Validation adequacy in safety-critical software depends on more than the system under test. Critical scenarios must be constructed under controlled conditions, execution evidence must be aligned into verdict-ready form, and abnormal outcomes must be attributable to actionable causes. Existing testability research remains largely artifact-centric and offers little architectural support for reasoning about the combined capability of the scenario, the test system, and the system under test. Joint Testability Architecture (JTA) addresses this gap by treating those three elements as a single object of analysis and design. It characterizes validation capability along three dimensions–controllability, observability, and isolability–and organizes them through three domains, three bridges, and an analysis-design-evaluation-refinement loop. JTA also introduces scenario contracts, joint capability assessment, validation blind-spot identification, and bridge-oriented design actions that map capability gaps to concrete improvements in control points, evidence organization, and attribution boundaries. An illustrative analysis of ArduPilot failsafe validation shows that link-loss scenarios are comparatively mature, whereas state-estimation anomaly scenarios remain harder to validate because evidence alignment and attribution semantics are weaker. JTA is not a replacement for existing testing or safety-analysis techniques; it provides an architectural basis for modeling, designing, and assessing scenario-based validation in safety-critical software.

AI-driven intelligent testing has advanced rapidly, enabling automated test-case generation, defect prediction, and risk assessment. However, the absence of explicit integration of human factors into the testing process often leads to the neglect of testers' cognitive attributes and domain expertise, thereby amplifying cognitive biases and exacerbating safety risks. This paper proposes a three-layer Human-in-the-Loop Intelligent Testing (HITL-IT) framework that systematically incorporates human factors into AI-based testing for safety-critical software. The framework consists of a human-factor modeling layer, an AI testing core, and an interactive feedback loop, collectively forming a closed-cycle mechanism of “suggestion-challenge-refinement-relearning.” In the context of AI test-case generation, this framework is designed to substantially improve both the quality and efficiency of generated cases. Preliminary applications show promising potential for this approach. By embedding explicit human-factor models and closed-loop feedback into the testing workflow, HITL-IT provides a novel and practical paradigm for building more trustworthy, resilient, and safety-critical AI testing systems.

Magnetic random-access memory (MRAM) is an emerging non-volatile memory technology distinguished by its radiation hardness, high endurance, and CMOS process compatibility, positioning it as a promising candidate to replace static random-access memory (SRAM). However, mainstream MRAM devices exhibit low cache efficiency in digital computing-in-memory (CIM) architectures due to the inherent incompatibility between their writing mechanism and the ReLU output data structure. This pa-per proposes a NAND-Like SAS-MRAM (NLSAS-MRAM), which shows extremely high efficiency when writing zero-dominated data, and is therefore highly suitable for caching ReLU outputs. To rigorously verify this advantage, we developed a high-fidelity co-simulation flow. Simulation results show that the proposed NLSAS-MRAM achieves a writing power consumption of only 35% that of STT-MRAM and 63% of SOT-MRAM. Furthermore, it delivers a 52.6× improvement in cache potential FOM over STT-MRAM and a 2.6× improvement over SOT-MRAM.

Software requirements specifications play a critical role in ensuring the reliability of software systems. This paper examines how two novel requirements scenarios-Selectivity and Post-completion Error-can trigger developers to introduce defects into programs. These scenarios are grounded in psychological theories of human error and are not addressed by conventional requirements quality standards. A case study was conducted to investigate whether and how these human errorbased criteria manifest in real-world aviation industry projects. The study analyzed ten aviation-related software systems, with defect data collected by an independent third-party testing center during certification tests. The results indicate that violations of the proposed criteria in requirements specifications do, in fact, lead to corresponding errors during the design and coding phases. Based on these findings, we recommend incorporating these two criteria to proactively prevent and detect software defects.

Spin-orbit torque (SOT)-based in-memory logic designs offer a promising approach to alleviating the memory wall problem. However, in existing schemes, the basic cells mainly rely on auxiliary physical mechanisms or transistors, which considerably limit their applicability. In this work, we demonstrate a novel SOT-based device fabricated on an 8-inch wafer platform, featuring dual canted magnetic tunnel junctions (MTJs) monolithically integrated via a well-designed SOT channel. This device breakthrough simultaneously achieves fast switching (current density is 19.8 MA/cm(2)@5 ns), multi-bit storage, and in-memory logic operations via SOT alone, while maintaining compatibility with CMOS technology ( sigma % under 7.25% in arrays). By exploiting synergistic voltage control, we achieve in situ XOR operation and edge detection with error tolerance and competitive F-measure. Our proposed scheme provides a potential pathway toward next-generation large-scale, ultra-fast, and energy-efficient in-memory logic.

As process nodes shrink to deep nanometer scales, the heightened sensitivity of digital circuits to radiation and the dramatic rise in leakage power have become pressing concerns. Magnetic tunnel junction (MTJ) possesses intrinsic radiation resistance and nonvolatility and can be integrated with CMOS processes. Therefore, designing MTJ-based radiation-hardened nonvolatile storage elements is a promising solution to address these issues. In this brief, we propose a novel MTJ-based radiation-hardened, speed and power optimized nonvolatile (RH-SPO) latch and compares it with existing designs to evaluate its performance. Using a 28 nm CMOS process, simulation results confirm that the proposed RH-SPO latch offers moderate radiation resistance and high robustness performance with backup and restore operations. Additionally, compared to state-of-the-art design, named M-8C, it can save up to 50% in area overhead, reduce transmission-restore power consumption and restore time by 98% and 75%, respectively.

In this paper, the total ionizing dose (TID) effects and the heavy-ion effects of NAND-like spin-orbit torque (SOT) devices are evaluated. Experimental results show that these devices are highly radiation tolerant, which have negligible resistance changes and small critical switching voltage changes of about 3% to 4% under 1 Mrad (Si) Cobalt-60 irradiation. The heavy-ion irradiation leads to similar results in the critical switching voltage and can eliminate the multi-state of the devices. These findings suggest that NAND-like SOT devices have potential for use in radiation-hardened applications.

Spin-orbit torque magnetic random access memory (SOT-MRAM) is widely recognized as a promising candidate for the next generation of high-performance nonvolatile memory solutions. Although characterized by its ultrafast operation speed and exceptional reliability, SOTMRAM presents significant challenges in fabrication process and circuit design due to its three-terminal topology and toppinned nano-pillar stack structure. In this invited paper, we provide a review of our recent advancements in the development of SOT-MRAM. Key issues concerning the process and design of SOT-MRAM are thoroughly discussed, meanwhile detailed optimization strategies and improvements are presented.

The use of model-based techniques for the development and testing of safety-critical software is a current research focus. However, the complete and accurate depiction of software safety attributes during model construction remains a problem that needs to be addressed. This study focuses on the integration of safety features in model-based testing technologies for safety-critical software. By using the MARTE modeling language and traditional as well as innovative safety analysis methods, this study demonstrates the evolution from basic to advanced models through an empirical project. The empirical results indicate that the analysis outcomes obtained through traditional safety methods (FMEA and FHA) can be modeled in the MARTE language through the extension of constructs or addition of stereotypes. The systematic safety analysis method (STPA) can utilize MARTE to extend the viewpoint layer, resulting in a new hybrid model. This facilitates a better realization of the depiction of software safety attributes based on models and provides a more robust technical foundation for automated modeling.

Magnetic random access memory (MRAM) is considered a potential candidate for next-generation memory due to its high density, high access speed, and high endurance. Compared with the previous two generations, the third-generation spin-orbit moment (SOT)-MRAM is more reliable and complex, and adequate irradiation effect studies are required before its application in aerospace. In this paper, the effect of heavy ion irradiation on the performance of SOT magnetic dot devices and SOT spin Hall devices is investigated. The results show that when the irradiated ion flux reaches 107 ions/cm2, the magnetic dot device has good robustness to the Ta+ ion beam of 1912 MeV, and the critical flip current of the spin Hall device undergoes a shift of about 16.7% after irradiation.

The majority of adversarial attack methods aim to enhance the quality of adversarial samples by decreasing the perceived distance between the adversarial samples and the original samples. However, the majority of these methods still accomplish this by restricting the size of the $L_{\mathrm{p}}-\text{norm}$ . On the other hand, we discover that the $L_{\mathrm{p}}-\text{norm}$ is inconsistent with the perceived similarity based on the literature and experiments. Crucially, we also prove that “Potential Adversarial Samples” exist. This result suggests that there is still room for improvement in the attack success rate and query efficiency of the current $L_{\mathrm{p}}-\text{based}$ attack methods, and the quantity of “Potential Adversarial Samples” can be utilized as an indicator to assess the attack methods' optimizable space. The discovery also offers a precise course and objective for the optimization of attack methods in the following years. Furthermore, we have derived seven image quality assessment metrics from literature research. We then weigh the benefits and drawbacks of each index against the $L_{2}$ norm in terms of human perception, and as one of the optimization goals of our next attack strategy, we choose the indexes that most closely match human perception.

The analysis of real software defects data not only helps to identify problems in a project, company, or industry but also enables continuous improvement in technologies and processes. In the Chinese aviation industry, a rigorous software quality assurance procedure is implemented, consisting of following a national process standard, conducting internal tests, expert milestone reviews, and formal documentation of all activities and results. After these quality assurance activities are completed, software systems undergo certification tests conducted by independent third-party centers. Discovering patterns of defects found in certification tests has significant practical implications for the industry on continuous improvement. This paper presents a taxonomy of defect forms based on the real defects found in 10 software systems, using Grounded Theory. The defect forms contain detailed information on how the defects are formed, compared to existing defect types. We then applied and validated this taxonomy on 9 additional software projects, using five software certification engineers independent of this paper. The results demonstrate that the developed taxonomy can describe the forms of 98% of defects found in certification tests. We recommend this taxonomy for process improvement and defect prevention in the aviation industry.

Regression testing is part of every testing phase in software testing and plays a key role in software quality assurance. With the increasing complexity of software systems, it is often necessary to design test cases for regression testing to balance test adequacy and test overhead. Some studies have proposed algorithms for sequence completion, which generates only one test sequence to be completed at a time, generating test sequences with excessive overheads, and the method relies on a large amount of data flow information, and the degree of automation is also not high. To this end, this paper proposes a coverage criterion based on ‘test obligation’, which integrates the control flow and data flow information of the EFSM model as the coverage basis, and is used to guide the test sequence generation work to better reflect the dependency relationships on the EFSM, and accurately determine the range of the regression test. Based on the proposed test obligation coverage criterion, a genetic algorithm for regression testing test sequence generation is designed and implemented by combining test obligation coverage diversity and test sequence length as a fitness function. Experiments show that the test sequence set generated by this algorithm is better than the existing methods in terms of test sequence length and number of test sequences, which effectively reduces the regression test overhead and improves the test sequence generation efficiency.

Software testing is an indispensable step to ensure software quality, and the number of test cases will increase with the expansion of software scale. Sorting test cases is an important means to improve test efficiency. At present, there are many software requiring high safety, but the existing test case ranking index can not take software safety into account well, and then can not find the defects of software modules with high safety faster under the condition of limited test resources. To solve this problem, this paper proposes a new test case priority calculation method, which takes safety as the index to sort the test cases, and improves the test intensity and test priority of the safety-critical functional modules. In addition, to solve the problem of low sorting efficiency of greedy algorithm, this paper also proposes a test case priority sorting algorithm based on greedy algorithm and mountain climbing algorithm. The greedy algorithm and mountain climbing algorithm are combined by adding mountain climbing algorithm to each step of greedy algorithm to search adjacent nodes. Meanwhile, in view of the fact that search algorithms such as the mountain climbing algorithm require a certain data structure relationship between test case sets, this paper also proposes a graph building rule based on the coverage relationship, which can form a graph relationship between test cases, so that sorting algorithms of search type are able to search among test cases. The experimental results show that compared with the traditional greedy test case prioritization algorithm, the new algorithm has a shorter execution time on the premise of ensuring the quality of the prioritization.

Specification based testing is a common method in software testing. Specification based testing is divided into behavior-based testing, such as finite state machine method; and input-based testing, such as random testing, combination testing, decision table testing, equivalence class testing, boundary value testing method and so on. At present, there are many modeling methods for behavior-based testing and test cases can be generated automatically through these models. However, for these input-based testing methods, test cases are generated manually by searching for parameters and parameter values, which is too inefficient. To solve this problem, we want to improve the efficiency of input-based testing methods by automatically generating test cases. Therefore, this paper presents a software input space model that can represent parameter values and constraint relationships as the basis of automatic generation of test cases. Some safety-critical software tests need to consider the input timing of parameters, so we also express the input timing of parameters in the input space model. Then we propose the process of modeling the input space, which is finally expressed in the form of a directed graph. The values of parameters and the input timing of parameters are expressed on the edges of the directed graph, and the constraints between parameters are described by the way of node splitting, and all valid test sequences are represented by the path traversing the graph. Finally, the proposed input space modeling method is applied to a real test project to demonstrate the feasibility of the method.

Extended Finite State Machine(EFSM) has a wide range of application systems and protocols in control modeling. Genetic algorithms are commonly used in EFSM test data generation due to their good performance. Studies have been done to generate regression test sequences to meet the requirements based on changes in EFSM, there is a large amount of historical test data to choose from when using genetic algorithms to generate test data for this part of the test sequence, and there is no research on how to fully utilize the existing test data to generate regression test data for the EFSM. In this paper, a multi-population genetic algorithm is designed to generate test data to meet the requirements. Firstly, some test data that can cover the regression test sequences are selected from the historical test data, and for the rest of the test sequences, a mathematical model for the simultaneous optimization of multiple test sequences is constructed, secondly, in order to make full use of the historical test data, the relationship between the existing test data and the target test sequences is investigated, and the initial population is constructed based on the similarity of the test sequences and the FSCS-ART algorithm, which makes full use of the The initial population is constructed based on the similarity of test sequences and the FSCS-ART algorithm, which makes full use of the existing test data and ensures the diversity of the initial population. Finally, the adaptive crossover operator and variation operator were designed and the fitness function was optimized. The experimental results show the feasibility and efficiency of this method in test data generation.

As an emerging non-volatile memory technology, the spin orbit torque magnetic random access memory (SOT-MRAM) has attracted intensive research interest due to its advanced performance. However, the binary storage feature of the SOT-MRAM has become one of the obstacles. In this paper, we present a study of two-bit multi-level SOT-MRAM where two canted in-plane-anisotropy magnetic tunnel junctions (MTJs) store a pair of data. Compared with the previous schemes of multi-level SOT-MRAMs, our proposal enables fully one-step writing without the need of the preset operation. Micromagnetic simulation is performed to validate the functionality of the proposed multi-level cell (MLC) SOT-MRAM, meanwhile, the details of magnetization switching are clearly shown. Simulation results also demonstrate that the device could accomplish the magnetization switching at the sub-nanosecond speed and continuously decreasing power consumption with the size scaling down. In addition, the dipolar field between two cells has little influence on the switching process.

Time-triggered Ethernet (TTE) places stringent requirements on communication real-time, message security and uses efficient static scheduling algorithms for TT messages to guarantee the service quality of the network. In order to improve the solving performance of TT message schedule, a hybrid schedule technology based on genetic algorithm and simulated annealing is innovatively used in the TTE scheduling solution process. It includes the optimization of coding, selection and crossover. The TT message transmission constraint in the form of penalty functions is combined with TTE objective function to devise an adaptability function with the configurable parameters. In addition, the individual update process is designed by using the idea of annealing. Messages scenarios are set up to comprehensively evaluate the algorithm performance in operational efficiency and solution quality. Experimental results indicate that the hybrid technology combined genetic algorithm and simulated annealing is well qualified for TTE scheduling, which performs well in time consumption and global solution search.

The application of Service-Oriented Architecture (SOA) system software in safety-critical areas such as aerospace is gradually increasing, and the demand of high-quality SOA system software is increasing too. Due to the feature of loose coupling, SOA system software depends strongly on the correctness of the interfaces among services. Web service composition (WSC) testing in the integration phase of service composition thus becomes exceptionally critical. In a model-based testing framework., the test model lays a foundation for the subsequent test activities to proceed smoothly. In this paper, we propose an approach to automate modeling for WSC testing. The idea is to automate the construction of a highly visualized model by fetching adequate test information from two documents in the design phase. Compared to the existing modeling methods, the proposed method is more sufficient in WSC test information extraction. In addition, we also promote the degree of visualization and automation for test modeling. In the end, a case study is conducted to verify the effectiveness of the proposed modeling approach.

Software testing is an indispensable part of the software life cycle, and the quality of software testing largely affects the quality of software delivered to users. However, in the current stage of software testing quality research, the focus on testing quality influencing factors is still limited to theoretical analysis and model application, and there is a lack of empirical research based on real data and samples. To address this problem, this paper proposes the idea of using natural experiments to conduct empirical research in the field of software testing quality for the first time, and analyzes the feasibility of the research proposal through literature research. In this paper, the empirical study was conducted using real data provided by enterprises, through the research of cooperative enterprise project data and event records, selected the "CNAS expansion of the on-site review and audit training" as an exogenous event, team capacity as the explanatory variable to build a natural experimental model of test quality. After completing the empirical model, we analyze the results, which show that the exogenous events have a significant disposition effect on test quality, and the empirical results pass the four commonly used robustness tests, indicating that the experimental results have a high 95% confidence level. In addition, this paper also analyzes the control variables in the empirical model and finds that test team size can have an impact on test quality by affecting test diversity, which provides ideas for subsequent research. Finally, based on the experimental ideas and empirical results of this study, this paper summarizes the methodological paradigm of applying the natural experiment method to empirically study and analyze the factors affecting test quality, which provides an important reference example for future empirical studies by introducing the natural experiment method in the study of software quality and test quality, and greatly expands the research horizon of related fields.