Marko Misic

Transition from traditional to LMS supported examining: A case study in computer engineering

Modern education is hard to imagine without the use of e‐learning tools, but still the transition from traditional “paper‐and‐pencil” examining to electronic exams is to some extent cautious. Although course administration and exam evaluation are much easier with learning management systems, there are concerns among teaching staff, that electronic tests simplify examination process compared to paper tests and classic examining. At the University of Belgrade, School of Electrical Engineering, e‐learning tools are used at several exams, mostly as a secondary tool to aid with laboratory exercises. In this paper, we show our experience with four courses from the computer engineering study program that are to various extent moved to Moodle LMS, and electronic examining. We mostly concentrate on certain aspects of transition from paper tests to electronic exams. We present 12 different transformations needed to conform to electronic examining and automated evaluation, and discuss benefits and drawbacks of such a transition.

Data sorting using graphics processing units

This paper presents a short survey and performance analysis of parallel sorting algorithms on graphics processing units. Three implementations of the representative sorting algorithms (Quicksort, Merge sort and Radix sort) were evaluated on CUDA platform which is used to execute programs on NVIDIA graphics processing units. Algorithms were carefully tested and evaluated using automated test environment with different datasets, especially those important for particular applications. Finally, the results of this analysis are briefly discussed.

Application of Moodle platform in computer engineering courses

In the modern educational process e-learning tools have an increasing application. There are a number of systems that can be used for various purposes and at various levels of education. This paper presents experiences in using one such system as a platform for teaching in higher education. Moodle platform has been successfully used at several computer engineering courses at the School of Electrical Engineering, University of Belgrade. The paper explaines to what purpose the system was used and how the transition from traditional teaching was made. The perceived advantages and disadvantages of the use of such a system are given.

Application of graphics processing units in audio signal processing

Intensive development of graphics processing units in recent years has led to their use in variety of applications, including the processing of audio signals. This paper presents the viability of those processors and their programming model in dealing with two audio effects - audio amplifier and digital equalizer. The parallel application is implemented as a plug-in for Audacity using CUFFT library for the fast Fourier transform. Evaluation of parallel implementation has shown a significant speedup compared to its sequential counterpart.

Chapter One – A New Course on R&D Project Management in Computer Science and Engineering: Subjects Taught, Rationales Behind, and Lessons Learned

Abstract This chapter describes the essence of a course for senior level undergraduate students and for master students of computer science and engineering, and analyzes its effects. The course prepares students for their professional life after graduation, and especially, it prepares them for the challenges related to efforts to bring new paradigm-shifting ideas into the commercial world. This course was developed to complement a DataFlow course and to teach DataFlow researchers about issues of importance for promotion of their results with a commercial potential. Consequently, course examples and homework assignments were chosen to reflect issues of importance in the commercialization of the DataFlow concept. The course includes the following subjects (presented with DataFlow-related issues in mind): (a) Writing proposals for Research and Development in industry and academia, (b) Understanding the essence of the MBA/PhD degrees and preparing the GMAT/GRE analytical exam, (c) Understanding Capability Maturity Model Integration and learning how to write holistic strategic project plans, (d) Understanding Project Management and learning how to write detailed tactical project plans, (e) Writing business plans for venture capital or business angels, (f) Writing patent applications, (g) Writing survey papers for SCI (Science Citation Index) journals, (h) Writing research papers for SCI journals, (i) Making an Internet shop, (j) Making a MindGenomics campaign for the Internet shop, (k) DataMining from project history and project experiments, and (l) Preservation of project heritage and skills related to brand making. Each subject matter is covered by a homework assignment to help deepen the practical knowledge of the subject matter covered. In addition to the above described, which is accompanied with homework, the following four subjects are also covered and accompanied with in-class discussions (oriented to DataFlow research): (m) Inventivity, (n) Creativity, (o) Effectiveness, and (p) Efficiency. Consequently, the analysis part concentrates on the following issues: (a) Inventivity: How different majors react to the subject matters, (b) Creativity: How efficiently the initial knowledge gaps get bridged, (c) Effectiveness: How the experience of the teacher helps, and (d) Efficiency: How the previous experiences of students help.

Extending valgrind framework with the MIPS MSA support

This paper presents an extension of Valgrind framework for dynamic binary code analysis to support MIPS MSA instruction set which includes instructions for vector (SIMD) processing of integer and floating-point data of different widths. First, a background on MIPS and its MSA extention is given. Then, Valgrind features for code instrumentation are described. Several changes have been made to Valgrind intermediate representation in order to support new instruction set. The implementation issues regarding the structures, formats, and functions needed for extension are discussed. Some low-level details and examples are also presented. Upgraded Valgrind tools are verified using dedicated regression test suite on the Linux platform.

Parallelization of GST algorithm for source code similarity detection

Source code is a frequent target for plagiarism in massive computing courses. Plagiarism detection requires a significant effort from the teaching staff, thus software tools have been used to detect similar source codes. This paper examines parallelization of source code similarity detection based on Greedy-String-Tiling and Karp-Rabin algorithms. CPU implementation is parallelized using Pthreads, while GPU implementation is based on CUDA. Observed speedups range from 4 to 6 over the sequential version of the code.

Simulation of Software Defined Networks in Mininet environment

Management of modern computer networks rapidly becomes more demanding due to complex user needs. Typical user tends to access content from different devices at any time. Due to multimedia content and mobile access, network traffic becomes more intensive, and network management more complex. A new paradigm, Software Defined Networking (SDN) enables network administrators to decouple control plane from data plane, and logically centralize control functions in a software controller. This paper describes SDN architecture and Mininet simulation environment for rapid SDN prototyping. A simple example of a real network is used to demonstrate programming of SDN controllers.

Use case analysis of OpenACC directives in the implementation of image processing algorithms

Graphics processing units have been intensively used in general-purpose computations in the recent years. Although new programming models (CUDA, OpenCL) have been developed and widely applied to support GPU programming, they still impose demanding environments for those users who want to speed up their applications without the knowledge of low level details of the underlying hardware. To cope with this problem, the paper analyzes and evaluates the usage of OpenACC directive-based programming model in the case of image processing algorithms that are amenable for GPU execution.