Most recent issue published online in the International Journal of Critical Computer-Based Systems.

1. An enhanced test case prioritisation with cyber-security detection using CMSANN for banking or finance software
   In the software development process, software testing executes a program to detect errors. During this process, security attacks may occur, posing significant threats. Incorporating AI into cybersecurity testing processes can help organisations protect sensitive financial data from unauthorised access. So, this research methodology proposed a CMSANN-based cybersecurity prediction for software testing. Initially, the test cases are generated and prioritised using the WW-ArCa-PA approach in the banking or finance application. Then, it enters the data structuring phase. Data are mapped using AKNMHC and reduced using the HS approach in this phase. Then, the features are extracted and given to the cyber-security detection phase. The anomaly dataset is given as input, and pre-processing is performed using MVI, numeralisation, and normalisation. After pre-processing, the features are extracted, and then data are balanced using the ACA-CTSYN approach. The balanced data is given to the CMSANN classifier for cyber security detection.
2. An analysis of secure software delivery validation using blockchain
   Due to a lack of security in the supply chain of the software delivery process, the proposed method aims to address the challenges encountered during software delivery and mitigate the attack surface of software supply chain attacks. Blockchain technology establishes trust by verifying software packages and their developers, thereby achieving this goal, reducing the risk of executing malicious or compromised software. The proof of security is rooted in the immutable nature of blockchain. This feature provides a tamperproof record of software developers and the packages they publish, instilling a high level of trust in the verification process. The system holds the promise of a future where organisations can confidently verify the identity of software developers, allowing them to trust only the software from vendors they trust. When a tampered package enters the delivery process, the installation checks will promptly detect the compromise, leading to the failure of the attack chain.
3. Enhancing IoMT security: spotted hyena optimised bi-layered attention adaptive recurrent network for intrusion detection
   Intrusion detection is essential to computer and network security because it detects unauthorised or malicious activity. It monitors and analyses network or system activity in real-time to alert or act on suspicious. Internet of medical things (IoMT) uses intrusion detection and attack detection systems to protect medical devices, healthcare systems, and patient data from cyberattacks. Researcher introduces a new IoMT-supported spotted hyena optimised bi-layered attention adaptive recurrent network (SHO-BAARNN) for intrusion detection. The spotted hyena optimisation (SHO) is used to adjust model parameters for better robustness and efficiency when processing IoMT data. The adaptive recurrent network enables real-time intrusion detection, while the BAARNN structure offers dual attention techniques to selectively focus on important data aspects. The incorporation of optimisation, attention mechanisms, and recurrent networks, among other practical implications, makes it an effective tool for real-time detection of intrusions, providing improved security for sensitive healthcare data and medical equipment.
4. Identifying security vulnerabilities in source code with safety verification
   Ensuring the security of modern software systems is critical due to increasing complexity and interconnectedness. While automated testing and bug-finding tools have made progress, detecting security flaws - especially those related to runtime behaviour - remains a significant challenge. Existing research in software security has concentrated on source code analysis and often ignores how the execution environment affects program semantics, leaving machine-dependent vulnerabilities undetected. To bridge this gap, our approach analyses source code with awareness of its execution environment, acknowledging that many flaws arise from subtle mismatches between the two. Our work targets arithmetic errors in C/C+ + programs by modelling how the execution context shapes program behaviour. This article presents the construction of our knowledge base and explains how it integrates with our algorithm to enable accurate, environment-aware vulnerability detection using precise logical formulas, enabling the reformulation of vulnerability detection as a satisfiability problem suitable for automated reasoning using formal methods.
5. Phishing attack detection and zero trust-based verification for secure data transfer using BTEL-GRU
   The authentication and access control of the user's data using zero trust architecture (ZTA) is necessary to prevent data breaches. Initially, the user registers in the cloud, and here, the hashcode is generated using the substitution cipher-based whirlpool hashing algorithm (SC-WHA). In the phishing attack detection model, word embedding is done using Kaiming normalised Xavier-based bidirectional encoder representations from Transformers (KNX-BERT) for content extracted from the email dataset, the features are extracted, and the dimensionality is reduced using linear discriminant analysis (LDA). The word embedded and reduced features are classified using Bernoulli-dropout TanhExp logish gated recurrent unit (BTEL-GRU). During testing, the attacked data is blocked, and the non-attacked data is secured using deltoid spiral curve cryptography (DS-CC) and uploaded to the cloud. Hence, in the proposed model, the phishing attack is identified with an accuracy of 98.77586945%, and the multifactor hashcode verification is done with a hashcode generation time of 1,326 ms for secured data transfer.