost, transmission delay, and energy consumption. While current research predominantly focuses on the optimization problem of local resource scheduling, there is a recognized need for increased attention to global resource scheduling. This paper contributes to the field by defining a global resource scheduling problem for distributed edge computing, demonstrating its NP-Hard nature through proof. To tackle this complex problem, the paper proposes a heuristic solution strategy based on the ant colony algorithm (ACO), with optimization of ACO parameters achieved through the use of particle swarm optimization (PSO). To assess the effectiveness of the proposed algorithm, an experimental comparative analysis is conducted. The results showcase the algorithm's notable accuracy and efficient iteration cost performance, highlighting its potential applicability and benefits in the realm of distributed edge computing resource scheduling. [ABSTRACT FROM AUTHOR]