ACCS text
Log In | Search | Sitemap
beta
Conferences
Institutional Members
image image
image image
image
ADCOM 2009 Proceedings
1. Parallel Implementation of Video Surveillance Algorithms on GPU Architecture using NVIDIA CUDA
Authors: Sanyam Mehta, Arindam Misra, Ayush Singhal, Praveen Kumar, Ankush Mittal, Kannappan Palaniappan
Abstract: At present high-end workstations and clusters are commonly used hardware for the problem of real-time video surveillance. Through this paper we propose a real time framework for a 640x480 frame size at 30 frames per second (fps) on an NVIDIA graphics processing unit (GPU) (GeForce 8400 GS) costing only Rs. 4000 which comes with many laptops and PC's. The processing of surveillance video is computationally intensive and involves algorithms like Gaussian Mixture Model (GMM), Morphological Image operations and Connected Component Labeling (CCL). The challenges faced in parallelizing Automated Video Surveillance (AVS) were: (i) Previous work had shown difficulty in parallelizing CCL on CUDA due to the dependencies between sub-blocks while merging (ii) The overhead due to a large number of memory transfers, reduces the speedup obtained by parallelization. We present an innovative parallel implementation of the CCL algorithm, overcoming the problem of merging. The algorithms scale well for small as well as large image sizes. We have optimized the implementations for the above mentioned algorithms and achieved speedups of 10X, 260X and 11X for GMM, Morphological image operations and CCL respectively, as compared to the serial implementation, on the GeForce GTX 280.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
2. Adapting Traditional Compilers onto Higher Architectures incorporating Energy Optimization Methods for Sustained Performa
Authors: Prahlada Rao B B, Mangala N, Amit K S Chauhan
Abstract: Improvements in processor technology are offering benefits, such as large virtual address space, faster computations, non-segmented memory, higher precision etc., but require upgradation of system software to be able to exploit the benefits offered. The Authors in this paper present various tasks and constraints experienced to enhance compilers from 32-bit to 64- bit platform. The paper describes various aspects, ranging from design changes, to porting and testing issues that have been dealt with while enhancing the C-DAC's Fortran90 (CDF90) compiler to work with 64-bit architecture and I: 8 support. Features supported by CDF90, for energy efficiency of code, are presented. The regression testing carried out to test I: 8 supports added in the compiler are discussed.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
3. Is I/O virtualization ready for End-to-End Application Performance?
Authors: J. Lakshmi, S. K. Nandy
Abstract: Workload consolidation using system virtualization feature is the key to many successful green initiatives in data centers. In order to exploit the available compute power, such systems warrant sharing of other hardware resources like memory, caches, I/O devices and their associated access paths among multiple threads of independent workloads. This mandates the need for ensuring end-to-end application performance. In this paper we explore the current practices for I/O virtualization, using sharing of the network interface card as the example, with the aim to study the support for end-to-end application performance guarantees. To ensure end-to-end application performance and limit interference caused due to the sharing of devices, we present an evaluation of previously proposed end-to-end I/O virtualization architecture. The architecture is an extension to the PCI-SIGV specification of I/O hardware to support reconfigurable device partitions and uses VMM bypass technique for device access by the virtual machines. Simulation results of the architecture for application quality of service guarantees demonstrate the flexibility and scalability of the architecture.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
4. Eco-Friendly Features of a Data center OS
Authors: Surya Prakki
Abstract: This paper presents the different technologies a modern operating system like Open Solaris offers which will help data centers to become more eco-friendly. It starts off with the various virtualization technologies Open Solaris has got which help in driving consolidation of systems thus reducing system foot print in a data center. It introduces the power aware dispatcher (PAD), how it plays well with the various processor supported power states, and moves onto observability tools which prove how well the system is using the power management features.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
5. Low Power Biometric Capacitive CMOS Fingerprint Sensor System
Authors: B. Shankkar, Tarun Kumar and Roy Paily
Abstract: A charge sharing based sensor for obtaining fingerprint has been designed. The design used sub threshold operations of MOSFETs for achieving low power sensor device working at 0.5 V. Also the interfacing circuitry and a fingerprint matching algorithm were also designed to support the sensor and complete a fingerprint verification system.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
6. Particle Swarm Optimization for Feature Selection: An application to Fusion of Palm Print and Face
Authors: Raghavendra.R1 Bernadette Dorizzi2 Ashok Rao3 Hemantha Kumar.G1
Abstract: This paper relates to multimodal biometric analysis. Here we present an efficient feature level fusion and selection scheme that we apply on face and palm-print images. The features for each modality are obtained using Log Gabor transform and concatenated to form a fused feature vector. We then use a Particle Swarm Optimization (PSO) scheme, a random optimization method to select the dominant features from this fused feature vector. Final classification is performed in the projection space of the selected features using Kernel Direct Discriminant Analysis. Extensive experiments are carried out on a virtual multimodal biometric database of 250 users built from the face FRGC and the palm print PolyU databases. We compare the proposed selection method with the well known feature selection methods such as Sequential Floating Forward Selection (SFFS), Genetic Algorithm (GA) and Adaptive Boosting (AdaBoost) in terms of both number of features selected and performance. Experimental results show that the proposed method of feature fusion and selection using PSO outperforms all other schemes in terms of reduction of the number of features and it corresponds to a system that is easier to implement, while showing the same or even better performance.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
7. OpenPEX: An Open Provisioning and Execution System for Virtual Machines
Authors: Srikumar Venugopal, James Broberg and Rajkumar Buyya
Abstract: Virtual machines (VMs) have become capable enough to emulate full-featured physical machines in all aspects. Therefore, they have become the foundation not only for flexible data center infrastructure but also for commercial Infrastructure-as-a-Service (IaaS) solutions. However, current providers of virtual infrastructure offer simple mechanisms through which users can ask for immediate allocation of VMs. More sophisticated economic and allocation mechanisms are required so that users can plan ahead and IaaS providers can improve their revenue. This paper introduces OpenPEX, a system that allows users to provision resources ahead of time through advance reservations. OpenPEX also incorporates a bilateral negotiation protocol that allows users and providers to come to an agreement by exchanging offers and counter-offers. These functions are made available to users through a web portal and a REST based Web service interface.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
8. Exploiting Heterogeneity in Grid Computing for Energy-Efficient Resource Allocation
Authors: Saurabh Kumar Garg and Rajkumar Buyya
Abstract: The growing computing demand from industry and academia has lead to excessive power consumption which not only impacting long term sustainability of Grids like infrastructures in terms of energy cost but also from environmental perspective. The problem can be addressed by replacing with more energy efficient infrastructures, but the process of switching to new infrastructure is not only costly but also time consuming. Grids being consist of several HPC centers under different administrative domain make problem more difficult. Thus, for reduction in energy consumption, we address the challenge by effectively distributing compute-intensive parallel applications on grid. We presented a meta-scheduling algorithm which exploits the heterogeneous nature of Grid to achieve reduction in energy consumption. Simulation results show that out algorithm HAMA can significantly improve the energy efficiency of global grids by a factor of typically 23% and as much as a factor of 50% in some cases while meeting user's QoS requirements.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
9. Intelligent Data Analytics Console
Authors: Snehal Gaikwad, Mihir Kedia, Aashish Jog & Bhavana Tiple
Abstract: The problem of integrating data analysis, visualization and learning is arguably at the very core of the problem of data intelligence. In this paper, we review our research work in the area of data analytics and visualization focusing on four interlinked directions of research-(1) data collection, (2) data analytics and visual evidence, (3) data visualization, and (4) intelligent user interfaces - that contribute and complement each other. Our research resulted in Intelligent Data Analytics Console (IDAC), an integration of the above four disciplines. The main objectives of IDAC are to - (1) provide a rapid development platform for analyzing the data and generating components that can be used readily in software applications, (2) provide visual evidences after each data analytical operation that will help user to learn the behavior of data, and (3) provide the user-centric platform for skilled data analytics experts as well as naive user. The paper presents development process of the user-centric intelligent software equipped with effective visualization. This approach should help business organizations to develop better data analytics software using open source technologies.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
10. Modified Greedy Search Algorithm for Bi-clustering Gene Expression Data
Authors: Shyama Das & Sumam Mary Idicula
Abstract: Bi-clustering refers to simultaneous clustering of both rows and columns of a data matrix. Bi-clustering is a highly useful data mining technique in the analysis of gene expression data. The problem of identifying the most significant bi-clusters in gene expression data has shown to be NP complete. In this paper a greedy search algorithm is developed for bi-clustering gene expression data. This algorithm has two steps. In the first step high quality bi-cluster seeds are generated using K-Means clustering algorithm. Then these seeds are enlarged using the greedy search method. Here the node that results in the minimum Hscore value when combined with the bi-cluster is selected. This is added to the bi-cluster. This selection and addition is continued till the Hscore value of the bi-cluster reaches the given threshold. Even though it is a greedy method the results obtained is far better than many of the meta-heuristic methods which are generally considered superior to greedy approach.
Access Full Text:
(Members Only)

(For Non ACCS members each paper is priced at Rs.100)
Pdf

 
FIRST PREV ( Page 1 of 4 ) NEXT LAST
UPCOMING EVENTS
The 18th edition of ADCOM 2011
ADCOM 2011 will be presented during October, 2011. The event this year will feature a new format of invited speakers and guest lectures. The ACCS-CDAC Foundation Award for 2011 will be conferred at ADCOM 2011. Please refer to the Call for Participation for more details on the event.

Stay tuned for further announcement

........................................................................