딥러닝을 이용한 HWSW 결함 구분 자동화 = HWSW Defects Classification using Deep Learning

다국어 초록 (Multilingual Abstract)

To improve the quality of software, defects must be found and resolved quickly. In embedded systems, hardware and software are tightly coupled. Therefore it is difficult to distinguish whether a defect is a hardware defect or a software defect when a defect occurs. Many studies have been conducted to classify defects; however, since network state or software execution is not considered when classifying defects, the accuracy of defects classification is lowered depending on the system. In this paper, we propose an automated method that improves the accuracy of defects classification by adjusting the metric used for defects classification using deep learning. The proposed method has a defect classification rate of 98.38%, which is higher than a defect classification rate of 94.85% of an existing method. These results show that by adjusting the metric using deep learning, reflecting the characteristics of the system in the defects classification effectively affects the performance.

국문 초록 (Abstract)

소프트웨어의 품질을 향상시키기 위해서는 결함을 빠르게 찾아내고 해결해야 한다. 그러나 임베디드 시스템에서는 하드웨어와 소프트웨어가 밀접하게 연관되어 동작하므로 결함이 발생하였을 때 그 결함이 하드웨어 결함인지 소프트웨어 결함인지를 구분하기가 어렵다. 결함을 구분해주기 위한 연구가 진행되었으나, 결함을 구분할 때 네트워크의 상태나 SW의 실행 상황 등을 고려하지 않아 시스템에 따라 결함 구분의 정확도가 떨어지게 된다. 본 논문에서는 결함을 구분할 때 딥러닝을 이용하여 결함 구분에 사용하는 메트릭을 조절하고 이를 결함 구분에 활용함으로써 결함 구분의 정확성을 높여주는 자동화 방안을 제안한다. 제안하는 방안은 기존의 결함 구분 방안의 결함 구분율 94.85%보다 높은 98.38%의 결함 구분율을 보여주었다. 이러한 결과는 딥러닝을 이용하여 메트릭을 변경함으로써 결함 구분에 시스템의 특성을 반영한 것이 결함 구분 성능에 효과적인 영향을 미치는 것을 보여준다.

참고문헌 (Reference)

1 A. C. Onal, "Weather data analysis and sensor fault detection using an extended IoT framework with semantics, big data, and machine learning" 2037-2046, 2017

2 Constantinides, K., "Software-based online detection of hardware defects: Mechanisms, architectural, and evaluation" 97-108, 2007

3 Bower, F. A., "Online diagnosis of hard faults in microprocessors" 4 (4): 8-es-, 2007

4 "ODROID-XU3"

5 Rahman Fahim, S, "Microgrid Fault Detection and Classification: Machine Learning Based Approach" 13 (13): 3460-, 2020

6 A. Makhshari, "IoT Bugs and Development Challenges" 460-472, 2021

7 Majid Dadashi, "Hardware-Software Integrated Diagnosis for Intermittent Hardware Faults" 363-374, 2014

8 "GEM5 Simulator"

9 Garvit Rajesh Choudhary, "Empirical analysis of change metrics for software fault prediction" 67 : 15-24, 2018

10 Han, Seungjae, "Doctor: An integrated software fault injection environment for distributed real-time systems" 204-213, 1995

11 Yan Cai, "Detecting concurrency memory corruption vulnerabilities" 706-717, 2019

12 Jihyun Park, "Automatic Method for Distinguishing Hardware and Software Faults Based on Software Execution Data and Hardware Performance Counters" 9 (9): 1815-, 2020

13 Park, J., "An embedded software reliability model with consideration of hardware related software failures" 207-214, 2012

14 Dhanalaxmi, B, "A review on software fault detection and prevention mechanism in software development activities" 17 (17): 25-30, 2015