Computer Science Review最新文献_第2页

Neuro-symbolic agentic AI: Architectures, integration patterns, applications, open challenges and future research directions 神经符号人工智能：架构、集成模式、应用、开放挑战和未来研究方向

IF 12.9 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-31 DOI: 10.1016/j.cosrev.2026.100902

Safayat Bin Hakim, Muhammad Adil, Alvaro Velasquez, Houbing Herbert Song

引用次数: 0

DDoS attack detection and defense techniques in software defined networks: A survey 软件定义网络中的DDoS攻击检测与防御技术综述

IF 12.9 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-31 DOI: 10.1016/j.cosrev.2026.100921

Wei Wang, Yong Liu, Qian Meng, Zihang Chen

引用次数: 0

AI-driven blockchain technology in smart healthcare system: Opportunities, challenges and future implications 智能医疗系统中人工智能驱动的区块链技术：机遇、挑战和未来影响

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-30 DOI: 10.1016/j.cosrev.2026.100909

Yunsheng Zhang , Syed Muhammad Mohsin , Hana Mujlid , Muhammad Sadiq , Syed Muhammad Abrar Akber , Sheraz Aslam , Junwei Liang

Blockchain technology in conjunction with artificial intelligence (AI) is transforming smart healthcare systems, by providing enhanced data security, interoperability, and transparency. Integration of AI along with blockchain into smart healthcare systems offers numerous benefits, including supporting decision-making processes, reducing administrative burdens, improving coordination of patient care and automated, trust-based execution of healthcare agreements. This study presents applications of AI-based blockchain technology in the field of smart healthcare and analyzes the state of affairs, highlights the key issues, and identifies perspectives to strengthen the reliability and trustworthiness of future medical systems. The study uses a structured framework to analyze the effectiveness of blockchain in healthcare by contrasting its advantages and disadvantages. Blockchain systems benefit healthcare by improving data security, streamlining data processing, ensuring trust, facilitating telemedicine and remote monitoring, and enabling efficient consent management, automated workflows and medication traceability. In this context, the study introduces a conceptual model namely the trust–automation–interoperability (TAI) synergy framework to guide the design, analysis, and deployment of AI-enabled blockchain solutions for smart healthcare aiming to achieve a sustainable digital health ecosystem by strengthening three fundamental dimensions: trust, automation, and interoperability. However, challenges such as scalability, interoperability, legal ambiguities, security concerns, user experience, acceptance barriers, long-term data storage, connectivity issues, discrepancies between data formats, user identity management, and cost considerations emphasize the importance of strong solutions.

区块链技术与人工智能（AI）相结合，通过提供增强的数据安全性、互操作性和透明度，正在改变智能医疗保健系统。将AI与区块链集成到智能医疗保健系统中可以带来许多好处，包括支持决策流程、减少行政负担、改善患者护理的协调以及基于信任的医疗保健协议的自动化执行。本文介绍了基于人工智能的区块链技术在智能医疗领域的应用，分析了现状，突出了关键问题，并确定了增强未来医疗系统可靠性和可信度的前景。本研究采用结构化框架，通过对比区块链的优缺点，分析区块链在医疗保健领域的有效性。区块链系统通过提高数据安全性、简化数据处理、确保信任、促进远程医疗和远程监控，以及实现高效的同意管理、自动化工作流程和药物可追溯性，使医疗保健受益匪浅。在此背景下，该研究引入了一个概念模型，即信任-自动化-互操作性（TAI）协同框架，以指导智能医疗的ai支持区块链解决方案的设计、分析和部署，旨在通过加强信任、自动化和互操作性三个基本维度来实现可持续的数字健康生态系统。然而，诸如可伸缩性、互操作性、法律模糊性、安全问题、用户体验、接受障碍、长期数据存储、连接问题、数据格式之间的差异、用户身份管理和成本考虑等挑战强调了强大解决方案的重要性。

{"title":"AI-driven blockchain technology in smart healthcare system: Opportunities, challenges and future implications","authors":"Yunsheng Zhang , Syed Muhammad Mohsin , Hana Mujlid , Muhammad Sadiq , Syed Muhammad Abrar Akber , Sheraz Aslam , Junwei Liang","doi":"10.1016/j.cosrev.2026.100909","DOIUrl":"10.1016/j.cosrev.2026.100909","url":null,"abstract":"<div><div>Blockchain technology in conjunction with artificial intelligence (AI) is transforming smart healthcare systems, by providing enhanced data security, interoperability, and transparency. Integration of AI along with blockchain into smart healthcare systems offers numerous benefits, including supporting decision-making processes, reducing administrative burdens, improving coordination of patient care and automated, trust-based execution of healthcare agreements. This study presents applications of AI-based blockchain technology in the field of smart healthcare and analyzes the state of affairs, highlights the key issues, and identifies perspectives to strengthen the reliability and trustworthiness of future medical systems. The study uses a structured framework to analyze the effectiveness of blockchain in healthcare by contrasting its advantages and disadvantages. Blockchain systems benefit healthcare by improving data security, streamlining data processing, ensuring trust, facilitating telemedicine and remote monitoring, and enabling efficient consent management, automated workflows and medication traceability. In this context, the study introduces a conceptual model namely the trust–automation–interoperability (TAI) synergy framework to guide the design, analysis, and deployment of AI-enabled blockchain solutions for smart healthcare aiming to achieve a sustainable digital health ecosystem by strengthening three fundamental dimensions: trust, automation, and interoperability. However, challenges such as scalability, interoperability, legal ambiguities, security concerns, user experience, acceptance barriers, long-term data storage, connectivity issues, discrepancies between data formats, user identity management, and cost considerations emphasize the importance of strong solutions.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100909"},"PeriodicalIF":12.7,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Cyber attack detection in smart grids: A survey of methods, challenges and future directions 智能电网中的网络攻击检测：方法、挑战和未来方向的调查

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-30 DOI: 10.1016/j.cosrev.2026.100915

Palanisamy Vigneshwaran, Selvarajah Thuseethan, Bharanidharan Shanmugam, Suresh Thennadil

Smart grids (SG) are the well-known cyber-physical systems responsible for processing, managing, controlling and optimising power energy for multi-scale organisations. Advancements in computing technologies have significantly revolutionised their performance with the integration of the Internet of Things (IoT), artificial intelligence (AI) and cloud technologies. However, such advancements also yield unforeseen vulnerabilities that pose serious threats to the grid management and control. In particular, the rapid advancement of AI and the increasing complexity of adversarial and other cyber attacks have created a highly vulnerable security landscape, despite years of countermeasure development. This study comprehensively reviews the recognition mechanisms utilised against such threats from the perspective of three layers of SG, such as sensing, communication and application layers. Additionally, it maps a wider range of cyber attacks with SG-related threat models and characteristics, along with vulnerable SG components. More importantly, it provides an in-depth comparative analysis of benchmark datasets and related evaluation metrics, including power system metrics, which is unique among other studies. This work also highlights the contemporary limitations and essential recommendations for future integration. Overall, the primary objective of the study is to guide scholars in understanding cutting-edge anomaly recognition techniques employed in SG and provide recommendations for mitigating challenges posed by advanced threats.

智能电网（SG）是众所周知的网络物理系统，负责处理、管理、控制和优化多规模组织的能源。随着物联网（IoT）、人工智能（AI）和云技术的融合，计算技术的进步极大地改变了它们的性能。然而，这些进步也产生了不可预见的漏洞，对电网管理和控制构成严重威胁。特别是，人工智能的快速发展以及对抗性和其他网络攻击的日益复杂，造成了一个高度脆弱的安全环境，尽管多年来一直在开发对策。本研究从传感层、通信层和应用层三个层面全面回顾了针对此类威胁的识别机制。此外，它还映射了与SG相关的威胁模型和特征以及易受攻击的SG组件的更广泛的网络攻击。更重要的是，它提供了对基准数据集和相关评估指标（包括电力系统指标）的深入比较分析，这在其他研究中是独一无二的。这项工作还强调了当代的局限性和对未来整合的重要建议。总体而言，本研究的主要目的是指导学者了解SG中使用的尖端异常识别技术，并为减轻高级威胁带来的挑战提供建议。

{"title":"Cyber attack detection in smart grids: A survey of methods, challenges and future directions","authors":"Palanisamy Vigneshwaran, Selvarajah Thuseethan, Bharanidharan Shanmugam, Suresh Thennadil","doi":"10.1016/j.cosrev.2026.100915","DOIUrl":"10.1016/j.cosrev.2026.100915","url":null,"abstract":"<div><div>Smart grids (SG) are the well-known cyber-physical systems responsible for processing, managing, controlling and optimising power energy for multi-scale organisations. Advancements in computing technologies have significantly revolutionised their performance with the integration of the Internet of Things (IoT), artificial intelligence (AI) and cloud technologies. However, such advancements also yield unforeseen vulnerabilities that pose serious threats to the grid management and control. In particular, the rapid advancement of AI and the increasing complexity of adversarial and other cyber attacks have created a highly vulnerable security landscape, despite years of countermeasure development. This study comprehensively reviews the recognition mechanisms utilised against such threats from the perspective of three layers of SG, such as sensing, communication and application layers. Additionally, it maps a wider range of cyber attacks with SG-related threat models and characteristics, along with vulnerable SG components. More importantly, it provides an in-depth comparative analysis of benchmark datasets and related evaluation metrics, including power system metrics, which is unique among other studies. This work also highlights the contemporary limitations and essential recommendations for future integration. Overall, the primary objective of the study is to guide scholars in understanding cutting-edge anomaly recognition techniques employed in SG and provide recommendations for mitigating challenges posed by advanced threats.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100915"},"PeriodicalIF":12.7,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146089496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A comprehensive review on quantum steganography techniques for data hiding 用于数据隐藏的量子隐写技术综述

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-29 DOI: 10.1016/j.cosrev.2026.100911

Ahana Patra , Ram Chandra Barik , Suvamoy Changder

Quantum steganography is an emerging field that blends the principles of quantum mechanics with the traditional art of steganography. It is an art of transmitting important information secretly without letting others know about the communication. Steganography is the practice of hiding information within other information. The exciting feature of quantum steganography is that it provides the capability to hide both classical and quantum qubits or quantum states. Thus, it can achieve better capacity, time complexity, and imperceptibility in comparison to the classical system. This survey paper describes various methodologies that have been applied in different steganography domains, such as text steganography, image steganography, audio steganography, and video steganography, and discusses the strengths and weaknesses of each methodology separately. The aim of this paper is to provide an overview of the theoretical foundation, recent advances, and open challenges in the fields of both classical and quantum steganography. It also helps to explore new research directions in quantum steganography across various domains, including quantum image steganography, quantum text steganography, quantum audio steganography, and quantum video steganography, depending on the requirements of imperceptibility, payload capacity, and robustness against attacks.

量子隐写术是将量子力学原理与传统的隐写术相结合的新兴领域。这是一种秘密传递重要信息而不让他人知道的艺术。隐写术是将信息隐藏在其他信息中。量子隐写术令人兴奋的特点是，它提供了隐藏经典量子比特和量子量子比特或量子态的能力。因此，与经典系统相比，它可以实现更好的容量、时间复杂度和不可感知性。这篇调查报告描述了在不同的隐写领域中应用的各种方法，如文本隐写、图像隐写、音频隐写和视频隐写，并分别讨论了每种方法的优缺点。本文旨在概述经典和量子隐写术领域的理论基础、最新进展和开放挑战。它还有助于探索量子隐写在各个领域的新研究方向，包括量子图像隐写、量子文本隐写、量子音频隐写和量子视频隐写，这取决于不可感知性、有效载荷能力和抗攻击的鲁棒性的要求。

{"title":"A comprehensive review on quantum steganography techniques for data hiding","authors":"Ahana Patra , Ram Chandra Barik , Suvamoy Changder","doi":"10.1016/j.cosrev.2026.100911","DOIUrl":"10.1016/j.cosrev.2026.100911","url":null,"abstract":"<div><div>Quantum steganography is an emerging field that blends the principles of quantum mechanics with the traditional art of steganography. It is an art of transmitting important information secretly without letting others know about the communication. Steganography is the practice of hiding information within other information. The exciting feature of quantum steganography is that it provides the capability to hide both classical and quantum qubits or quantum states. Thus, it can achieve better capacity, time complexity, and imperceptibility in comparison to the classical system. This survey paper describes various methodologies that have been applied in different steganography domains, such as text steganography, image steganography, audio steganography, and video steganography, and discusses the strengths and weaknesses of each methodology separately. The aim of this paper is to provide an overview of the theoretical foundation, recent advances, and open challenges in the fields of both classical and quantum steganography. It also helps to explore new research directions in quantum steganography across various domains, including quantum image steganography, quantum text steganography, quantum audio steganography, and quantum video steganography, depending on the requirements of imperceptibility, payload capacity, and robustness against attacks.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100911"},"PeriodicalIF":12.7,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A review of self-scheduled strategies for numerical linear algebra on GPU GPU上数值线性代数的自调度策略综述

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-29 DOI: 10.1016/j.cosrev.2026.100901

Manuel Freire, Ernesto Dufrechou, Pablo Ezzatti

The slowdown of Moore’s Law has generated a shift toward multicore and massively parallel architectures, making it essential to efficiently exploit parallelism for high performance. Among task-scheduling paradigms for shared-memory systems, self-scheduled (or sync-free) methods dynamically assign tasks to processors, balancing workload and improving utilization while managing data dependencies through synchronization mechanisms. Originally applied to sparse linear algebra on CPUs, these techniques have recently been adapted to GPUs. The introduction of the self-scheduled approach for sparse triangular solves (

S p T R S V

) marked a breakthrough, outperforming static level-set methods and motivating further work. While the first efforts related to this paradigm were focused on the

S p T R S V

operation, the efficiency of this strategy has extended to multiple Numerical Linear Algebra (NLA) kernels. This paper surveys the state of the art in self-scheduled algorithms for numerical linear algebra on GPUs. We classify contributions into three categories: (i) optimizations of triangular solvers, (ii) preprocessing techniques that enhance sync-free execution, and (iii) extensions of the paradigm to other linear algebra kernels.

摩尔定律的放缓导致了向多核和大规模并行架构的转变，这使得高效利用并行性来实现高性能变得至关重要。在共享内存系统的任务调度范式中，自调度（或无同步）方法动态地将任务分配给处理器，平衡工作负载并提高利用率，同时通过同步机制管理数据依赖性。这些技术最初应用于cpu上的稀疏线性代数，最近已被应用于gpu。稀疏三角解的自调度方法（SpTRSV）的引入标志着一个突破，它优于静态水平集方法，并激励了进一步的工作。虽然与此范式相关的第一个努力集中在SpTRSV操作上，但该策略的效率已扩展到多个数值线性代数（NLA）内核。本文综述了gpu上数值线性代数自调度算法的研究现状。我们将贡献分为三类：(i)三角求解器的优化，（ii）增强无同步执行的预处理技术，以及（iii）将范式扩展到其他线性代数内核。

{"title":"A review of self-scheduled strategies for numerical linear algebra on GPU","authors":"Manuel Freire, Ernesto Dufrechou, Pablo Ezzatti","doi":"10.1016/j.cosrev.2026.100901","DOIUrl":"10.1016/j.cosrev.2026.100901","url":null,"abstract":"<div><div>The slowdown of Moore’s Law has generated a shift toward multicore and massively parallel architectures, making it essential to efficiently exploit parallelism for high performance. Among task-scheduling paradigms for shared-memory systems, self-scheduled (or sync-free) methods dynamically assign tasks to processors, balancing workload and improving utilization while managing data dependencies through synchronization mechanisms. Originally applied to sparse linear algebra on CPUs, these techniques have recently been adapted to GPUs. The introduction of the self-scheduled approach for sparse triangular solves (<span><math><mi>S</mi><mi>p</mi><mi>T</mi><mi>R</mi><mi>S</mi><mi>V</mi></math></span>) marked a breakthrough, outperforming static level-set methods and motivating further work. While the first efforts related to this paradigm were focused on the <span><math><mi>S</mi><mi>p</mi><mi>T</mi><mi>R</mi><mi>S</mi><mi>V</mi></math></span> operation, the efficiency of this strategy has extended to multiple Numerical Linear Algebra (NLA) kernels. This paper surveys the state of the art in self-scheduled algorithms for numerical linear algebra on GPUs. We classify contributions into three categories: (i) optimizations of triangular solvers, (ii) preprocessing techniques that enhance sync-free execution, and (iii) extensions of the paradigm to other linear algebra kernels.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100901"},"PeriodicalIF":12.7,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Federated learning in cloud-edge-fog architectures: Enhancing privacy, efficiency, and scalability 云边缘雾架构中的联邦学习：增强隐私、效率和可扩展性

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-29 DOI: 10.1016/j.cosrev.2026.100917

Zahra Jalali KhalilAbadi, Najme Mansouri, Mohammad Masoud Javidi

In cloud, edge, and fog computing environments, Federated Learning (FL) enables model training across decentralized devices without sharing raw data. However, a comprehensive survey that integrates the unique challenges and solutions of FL across this hierarchical computing stack is lacking. This paper presents a systematic review of federated learning applications, algorithms, and frameworks within integrated cloud-edge-fog ecosystems. Our key contributions include: (a) a novel taxonomy classifying FL strategies across the computing continuum; (b) a comparative analysis of aggregation techniques and optimization methods tailored for system heterogeneity and communication bottlenecks; and (c) the identification of critical research gaps. We consolidate findings on key challenges such as model drift, adversarial threats, and scalability, concluding with promising future research directions to advance decentralized intelligence in modern distributed systems.

在云计算、边缘计算和雾计算环境中，联邦学习（FL）可以跨分散设备进行模型训练，而无需共享原始数据。然而，在这个分层计算堆栈中集成FL的独特挑战和解决方案的全面调查是缺乏的。本文系统地回顾了集成云-边缘-雾生态系统中的联邦学习应用、算法和框架。我们的主要贡献包括：(a)在计算连续体中对FL策略进行分类的新分类法；(b)针对系统异质性和通信瓶颈量身定制的聚合技术和优化方法的比较分析；(c)关键研究缺口的识别。我们整合了关键挑战的研究结果，如模型漂移、对抗性威胁和可扩展性，并总结了有希望的未来研究方向，以推进现代分布式系统中的分散智能。

引用次数: 0

A comprehensive review on expansion and statistical evaluation of Coati optimization algorithm Coati优化算法的扩展及统计评价综述

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-28 DOI: 10.1016/j.cosrev.2026.100922

Vimal Kumar Pathak

The Coati optimization algorithm (COA) is a recently introduced metaheuristic algorithm inspired from the hunting and escaping activities of Coati in their habitat. Since its development, COA has achieved new heights owing its efficacy in solving diverse optimization problems, flexibility and simple procedure. Alike other metaheuristic optimization algorithms, COA based on intelligent metaphor have seen rapid advancements and improvements, however with similar intrinsic weaknesses. The probable reason being inadequate investigations on COA procedure, limitations, statistical evaluation prior to its enhancement. The importance of the current review stems from its in-depth exploration of COA advancements, foundation principles and practical scenarios. This comprehensive review features 91 COA articles for analysing its adaptability and increasing interest in solving non-trivial optimization problems. It precisely explores the COA expansion, researching historical developments and modifications that enhances the algorithm capability in adapting to diverse landscapes of optimization problems. This review investigates the COA articles in terms of year wise publications, including enhancements, chaotic, opposition based learning variants, hybridizations, binary version, multi-objective variants, and different applications. It was shown that Elsevier and Springer have published highest number of COA papers having 54 and 51 counts, respectively. The published articles on classical COA confirm that it is mostly utilized for solving complex optimization problems in the field of electrical engineering (35%), I-o-T (22%) and medical diagnostics (17%) applications. Moreover, it is established that the COA performance has been improved by mostly hybridizing with other metaheuristic algorithms (29%), followed by introducing chaotic mapping (24%) and OBL scheme (15%), respectively. In addition, the present review significantly assesses the COA’s convergence behaviour and statistical performance in comparison to some well-known recent metaheuristic algorithms. Finally, concluding remarks with future research directions were recommended for scholars and scientists including improvements and critical suggestion for minimizing its limitations.

Coati优化算法（Coati optimization algorithm， COA）是最近提出的一种元启发式算法，其灵感来自于Coati在其栖息地的狩猎和逃跑活动。自发展以来，COA算法以其解决多种优化问题的有效性、灵活性和程序简单等优点，达到了新的高度。与其他元启发式优化算法一样，基于智能隐喻的COA也取得了快速的发展和改进，但也存在类似的内在弱点。可能的原因是对COA程序的调查不足，存在局限性，在其加强之前进行统计评估。本次审查的重要性在于其对COA进展、基本原则和实际情况的深入探讨。这篇综合综述收录了91篇COA文章，用于分析其适应性和对解决非平凡优化问题的兴趣。它精确地探索了COA的扩展，研究了历史发展和改进，提高了算法适应各种优化问题的能力。本综述根据年度出版物调查了COA文章，包括增强，混沌，基于对立的学习变体，杂交，二进制版本，多目标变体和不同的应用。结果表明，Elsevier和b施普林格发表的COA论文数量最多，分别为54篇和51篇。已发表的关于经典COA的文章证实，它主要用于解决电气工程（35%）、I-o-T（22%）和医疗诊断（17%）应用领域的复杂优化问题。此外，我们还发现，COA性能的提高主要是通过与其他元启发式算法的混合（29%），然后分别引入混沌映射（24%）和OBL方案（15%）。此外，本综述还将COA的收敛行为和统计性能与最近一些著名的元启发式算法进行了比较。最后，结束语对学者和科学家提出了未来的研究方向，包括改进和减少局限性的关键建议。

{"title":"A comprehensive review on expansion and statistical evaluation of Coati optimization algorithm","authors":"Vimal Kumar Pathak","doi":"10.1016/j.cosrev.2026.100922","DOIUrl":"10.1016/j.cosrev.2026.100922","url":null,"abstract":"<div><div>The Coati optimization algorithm (COA) is a recently introduced metaheuristic algorithm inspired from the hunting and escaping activities of Coati in their habitat. Since its development, COA has achieved new heights owing its efficacy in solving diverse optimization problems, flexibility and simple procedure. Alike other metaheuristic optimization algorithms, COA based on intelligent metaphor have seen rapid advancements and improvements, however with similar intrinsic weaknesses. The probable reason being inadequate investigations on COA procedure, limitations, statistical evaluation prior to its enhancement. The importance of the current review stems from its in-depth exploration of COA advancements, foundation principles and practical scenarios. This comprehensive review features 91 COA articles for analysing its adaptability and increasing interest in solving non-trivial optimization problems. It precisely explores the COA expansion, researching historical developments and modifications that enhances the algorithm capability in adapting to diverse landscapes of optimization problems. This review investigates the COA articles in terms of year wise publications, including enhancements, chaotic, opposition based learning variants, hybridizations, binary version, multi-objective variants, and different applications. It was shown that Elsevier and Springer have published highest number of COA papers having 54 and 51 counts, respectively. The published articles on classical COA confirm that it is mostly utilized for solving complex optimization problems in the field of electrical engineering (35%), I-o-T (22%) and medical diagnostics (17%) applications. Moreover, it is established that the COA performance has been improved by mostly hybridizing with other metaheuristic algorithms (29%), followed by introducing chaotic mapping (24%) and OBL scheme (15%), respectively. In addition, the present review significantly assesses the COA’s convergence behaviour and statistical performance in comparison to some well-known recent metaheuristic algorithms. Finally, concluding remarks with future research directions were recommended for scholars and scientists including improvements and critical suggestion for minimizing its limitations.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100922"},"PeriodicalIF":12.7,"publicationDate":"2026-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Advances in geospatial artificial intelligence for remote sensing applications 地理空间人工智能遥感应用研究进展

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-24 DOI: 10.1016/j.cosrev.2026.100913

Elias Dritsas, Maria Trigka

The convergence of artificial intelligence (AI) and remote sensing (RS) has accelerated the development of geospatial AI (GeoAI), a domain focused on extracting structured knowledge from spatially explicit high-dimensional observations. This survey synthesizes recent methodological advances in GeoAI with an emphasis on modern machine learning (ML), particularly deep learning (DL) architectures, spatiotemporal modeling, multimodal fusion, explainability frameworks, and large-scale pre-training paradigms. It examines the algorithmic foundations and operational scalability of GeoAI systems, and highlights how spatial priors, non-Euclidean structures, and dynamic environments challenge standard AI assumptions. The application domains are analyzed based on their geospatial properties and the modeling requirements imposed on them. The survey also discusses persistent challenges in spatial generalization, uncertainty quantification, ethical deployment, and benchmarking. By integrating advances in multimodal representation learning, scalable inference, and adaptive learning mechanisms, this study links theoretical innovation to practical deployment. Furthermore, it examines the role of the foundation and generative models in enabling cross-regional transfer and robustness under data scarcity. Finally, it outlines a roadmap for developing transparent, adaptive, and trustworthy GeoAI systems capable of supporting global-scale decision-making.

人工智能（AI）和遥感（RS）的融合加速了地理空间人工智能（GeoAI）的发展，这是一个专注于从空间显式高维观测中提取结构化知识的领域。本调查综合了GeoAI的最新方法进展，重点是现代机器学习（ML），特别是深度学习（DL）架构、时空建模、多模态融合、可解释性框架和大规模预训练范例。它考察了GeoAI系统的算法基础和操作可扩展性，并强调了空间先验、非欧几里得结构和动态环境如何挑战标准的人工智能假设。根据应用领域的地理空间属性和对其施加的建模需求对应用领域进行分析。该调查还讨论了在空间概化、不确定性量化、伦理部署和基准方面的持续挑战。通过整合多模态表示学习、可扩展推理和自适应学习机制的进展，本研究将理论创新与实际部署联系起来。此外，它还考察了基础模型和生成模型在数据稀缺条件下实现跨区域转移和鲁棒性方面的作用。最后，它概述了开发透明、自适应和可信赖的GeoAI系统的路线图，该系统能够支持全球规模的决策。

{"title":"Advances in geospatial artificial intelligence for remote sensing applications","authors":"Elias Dritsas, Maria Trigka","doi":"10.1016/j.cosrev.2026.100913","DOIUrl":"10.1016/j.cosrev.2026.100913","url":null,"abstract":"<div><div>The convergence of artificial intelligence (AI) and remote sensing (RS) has accelerated the development of geospatial AI (GeoAI), a domain focused on extracting structured knowledge from spatially explicit high-dimensional observations. This survey synthesizes recent methodological advances in GeoAI with an emphasis on modern machine learning (ML), particularly deep learning (DL) architectures, spatiotemporal modeling, multimodal fusion, explainability frameworks, and large-scale pre-training paradigms. It examines the algorithmic foundations and operational scalability of GeoAI systems, and highlights how spatial priors, non-Euclidean structures, and dynamic environments challenge standard AI assumptions. The application domains are analyzed based on their geospatial properties and the modeling requirements imposed on them. The survey also discusses persistent challenges in spatial generalization, uncertainty quantification, ethical deployment, and benchmarking. By integrating advances in multimodal representation learning, scalable inference, and adaptive learning mechanisms, this study links theoretical innovation to practical deployment. Furthermore, it examines the role of the foundation and generative models in enabling cross-regional transfer and robustness under data scarcity. Finally, it outlines a roadmap for developing transparent, adaptive, and trustworthy GeoAI systems capable of supporting global-scale decision-making.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100913"},"PeriodicalIF":12.7,"publicationDate":"2026-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A multi-dimensional study of IoT DDoS in smart environments: SDN integration, taxonomy, security gaps, and emerging defenses 智能环境下物联网DDoS的多维研究：SDN集成、分类、安全漏洞和新兴防御

IF 12.7 1区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Computer Science Review

Pub Date : 2026-01-23 DOI: 10.1016/j.cosrev.2026.100903

Shivani Rathore, Abhinav Bhandari, Raman Maini

The proliferation of Internet of Things (IoT) devices in Software-Defined Networking (SDN)-enabled smart environments has significantly expanded the threat landscape, especially for Distributed Denial of Service (DDoS) attacks. This issue has gained scientific attention due to its complex socio-technical nature. This article examines the architecture of emerging IoT technologies, provides visual illustrations of vulnerabilities in smart environments, and discusses the increase in DDoS attacks related to IoT botnets. Also, we provide a market study of the current IoT and SDN deployments and DDoS landscape that reveals critical gaps between theoretical frameworks and practical adoption. This industry-driven perspective strengthens the motivation for conducting a structured review that aligns emerging academic research with evolving market statistics and deployment challenges. Additionally, we introduce a new IoT DDoS attack taxonomy designed to address protocol-level vulnerabilities in commonly used IoT communication protocols, including MQTT, CoAP, and SSDP. Our taxonomy is comprehensive and innovative, representing the first framework dedicated solely to analyzing IoT-driven DDoS attacks. It offers researchers a better understanding of this problem. By developing this framework, our goal is to help practitioners and researchers identify vulnerabilities, develop mitigation strategies, and strengthen the resilience of IoT environments against DDoS attacks.

Furthermore, we critically examine state-of-the-art DDoS mitigation techniques within SDN-enabled smart infrastructure. We conduct an incisive evaluation of contemporary DDoS mitigation strategies presented by many researchers, scrutinizing their operational potency, deployment feasibility, and resilience under dynamic, heterogeneous smart surroundings. Our review reveals that while a significant body of research claims to address IoT-based DDoS attacks, the majority of defenses remain centered on conventional network protocols, overlooking the distinct characteristics and vulnerabilities of IoT-specific protocols such as MQTT, CoAP, and SSDP. Building on this study, we identify critical research gaps and highlight novel possibilities for enhancing the resilience of next-generation smart networks against evolving DDoS adversaries.

在支持软件定义网络（SDN）的智能环境中，物联网（IoT）设备的激增极大地扩展了威胁格局，尤其是分布式拒绝服务（DDoS）攻击。这一问题由于其复杂的社会技术性质而引起了科学界的关注。本文研究了新兴物联网技术的架构，提供了智能环境中漏洞的可视化说明，并讨论了与物联网僵尸网络相关的DDoS攻击的增加。此外，我们还提供了当前物联网和SDN部署以及DDoS格局的市场研究，揭示了理论框架与实际采用之间的关键差距。这种行业驱动的观点加强了进行结构化审查的动机，将新兴的学术研究与不断变化的市场统计和部署挑战结合起来。此外，我们引入了一种新的物联网DDoS攻击分类法，旨在解决常用物联网通信协议（包括MQTT， CoAP和SSDP）中的协议级漏洞。我们的分类是全面和创新的，代表了第一个专门用于分析物联网驱动的DDoS攻击的框架。它让研究人员更好地理解了这个问题。通过开发此框架，我们的目标是帮助从业者和研究人员识别漏洞，制定缓解策略，并加强物联网环境对DDoS攻击的弹性。此外，我们在支持sdn的智能基础设施中严格检查了最先进的DDoS缓解技术。我们对许多研究人员提出的当代DDoS缓解策略进行了深刻的评估，仔细检查了它们在动态、异构智能环境下的操作效力、部署可行性和弹性。我们的审查显示，虽然大量研究声称可以解决基于物联网的DDoS攻击，但大多数防御仍然集中在传统的网络协议上，忽视了MQTT、CoAP和SSDP等物联网特定协议的独特特征和漏洞。在这项研究的基础上，我们确定了关键的研究差距，并强调了增强下一代智能网络抵御不断变化的DDoS攻击的弹性的新可能性。

{"title":"A multi-dimensional study of IoT DDoS in smart environments: SDN integration, taxonomy, security gaps, and emerging defenses","authors":"Shivani Rathore, Abhinav Bhandari, Raman Maini","doi":"10.1016/j.cosrev.2026.100903","DOIUrl":"10.1016/j.cosrev.2026.100903","url":null,"abstract":"<div><div>The proliferation of Internet of Things (IoT) devices in Software-Defined Networking (SDN)-enabled smart environments has significantly expanded the threat landscape, especially for Distributed Denial of Service (DDoS) attacks. This issue has gained scientific attention due to its complex socio-technical nature. This article examines the architecture of emerging IoT technologies, provides visual illustrations of vulnerabilities in smart environments, and discusses the increase in DDoS attacks related to IoT botnets. Also, we provide a market study of the current IoT and SDN deployments and DDoS landscape that reveals critical gaps between theoretical frameworks and practical adoption. This industry-driven perspective strengthens the motivation for conducting a structured review that aligns emerging academic research with evolving market statistics and deployment challenges. Additionally, we introduce a new IoT DDoS attack taxonomy designed to address protocol-level vulnerabilities in commonly used IoT communication protocols, including MQTT, CoAP, and SSDP. Our taxonomy is comprehensive and innovative, representing the first framework dedicated solely to analyzing IoT-driven DDoS attacks. It offers researchers a better understanding of this problem. By developing this framework, our goal is to help practitioners and researchers identify vulnerabilities, develop mitigation strategies, and strengthen the resilience of IoT environments against DDoS attacks.</div><div>Furthermore, we critically examine state-of-the-art DDoS mitigation techniques within SDN-enabled smart infrastructure. We conduct an incisive evaluation of contemporary DDoS mitigation strategies presented by many researchers, scrutinizing their operational potency, deployment feasibility, and resilience under dynamic, heterogeneous smart surroundings. Our review reveals that while a significant body of research claims to address IoT-based DDoS attacks, the majority of defenses remain centered on conventional network protocols, overlooking the distinct characteristics and vulnerabilities of IoT-specific protocols such as MQTT, CoAP, and SSDP. Building on this study, we identify critical research gaps and highlight novel possibilities for enhancing the resilience of next-generation smart networks against evolving DDoS adversaries.</div></div>","PeriodicalId":48633,"journal":{"name":"Computer Science Review","volume":"60 ","pages":"Article 100903"},"PeriodicalIF":12.7,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033304","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0