• Research Associate Projects
  • Disconnection handling in real-time groupware systems The focus and methodology of my work is exemplified in my dissertation where I focus on how disconnection can be handled by the groupware developers while engineering a real-time collaborative systems or synchronous groupware in order to ensure usability of the systems. Banani Roy
  • Adapting Real-time Collaboration in Scientific Workflow Management SystemsScientific Workflow Management systems (such as Galaxy, iPlant, Taverna, Keplar, etc) is different than usual software systems as the workflows are executed in a very structured way and processes which form the workflows are dependent with a particular dataflow direction. In such systems, we have been investigating whether real-time collaboration can increase the usability and efficiency of the systems.  Banani Roy
  • Consistency Handling in Collaborative Scientific WorkflowOne of the main challenges of scientific collaborative system is consistency management - in the face of conflicting concurrent operations by the collaborators [23], [24]. The existing research works use locking techniques where a collaborator gets exclusive Write access to a part of the workflow to facilitate the consistency management [19], [14]. I want to work on efficient locking algorithms that can reduce average waiting time of the collaborators and thus can improve the usability of a collaborative scientific workflow management system. Banani Roy
  • Provenance Supported Workflow Programming Model (ProvMod-Viz)Workflow provenance is important for workflow behavior analysis, data quality measurement, usage pattern mining, fault detection, monitoring, providing user recommendations, resource management and so on. Data intensive workflow systems are never complete without provenance support. We have been developing a workflow programming model that is based on the Python Programming Language, extendable to a broad range of use cases, adaptable to third party tools and offers automated provenance, easy configuration and provenance querying via data visualizations. Banani Roy
  • Dynamic Hadoop/Spark Cluster Creation in Openstack for Supporting Distributed High Performance ComputingIn order to support high performance computing and fault tolerance, we are working on a script-based platform using which we should be able to create Hadoop/Spark Cluster on the fly. We are utilizing Compute Canada’s Openstack platform to host our dynamic clusters. Using the HOT script (heat orchestration script), we create virtual machines or a stack in the openstack environment. After that we extract an inventory file from the output of the heat script that contains host names and ip addresses of the VMs using a python script. We used Ansible, which used the inventory file to setup Hadoop/Spark and all other necessary software stacks (such as opencv) in the VMs. Banani Roy
  • Plugin-based Architecture for Building Distributed High-throughput Image Processing Pipelines We have been working on a plug-in based architecture for supporting processing of large sets of image data. As part of this we are designing high throughput image-based APIs that developers can use for programming image processing tasks in Spark. Banani Roy
  • Cross Language Software Similarity Detection (CRopSIM)As workflow management systems include software tools across various programming, languages, we are working on developing a tool that can detect similar software applications written in various programming languages. Banani Roy
  • Intermediate Big Data Management in Distributed Programming PlatformsWe are developing a data management scheme that will allow us handle intermediate states intelligently or optimally. This scheme will make sure whether intermediate states should be reused by a workflow or regenerated during the execution time. Banani Roy
  • Workflow RecommendationIn a workflow management system, users can build workflows by manually selecting and interconnecting the available processing modules for performing their desired investigations. An essential but time consuming part of a workflow building process is to make a decision about which of the existing modules can be the most suitable one to add after the last module in the pipeline under progress given that a number of modules might exist for the same purpose. A system for automatically suggesting the next module during building a pipeline can be much helpful in this decision making process. We have been working on a technique for automatically suggesting the next module through mining and analyzing association rules from the previously developed workflows. Banani Roy
  • Meta data handlingWe have been working on creating a dictionary based website for describing P2IRC-metadata. Banani Roy
  • CRHM MigrationWe have been working on migrating and re-engineering a legacy hydrological software, named CRHM. Banani Roy
  • Post-Doc Projects
  • Manishankar Mondal
  • MSc Projects
  • Deep Method Argument Reccomendation:
    A deep learning based method's argument reccomendation system. C M Khaled Saifullah
  • ProvMod:
    Provmod is a workflow provenance programming model implemented in Python. Rayhan Ferdous
  • DPLib:
    An automated and configurable provenance logging library in Python. Rayhan Ferdous
  • Consistency Management in Real-Time Collaborative Scientific Workflow Management Systems
    One of the main challenges of a real-time collaborative system is consistency management - in the face of conflicting concurrent operations by the collaborators. In this project, we are working on a novel fine-grained attribute level locking scheme for collaborative scientific workflow management system. Golam Mostaeen
  • SciWorCS: Towards a Collaborative Scientific Workflow Management System
    The data-intensive experiments open up several possibilities of new interesting knowledge discoveries. However, they also impose several challenges for a successful analysis process such as data-intensive job management, efficient use of machine learning algorithms, real-time job monitoring, collaborative workflow composition and so on. In this project, we are working towards an architecture in an attempt to address the above challenges. Golam Mostaeen
  • Automatic Code Clone Validation Using Machine Learning
    Since code clones often negatively impact the maintainability of a software system, a great many numbers of code clone detection techniques and tools have been proposed and studied over the last decade. To detect all possible similar source code patterns in general, the clone detection tools work on syntax level (such as texts, tokens, AST and so on) while lacking user-specific preferences. In this project, we worked a machine learning based approach for automating the validation process. Golam Mostaeen
  • A Data Management Scheme for Micro-Level Modular Computation-intensive Programs in Big Data Platforms, International Symposium on Big Data Management and Analytics BIDMA 2018.
    Big-data analytics or systems developed with parallel distributed processing frameworks (e.g., Hadoop and Spark) are becoming popular for finding important insights from a huge amount of heterogeneous data (e.g., image, text and sensor data). These systems offer a wide range of tools and connect them to form workflows for processing Big Data. In this work, we investigate whether management of the intermediate states can speed up the execution of an image processing pipeline consisting of various image processing tools/APIs in Hadoop Distributed File System (HDFS) while ensuring appropriate reusability, and error monitoring. Debasish Chakroborti
  • Automatically Suggesting Intermediate Modules in a workflow.
    A workflow can be defined as a systematic organization of processing modules that sequentially work on an input dataset in order to produce a desired output. In a workflow management system, users can build workflows (also known as pipelines) by manually selecting and interconnecting the available processing modules for performing their desired investigations. Our research in this work deals with semi-automatic generation of workflows. Debasish Chakroborti
  • Web service performance enhancement for portable devices modifying SOAP security principle, 2017 20th International Conference of Computer and Information Technology (ICCIT).
    Nowadays most of the web applications are bound to Service Oriented Architecture (SOA) such as - inter-operable web services, and their performance and security greatly depend on the backend structure of web service. Web Service has two layers of security - message level security (WS-Security) and transport level security. In WS-Security, embedding of security modules within SOAP message increases the line of code which lengthens the response time and thus decreases the overall performance of web services. Our experiments on different web services showed that, among three security modules of WS-Security of SOAP message, authentication module requires processing time the most, in Table I. Another essential part of web service is WSDL and security for WSDL is emphasized by few researchers recently as it is open to access and vulnerable for presenting SOAP messaging information. Restricted access of WSDL file can play a strong role to increase the authentication and overall security of web service as in. In this paper, we have proposed a new model to improve web service performance modifying the security principle of SOAP message and tethering WSDL file to facilitate authentication. Experimental results on our model have shown significant improvements in the line of code and response time over the traditional approach without compromising any WS-Security module's contrivances (Authentication, Signature, and Encryption).  Debasish Chakroborti
  • Trust-based D2D communications for accessing services in Internet of Things, 2015 18th International Conference on Computer and Information Technology (ICCIT).
    Internet of Things (IoT) is a vision of connecting everything for providing better services efficiently. IoT consists of enormous number of heterogeneous computing devices with different capabilities to provide a diverse range of services situated around the globe. Increasing number of IoT devices and availability of different services in the edge of the networks makes it inevitable to interact among devices. As a result, this open, non-homogeneous and distributed environment also breaches the integrity of secure and reliable device to device (D2D) communication. Traditional access control mechanisms are not prolific to the itinerant, decentralized and dynamic scenarios in the IoT. Trust management is a proven technology for applications like P2P, Grid, and ad hoc network. Hence, this technology can also be used to increase the user reliability in IoT. In this paper, we propose a trust based D2D communication mechanism for accessing different services to meet the growing transactions and successful operations of IoT. We also analyze the effect of adaptive trust parameters for IoT device to device communication in order to access the services. Debasish Chakroborti