Shiyi Kong

Dr. Shiyi Kong

orcid

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 October 2023 in Neural Networks
Verified
Journal Article
Closed access
SKShiyi Kong
Shiyi Kong
JAJun Ai
Jun Ai
MLMinyan Lu
Minyan Lu
YGYiang Gong
Yiang Gong

17 June 2020 in IEEE Access
Verified
Journal Article
Open access
SKShiyi Kong
Shiyi Kong
MLMinyan Lu
Minyan Lu
LLLuyi Li
Luyi Li
LGLihua Gao
Lihua Gao
As with the extensive use of complex software in many fields, such as finance, transportation, aeronautics, and astronautics, software plays an increasingly more important role in society. Software reliability becomes a critical bottleneck for system reliability and draws increasingly greater attention from software engineers. Many researchers find that traditional software verification and validation techniques are not sufficient to ensure complex software reliability, especially after being deployed. In this paper, therefore, we propose a novel software runtime monitoring method that can help with software runtime verification and software failure prediction. This method aims to monitor software execution trace that reveal the software runtime status. First, the framework of this method is proposed to introduce the entire process and two key problems. Second, an instrumentation method based on srcML, an open source tool used to extract the software abstract syntax tree, is presented. By this instrumentation technique, monitoring code can be inserted into source code and runtime data can be exported during the software execution process. Then, a runtime data-collection technique is proposed to collect runtime data exported by monitoring code. On one hand, the file stream mechanism is used to export data files that can be used to support offline analysis, while on the other hand a client-server structure is proposed to support online analysis. Finally, a case study on Nginx is used to show the feasibility of the proposed method.

27 October 2020 in Entropy
Verified
Journal Article
Open access
PDF file
Shuguang WangShuguang Wang
Shuguang Wang
MLMinyan Lu
Minyan Lu
SKShiyi Kong
Shiyi Kong
JAJun Ai
Jun Ai
Software aging is a phenomenon referring to the performance degradation of a long-running software system. This phenomenon is an accumulative process during execution, which will gradually lead the system from a normal state to a failure-prone state. It is a crucial challenge for system reliability to predict the Aging-Related Failures (ARFs) accurately. In this paper, permutation entropy (PE) is modified to Multidimensional Multi-scale Permutation Entropy (MMPE) as a novel aging indicator to detect performance anomalies, since MMPE is sensitive to dynamic state changes. An experiment is set on the distributed database system Voldemort, and MMPE is calculated based on the collected performance metrics during execution. Finally, based on MMPE, a failure prediction model using the machine learning method to reveal the anomalies is presented, which can predict failures with high accuracy.

11 March 2023 in Applied Sciences
Verified
Journal Article
Open access
PDF file
SKShiyi Kong
Shiyi Kong
JAJun Ai
Jun Ai
MLMinyan Lu
Minyan Lu
Software plays a critical role in the infrastructure of modern society. Due to the increasing complexity, it suffers runtime reliability issues. Online anomaly detection can detect partial failures within the program based on manifestations exhibited internally or externally before serious failures occur in the software system, thus enabling timely intervention by operation and maintenance staff to avoid serious losses. This paper introduces CL-MMAD, a novel anomaly detection method based on contrastive learning using multimodal data sources. CL-MMAD uses ResNet-18 to learn the comprehensive feature spaces of software running status. MSE loss is used as the objective to guide the training process and is taken as the anomaly score. Empirical results highlight the superiority of MSE loss over InfoNCE loss and demonstrate CL-MMAD’s effectiveness in detecting both functional failures and performance issues, with a greater ability to detect the latter.

1 July 2018
Verified
Conference Paper
Closed access
SKShiyi Kong
Shiyi Kong
MLMinyan Lu
Minyan Lu
LLLuyi Li
Luyi Li
Software error propagation behaviors often cause system failure by propagating a local error to other components till system failure. Error propagation path consists of faulty modules that fault spread via. Study on error propagation path can assist in fault localization, evaluating the validity of fault injection, and understanding mechanism of error propagation. This paper proposes an approach for tracing error propagation in C/C++ applications working on the Linux platform. A framework on how to trace error propagation in applications is proposed as a guideline. This paper proposes a software execution path modeling method, which can be used to extract error propagation in our approach. Then, an automatic instrumentation and data collection technique is proposed to obtain execution path model parameters. Finally, a method on how to extract error propagation path from execution path model is proposed. A case study with grep shows the feasibility of this approach.

1 November 2020
Verified
Conference Paper
Closed access
SKShiyi Kong
Shiyi Kong
MLMinyan Lu
Minyan Lu
CTChao Tang
Chao Tang
SWShuguang Wang
Shuguang Wang
Software error propagation is a crucial factor leading to software failures. Researchers study on the relationships between software structure measures and error propagation influences, aiming to guide software designing to mitigate the impact brought by error propagation behaviors. Unlike existing researches focusing on the number of nodes that involved in error propagation chains, this paper proposes a novel measure from the aspect of path deviation caused by error propagation, Error Execution Path Deviation (EEPD), to measure the influence of error propagation. It can measure the influences in multi-task systems more precisely because different paths in those systems normally represent different missions. We design fault injection experiments to collect the run-time data supporting the correlation study. The empirical study shows that there are strong relationships between several structure measures and EEPD. The results can assist in finding the error propagation critical nodes and guiding software design to reduce the impact on system caused by error propagation.