Publicații

Lucrǎri Publicate la Conferințe

  1. Nandra, C., Bacu, V., Gorgan, D., „Parallel Earth Data Tasks Processing on a Distributed Cloud Based Computing Architecture”. Proceedings of 21st International Conference on Control Systems and Computer Science (CSCS), ISBN: 978-1-5386-1840-0, ISSN 2379-0482, pp. 677-684, (2017).

Abstract: We now live in a society flooded by increasing volumes of data. These data are being generated by our everyday actions and the plethora of machines and sensors which surround us. This is especially true for the fields of study related to Earth Observation. Over the last few decades, new government policies have dramatically increased the volumes of data available to the general public. This increase in availability gives rise to new opportunities of analyzing and gaining knowledge from publicly available data stores. This paper is intended to show the benefits of combining modular processing description methodologies with a distributed execution environment, in order to facilitate the processing of large Earth Observation data volumes. It will briefly touch upon our proposed solution – in the form of the BigEarth platform – highlighting its main characteristics, functionality and the potential benefits of using it in order to speed up the processing of large Earth Observation data sets. To this end, we will present a set of experiments conducted to demonstrate the capabilities of our system to take advantage of the inherent task-level parallelism, thus shortening the overall task execution time.

  1. Nandra, C., Gorgan, D., „Defining Earth Data Batch Processing Tasks by Means of a Flexible Workflow Description Language”, ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume III-4, pp.59-66, (2016)

Abstract: This paper aims to present some of the main features of the Workflow Description Language (WorDeL) and demonstrate their usage in defining Earth Data processing tasks. This description language is based on the flexible description of processing tasks as workflows, composed of basic processing operators. This approach allows the language to offer an intuitive way of representing processing tasks, without requiring programming expertise from its users. It also allows its users to employ and integrate existing functionality into their design, thereby reducing the design complexity and development effort of newly defined processing workflows. WorDeL supports the transparent adaptive parallelization of the processing tasks over high performance computation architectures, such as cloud-based solutions. Throughout the paper, we will exemplify this language’s use in creating flexible, reusable and easy-to-understand earth data processing descriptions, with an emphasis on satellite image processing.

  1. Bacu, V., Stefanut, T., Gorgan, D. “Adaptive Processing of Earth Observation Data on Cloud Infrastructures Based on Workflow Description”, Proceedings of the Intelligent Computer Communication and Processing (ICCP), IEEE-Press, pp.444-454, (2015).

Abstract: The analysis of Earth Observation data is a challenging task due to the variety, velocity and volume of incoming data from various sources. As storing all the raw data is almost impossible, knowledge extraction would be a recommended approach in reducing data size without losing valuable information. For describing the complex processing required to extract knowledge we propose a flexible solution based on workflows and an adaptive execution platform. The main focus of this paper is the Executor component that is oriented on scalability and isolation from the virtual resources management that can be dedicated to a specific cloud infrastructure.

  1. Mihon, D., Bacu, V., Colceriu, V., Gorgan, D. “Modeling of Earth Observation Use Cases through the KEOPS System”, Proceedings of the Intelligent Computer Communication and Processing (ICCP), IEEE-Press, pp. 455-460, (2015).

Abstract: The description of natural phenomena from different Earth Observation (EO) activity domains is represented by complex processes that involve a solid understanding of the phenomena, the syntactic and semantic description of the proposed solutions, the experimental data collection, and the analysis and interpretation of the results. Such a use case scenario is modeled as a collection of operators that are able to generate in a finite amount of time a valid output, based on a range of inputs data sets. The current paper aims at identifying the main EO data processing types and providing a set of basic operators that represent the core of the KEOPS (Kernel Operators) system. At the moment, several researches are conducted to find the best solution of integrating this system within the BigEarth platform, but the main idea is to use KEOPS as a plugin that can fit within any EO related platform that aims at processing spatial data. One main advantage of using the KEOPS system is the possibility of easily extending its core dataset with new operators that fulfill the needs of developing complex use case scenarios.

  1. Nandra, C., Gorgan, D. “Workflow Description Language for Defining Big Earth Data Processing Tasks”, Proceedings of the Intelligent Computer Communication and Processing (ICCP), IEEE-Press, pp. 461-468, (2015).

Abstract: In this paper we are presenting an approach for the flexible description of Earth Observation (EO) data processing tasks with varying degrees of complexity. This approach is based on a simple and compact description language which offers the user the possibility of creating complex processing tasks by representing them in the form of a network or directed graph consisting of independent functional units, called operators. This kind of description allows for the partitioning of the processing tasks into smaller, more manageable processing units which can potentially be run in parallel. By relying on collections of readily available functional units to be employed in the creation of processing tasks, our system attempts to maximize the component reuse rate, while at the same time taking advantage of the design’s modular nature in order to exploit opportunities for task parallelism. To make all of this possible, we envisaged the use of the specially-designed description language, called WorDeL, which we will be describing in detail throughout this paper.

  1. Ilies D., Sabou A., Gorgan D. „Real Time Visualization of Crowd Dynamics Scenarios”. Proceedings of the 12th Romanian Conference on Human-Computer Interaction, ISSN 2344-1690, pp.121-128, (2015).

Abstract: This paper presents an approach to real-time simulation of crowd dynamics on GPU enabled computing architectures. We discuss challenges with parallelization of agent-based models, implementing parallel simulation algorithms, visualization and interaction with the simulated scene and, most importantly, ensuring communication and synchronization between all these processes.  Our main objective is to provide interactive simulation of realistic models such as pedestrian dynamics, in which large crowds move and interact among themselves and with the environment.  Simulation parameters like scene complexity, scene composition, as well as the number of agents are varied in order to simulated different scenarios and to assess the impact on performance.

Publicații in Jurnale

  1. Gorgan D., Catana M.C., Stefanut T. “Tehnici vizuale de analizǎ a datelor masive multidimensionale”. Revista Romana de Interactiune Om-Calculator, Vol. 8(3), pp.237-255, (2015).

Abstract: Lucrarea propune tehnicile de analizǎ vizualǎ ca o soluție pentru comunitatea științificǎ preocupatǎ de dezvoltarea unor algoritmi performanți pentru explorarea și analizarea depozitelor de date masive multidimensionale. Infrastructurile actuale de calcul de înaltǎ performanțǎ cum sunt cloud, grid, multicore sau cluster sunt capabile mai mult sau mai puțin sǎ satisfacǎ cerințele de calcul pentru transformarea, clasificarea și evidențierea datelor semnificative. Este elaboratǎ metrica de evaluare a strategiilor și sunt experimentate tehnici de navigare vizualǎ interactivǎ în spațiul valorilor, pentru a asigura convergența spre soluții optime.

Prezentǎri la Evenimente Științifice

  1. Dorian Gorgan, “Flexible and Adaptive Processing of Earth Observation Data over High Performance Computation Architectures”, invited speaker to International Conference and Exhibition Satellite 2015, August 17-19, Houston, Texas, USA.
  2. Dorian Gorgan, “Visual Analytics on Multidimensional Big Data”, invited speaker to International Conference and Exhibition Satellite 2015, August 17-19, Houston, Texas, USA.
  3. Dorian Gorgan, “Presentation of the PECSA Project”, IASON Project meeting, 22 April, 2015, Thessaloniki, Greece.
  4. Dorian Gorgan, Teodor Stefanut, Victor Bacu, Cosmin Nandra, “Presentation of the PECSA Project”. KEYSTONE Cost Project meeting, 11-12 May, 2015, Kosice, Slovakia.
  5. Teodor Stefǎnuț, Victor Bȃcu, Dorian Gorgan, “PECSA Project Presentation”, ICT2015 Conference, Lisbon, 20-22 October, 2015.
  6. Constantin Ioan Nandra, “The Flexible Description of Geospatial Data Processing“, Universitatea Tehnica din Cluj-Napoca, iunie 2015.
  7. Mihai Bica, Dorian Gorgan, “Domain Analysis and Bibliographic Study about Massive Data Processing Over HPC“, PhD Research Report, UTCN, ianuarie 2015.
  8. Ilies D., Sabou A., Gorgan D., „Simularea unui model socio-fizic de dinamică a mulțimilor folosind tehnici de paralelizare pe GPU”, Raport cercetare UTCN, 2015.

Rapoarte Cercetare la Licențǎ, Masterat și Doctorat

  1. Constantin Ioan Nandra, “The Flexible Description of Geospatial Data Processing“, Universitatea Tehnica din Cluj-Napoca, iunie 2015.
  2. Mihai Bica, Dorian Gorgan, “Domain Analysis and Bibliographic Study about Massive Data Processing Over HPC“, PhD Research Report, UTCN, ianuarie 2015.
  3. Ilies D., Sabou A., Gorgan D., „Simularea unui model socio-fizic de dinamică a mulțimilor folosind tehnici de paralelizare pe GPU”, Raport cercetare UTCN, 2015.