Earticle

In digital forensics, searching for specific patterns in binary data is a crucial task almost always performed following data acquisition. As the volume of digital data surges and the capacity of storage media expands, there is an increasing demand for search techniques that ensure accuracy, completeness, and speed. In this paper, we explore methods to enhance search efficiency through multi-threading and memory-mapped files, while also improving the completeness of searches in corrupted data by adjusting the search resumption position after a match is found. By employing multi-threading, the parallel utilization of multiple processor cores increases the computational speed. Additionally, while some threads occupy the processor, others can access the storage medium, reducing idle times for the storage medium, and thereby maximizing the utility of the storage medium's read speed. Memory-mapped files enable direct access to the operating system’s memory region associated with the storage medium, minimizing the caching and copying stages. This reduces computational overhead and memory load compared to standard file input/output operations. In 64-bit systems, memory-mapped files allow access to files larger than physical memory as a contiguous virtual memory space of up to 2⁶⁴ bytes (16,777,216 terabytes), enabling memory-based implementations of search algorithms to be applied to large files without modification. Experimental results show that the proposed approach significantly enhances both the speed and accuracy of searches in large files, outperforming a method that relies on single-threading and standard file input/output.

이 논문에서는 여러 대의 로봇들이 협력하여 미로를 탈출하는 방법을 제안한다. 논문에서 사용된 교육용 로봇들은 ZigBee로 서로 통신할 수 있으나, 로봇들의 연산기능이 낮아 서로 협력하여 문제를 해결할 수 없다. 로봇의 모션제어로 통로를 직립 보행하도록 하였으며, 절대거리 센서를 이용하여 교차로와 막다른 골목을 인식하면 중앙제어 시스템에 전송하여 명령을 받는다. 여러 로봇들이 동시에 미로에 들어가서 효과적으로 미로를 탐색하도록 하는 미로 탐색 알고리즘을 수정하였다.

A maze-escaping method with cooperating work of robots alongside one another will be proposed in this paper. Educational robots can communicate each other using Zigbee; however, they can't solve problems together due to their lack of arithmetic function. The robots walk upright controlled by a motion program; furthermore, they recognize an intersection or a dead-end in the use of distant sensors with sending data and receiving commands from the central control system. The maze-search algorithms were modified so that all robots can effectively navigate the maze.

In small embedded systems including IoT devices, memory size is very small and it is important to reduce memory amount for execution of application programs. For multi-threaded applications, stack may consume a large amount of memory because each thread has its own stack of sufficiently large size for worst case. This paper presents an implementation of single stack multi-threading, called SSThread (Single Stack Thread), by sharing a stack for all threads to reduce stack memory size. By using SSThread, multi-threaded applications can be programmed based on normal C language environment and there is no requirement of transporting multi-threading operating systems. It consists of several library functions and various C macro definitions. Even though some functional restrictions in comparison to operating systems supporting complete multi-thread functionalities, it is very useful for small embedded systems with tiny memory size and it is simple to setup programming environment for multi-thread applications.

소프트웨어로 MPEG-1 디코더의 구현이 가능하다. 소프트웨어 MPEG-1 디코더의 문제 중 하나는 MPEG-1 압축 알고리즘의 특징상 각각의 영상들이 서로 다른 압축율로 압축되기 때문에 재생시에 디코더에 걸리는 부하가 매우 불규칙적이라는 점이다. 이 문제는 MPEG-1 디코더를 보다 작은 실행 단위인 쓰래드로 나누어 처리함으로써 효율적으로 해결할 수 있다. 이때 독립적인 실행 흐름을 가지는 쓰래드들간의 데이터 전달을 위하여 공유버퍼를 사용하게 된다. 본 논문에서는 다중 쓰래드로 구성된 소프트웨어 MPEG-1 디코더에서 쓰래드들 간의 데이터 전달에 사용되는 공유 버퍼를 이용하여 오디오와 비디오 스트림의 동기화를 효과적으로 수행하는 기법을 소개한다.

지금가지 인지과학적인 연구는 주로 인간의 지능이나 신경망, 그리고 언어를 주 연구 대상으로 다루어 왔다. 그런데 인공지능은 철학의 고유영역에 속하던 많은 문제를 다루게 되었고, 그 주제들을 다뤄온 철학적 방법들을 여러 측면에서 채용하고 있다. 따라서 인지과학과 철학이라는 두 분야가 접목되어야 할 필요성이 있을 것이다. 본 연구는 위에 바탕을 두어 1) 인간 사유에 대한 데카르트의 성찰(cogito ergo sum)을 소개하고, 2) 이를 MFC를 이용한 Multi-threading으로 구현하고 실험하여, 3) 인간의 철학적 사유체계와 사고 중 이성과 오성에 관한 부분은 인공적으로 구현 가능하다는 사실과 응용 가능성을 검토하도록 한다.

효과적인 시각화는 일반적으로 대용량의 디버깅 정보와 프로그램의 추상적 수행모델을 표현하는 것이 중요하다. 본 논문에서는 스레드들간의 논리적 병행성 관계를 나타내는 부분순서 수행그래프를 이용하여 내포병렬성과 임계구역을 가진 OpenMP 병렬 프로그램의 수행양상과 경합정보의 효과적 디버깅을 위해서 효과적으로 제공하는 추상적 시각화 도구를 제안한다. 본 도구는 수행중 추적선택 기법으로 시각화 정보를 위한 공간적 복잡성을 줄이고, 추적된 시각화 정보에 프로그램의 내포병렬성과 임계구역 동기화를 위한 그래프 추상화를 제공하여 시각적 복잡성을 해결한다. 그래프 추상화를 통한 스레드들간의 부분순서 관계와 경합탐지 정보는 프로그램의 제어흐름과 경합의 위치를 구조적으로 파악할 수 있게 하므로 프로그램 수행의 이해와 경합 탐지 및 수정에 효과적이다.

It is important for effective visualization to summarize not only a large amount of debugging information but also the mental models of abstract ideas. This paper presents an abstract visualization tool which provides effective visualization of thread structure and race information for OpenMP programs with critical sections and nested parallelism, using a partial order execution graph which captures logical concurrency among threads. This tool is supported by an on-the-fly trace-filtering technique to reduce space complexity of visualization information, and a graph abstraction technique to reduce visual complexity of nested parallelism and critical sections in the filtered trace. The graph abstraction of partial-order relation and race information is effective for understanding program execution and detecting to eliminate races, because the user can examine control flow of program and locations of races in a structural fashion.

깊이지도를 구하는 방법 중 많이 사용되어지는 방법으로 줄무늬 패턴을 이용하는 방법이 존재한다. 이 방법은 프로젝터-카메라 시스템(Pro-Cam System)을 이용하며 프로젝터로 조사한 패턴을 카메라로 촬영하여 원래의 패턴과 촬영된 패턴간의 기하학적인 관계를 구하여 깊이지도를 구하는 방법이다. 본 논문에서는 이와 같이 구조광을 이용하는 깊이지도 획득 시스템에서 효과적으로 멀티 쓰레드를 사용하여 실시간 처리하는 것을 제안한다. 일반적으로 자주 사용되는 멀티 쓰레딩에는 CPU의 쓰레드를 이용하는 OpenMP와 GPU의 쓰레드를 이용하는 CUDA가 있다. 이 두 가지 기법은 수행하는데 차이점이 존재하기 때문에 상황에 따라 OpenMP가 더 좋은 효율을 보이는 부분이 있고 CUDA가 더 좋은 효율을 보이는 부분이 있다. 따라서 본 논문에서는 이 두 가지에 대해서 각 부분의 특성에 맞게 더 좋은 효율을 보이는 멀티 쓰레드를 적용하였다. 결과적으로 제안된 방법은 $1280{\times}800$의 영상에 대해 25fps 이상의 깊이지도를 획득할 수 있었다.

A depth map can be obtained by projecting/capturing patterns of stripes using a projector-camera system and analyzing the geometric relationship between the projected patterns and the captured patterns. This is usually called structured light technique. In this paper, we propose a new multi-threading scheme for accelerating a conventional structured light technique. On CPUs and GPUs, multi-threading can be implemented by using OpenMP and CUDA, respectively. However, the problem is that their performance changes according to the computational conditions of partial processes of a structured light technique. In other words, OpenMP (using multiple CPUs) outperformed CUDA (using multiple GPUs) in partial processes such as pattern decoding and depth estimation. In contrast, CUDA outperformed OpenMP in partial processes such as rectification and pattern segmentation. Therefore, we carefully analyze the computational conditions where each outperforms the other and do use the better one in the related conditions. As a result, the proposed method can estimate a depth map in a speed of over 25 fps on $1280{\times}800$ images.

A variety of input parameters are taken into consideration while performing a Level 3 PSA. Some parameters related to plume segments, spatial grids, and particle size distribution have flexible input formats. Fine modeling performed by splitting a number of segments or grids may enhance the accuracy of analysis but is time-consuming. Analysis speed is highly important because a considerably large number of calculations is required to handle Level 2 PSA scenarios for a single-unit or multi-unit Level 3 PSA. This study developed a sensitivity analysis supporting interface called MACCSsense to compare the results of the trials of plume segmentation with the results of the base case to determine its impact (in terms of time and accuracy) and to support the development of a modeling approach, which saves calculation time and improves accuracy. MACCSense is an automation tool that uses a large amount of plume segmentation analysis results obtained from MUST Converter and Mr. Manager developed by KAERI to generate a sensitivity report that includes impact (time and accuracy) by comparing them with the base-case result. In this study, various plume segmentation approaches were investigated, and both the accuracy and speed of offsite consequence analysis were evaluated using MACCS as a consequence analysis tool. A simultaneous evaluation revealed that execution time can be reduced using multi-threading. In addition, this study can serve as a framework for the development of a modeling strategy for plume segmentation in order to perform accurate and fast offsite consequence analyses.

The demand for mass calculation of offsite consequence analysis to conduct exhaustive single-unit or multi-unit Level 3 PSA is increasing. In order to perform efficient offsite consequence analyses, the Korea Atomic Energy Research Institute is conducting model optimization studies to minimize the analysis time while maintaining the accuracy of the results. A previous study developed a model optimization method using efficient plume segmentation and verified its effectiveness. In this study, we investigated the possibility of optimizing the model through particle size distribution setting by checking the reduction in analysis time and deviation of the results. Our findings indicate that particle size distribution setting affects the results, but its effect on analysis time is insignificant. Therefore, it is advantageous to set the particle size distribution as fine as possible. Furthermore, we evaluated the effect of multithreading and confirmed its efficiency. Future optimization studies should be conducted on various input factors of offsite consequence analysis, such as spatial grid settings.