In the industrial landscape, cyber threats challenge information security, especially in industrial cybersecurity. Attack vectors like malware, phishing, and ransomware complicate data protection in manufacturing. Data breaches risk critical information and financial implications. QR codes, convenient for tasks like inventory management, introduce vulnerabilities. Traditional detection struggles with data volume and evolving hacking techniques. Recent advancements explore lightweight deep learning for industrial cybersecurity, focusing on QR codes. This research introduces a hybrid approach, using multi-objective optimization to enhance QR code-based cyber-attack detection. QR code images are trained with advanced CNN models like MobileNetV2 and ShuffleNet. The monarch butterfly optimization algorithm strategically selects impactful features. In practical use, the hybrid model achieved 99.82% accuracy, surpassing traditional CNN models. It proves effective for industrial cybersecurity, addressing vulnerabilities in QR code usage.
TopIntroduction
In the contemporary landscape of cybersecurity, an array of studies has been conducted to counteract the ever-evolving and sophisticated nature of cyber-attacks and hacking techniques (Cil, A.E., Yildiz, K. & Buldu, A., 2021). Denial-of-Service (the DoS) as well as Distributed Denial-of-Service (the DDoS) are two assaults of notable concern due to their capacity to disrupt continuously operational systems, specifically targeting servers, system networks, and websites as potential victims (Khader, R. & Eleyan, D., 2021).
Given the diversification and advancement of technologies, various types of DoS/DDoS attacks have emerged. To effectively address the findings of cyber-attacks, a hybrid multi-objective bio-inspired optimization-based feature selection (FS) method is employed, aiming to enhance performance and accuracy by strategically selecting pertinent features for cyber-attack detection.
Research works on lightweight pre-emptive padding-based cryptographic security arrangements tailored for distributed cloud environments have been proposed to address vulnerabilities to cyber-attacks (Indira, N., & Rukmanidevi, S, 2019). These arrangements employ efficient padding techniques, offering robust protection against potential threats in a distributed cloud setting.
Machine learning methods and techniques have also been incorporated to facilitate real-time recognition of cyber-attacks across multiple systems (Awan et.al, 2021). As cyber-attacks grow more intricate and intelligent, particularly with the limited resources of physical and ever-growing IoT Edge-based devices in modern Edge/Cloud technologies, cloud-based architectures are utilized for substantial computational procedures like deep traffic control and classification. Extreme Learning Machines (ELM) play a crucial role in the proposed traffic classification model, overcoming challenges related to training data volume, storage, and processing capabilities (Kozik, R., et. al, 2018).
In response to the challenges posed by these sophisticated cyber threats, various intrusion detection systems have been developed. One innovative approach involves enhancing user security in computer networks by creating unique QR codes for each user. Users are granted access to the system by logging in with their personalized QR codes, providing an additional layer of protection against cyber-attacks.
Secure and reliable routing, considering node mobility patterns and behaviours, has been proposed to contribute to a more secure and trustworthy communication framework in dynamic Wireless Sensor Networks (WSNs) (Kumar, B.P., et. al, 2023). The authors suggest a unique way of communication to protect against attacks.
In the era of rapid information technology development, servers have become prime targets for cyber-attacks, given the substantial value associated with components such as mobile and desktop applications, computer networks, and websites (Al-talak et. al, 2021). Servers are susceptible to a variety of attacks, including server attacks, network attacks, TP-Brute Force attacks targeting web security, and more sophisticated threats such as SQL injection as well as cross-site scripting (the XSS) attacks (Liu et. al, 2021). The diverse characteristics of these components make servers vulnerable to a range of cyber threats, from basic attacks to more advanced and targeted intrusions.
Secure data sharing mechanisms employing Attribute-based Encryption (the ABE) for Remote Data Checking (the RDC) within a cloud-based environment have been suggested as an approach to ensure protected data sharing while facilitating remote verification in cloud-based data interactions (Ramprasath et. al, 2023). This signifies the importance of implementing advanced encryption techniques to safeguard data in cloud environments, where remote data checking is essential for maintaining the integrity of shared information.