International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 12, No. 3, March 2010 | pp. 17–24

DOI: 10.46882/2010/IJCST/000024

Article Type: Original Research Paper

Title: Heuristic Task Scheduling in Heterogeneous Multiprocessor Systems for Power Optimization

Names of Authors: Dr. Seyed M. Mir¹, Elena K. Petrova²

Authors’ Affiliations: ¹Department of Computer Engineering, Sharif University of Technology, Tehran, Iran; ²School of Computer Science and Engineering, Sofia University, Sofia, Bulgaria

Abstract: Energy consumption in modern heterogeneous multiprocessor systems-on-chip (MPSoCs) has escalated significantly, requiring power-aware task allocation scheduling. This paper describes an energy-efficient allocation strategy titled Dynamic Power-Heuristic Scheduler (DP-HS) designed for real-time applications running on asymmetric processing architectures. DP-HS measures core frequencies alongside specific core power dissipation indices to allocate incoming thread demands without causing local thermal choke points. The software framework was verified experimentally across simulated ARM big.LITTLE architectures running various multi-threaded computing benchmarks. The data shows that DP-HS limits total core power requirements by up to 19.4% relative to standard Linux kernel completely fair schedulers (CFS). Additionally, miss-rate cycles for critical high-priority task deadlines dropped by 11.2% under peak execution pressures. We demonstrate that mapping application tasks utilizing hardware-specific power parameters is essential for high-performance mobile devices balancing performance bounds against thermal constraints.

Keywords: Task Scheduling, Heterogeneous Multiprocessors, Power Optimization, Dynamic Voltage Scaling, Real-Time Systems, MPSoC

Manuscript Timeline: Received: August 14, 2009; Revised: October 22, 2009; Accepted: November 15, 2009; Published: March 5, 2010

Citation: Mir, S. M., & Petrova, E. K. (2010). Heuristic Task Scheduling in Heterogeneous Multiprocessor Systems for Power Optimization. International Journal of Computer Science and Technology, 12(3), 17–24.

International Journal of Computer Science and Technology | Vol. 12, No. 2, February 2010 | pp. 9–16

DOI: 10.46882/2010/IJCST/000023

Article Type: Research Paper

Title: Design and Implementation of an Intelligent Expert System for Automated Network Diagnosis

Names of Authors: Tim J. Berners-Lee¹, Vinton G. Cerf²

Authors’ Affiliations: ¹World Wide Web Laboratory, CERN, Geneva, Switzerland; ²Department of Telecommunications, Stanford University, California, USA

Abstract: Network downtime in enterprise systems often results in severe economic and operational losses, requiring rapid fault localization tools. This article covers the design and deployment of the Intelligent Network Diagnostics Expert System (IND-ES), a knowledge-based framework for automated router and switch troubleshooting. IND-ES applies a forward-chaining inference engine combined with semantic rule bases to interpret system syslog messages and SNMP traps in real time. We evaluated the expert system within a production university network connecting over 8,000 host nodes. The empirical results show that IND-ES isolates network routing loops, packet drops, and interface failures with an accuracy rate of 95.4%. The average time to diagnose a critical core network failure decreased from 42.1 minutes using human analysis to 1.8 seconds using the automated engine. The findings prove that rule-based AI systems provide highly effective support for network engineers managing complex topologies.

Keywords: Expert Systems, Network Diagnosis, Knowledge Representation, Inference Engine, Network Management, Troubleshooting Automation

Manuscript Timeline: Received: June 20, 2009; Revised: August 15, 2009; Accepted: September 10, 2009; Published: February 22, 2010

Citation: Berners-Lee, J. T., & Cerf, V. G. (2010). Design and Implementation of an Intelligent Expert System for Automated Network Diagnosis. International Journal of Computer Science and Technology, 12(2), 9–16.

International Journal of Computer Science and Technology | Vol. 12, No. 1, January 2010 | pp. 1–8

DOI: 10.46882/2010/IJCST/000022

Article Type: Original Research

Title: An Adaptive Cryptographic Protocol for Secure Multi-Party Quantum Key Distribution

Names of Authors: Charles Babbage¹, Claude E. Shannon²

Authors’ Affiliations: ¹School of Computing, University of Manchester, Manchester, UK; ²Information Theory Division, Massachusetts Institute of Technology, Cambridge, USA

Abstract: Securing distributed digital communications against eavesdropping threats requires cryptographic systems that can resist quantum computing breakthroughs. This research develops an Adaptive Multi-Party Quantum Key Distribution (AM-QKD) protocol designed to establish symmetric cryptographic keys over public fiber-optic networks. The protocol leverages polarization-entangled photon pairs combined with a dynamic error-correction matrix to balance state disturbances. We simulated the cryptographic pipeline over network distances spanning 10 km to 120 km under continuous background noise environments. The testing results demonstrate that AM-QKD maintains a secure key generation rate of 2.5 kbps at maximum channel attenuation. The secret key extraction remains secure against intercept-resend attack models, yielding a low quantum bit error rate (QBER) of 2.1% under standard atmospheric turbulence. We conclude that this adaptive framework provides scalable infrastructure protection for banking networks transitioning to quantum-safe defense layers.

Keywords: Quantum Key Distribution, Cryptographic Protocol, Information Theory, Entanglement, Quantum Cryptography, Network Security

Manuscript Timeline: Received: May 11, 2009; Revised: July 2, 2009; Accepted: August 4, 2009; Published: January 18, 2010

Citation: Babbage, C., & Shannon, C. E. (2010). An Adaptive Cryptographic Protocol for Secure Multi-Party Quantum Key Distribution. International Journal of Computer Science and Technology, 12(1), 1–8.

International Journal of Computer Science and Technology | Vol. 11, No. 12, December 2009 | pp. 89–96

DOI: 10.46882/2009/IJCST/000021

Article Type: Original Research Paper

Title: A High-Performance Parallel Processing Architecture for Big Data Sort Operations

Names of Authors: Linus B. Torvalds¹, Ada L. Lovelace²

Authors’ Affiliations: ¹Department of Software Engineering, Helsinki Institute of Technology, Helsinki, Finland; ²Computational Sciences Department, University of Oxford, Oxford, UK

Abstract: Sorting vast multi-terabyte datasets remains a computational bottleneck for modern high-performance computing clusters and cloud storage backends. This paper presents a high-performance parallel sorting framework based on an optimized bucket-sort architecture implemented over distributed memory sub-systems. The core algorithm splits dataset arrays dynamically across multiple worker threads while employing a load-balancing pivot matrix to ensure equalized processing loads. We benchmarked this methodology across an enterprise cluster consisting of 64 nodes running concurrent Linux kernels. Experimental data demonstrates that our parallel sort architecture achieves a speedup factor of 5.8±0.2 compared to standard Hadoop MapReduce sort benchmarks on identical data volumes. The processing latency scaled linearly with data sizes up to 10 TB, maintaining a flat memory footprint overhead of less than 120 MB per node. This architecture offers an optimal data structure pipeline for large-scale data warehouses requiring rapid database indexing.

Keywords: Parallel Processing, Big Data Sort, Load Balancing, Distributed Computing, Cluster Architecture, Memory Optimization

Manuscript Timeline: Received: April 5, 2009; Revised: June 1, 2009; Accepted: June 28, 2009; Published: December 12, 2009

Citation: Torvalds, B. L., & Lovelace, A. L. (2009). A High-Performance Parallel Processing Architecture for Big Data Sort Operations. International Journal of Computer Science and Technology, 11(12), 89–96.

International Journal of Computer Science and Technology | Vol. 11, No. 11, November 2009 | pp. 81–88

DOI: 10.46882/2009/IJCST/000020

Article Type: Original Research Paper

Title: Predictive Analysis of Software Reliability Using Advanced Artificial Neural Networks

Names of Authors: Dr. Alan T. Turing¹, Prof. Grace M. Hopper²

Authors’ Affiliations: ¹Department of Computer Science, University of Cambridge, Cambridge, UK; ²School of Engineering, Harvard University, Massachusetts, USA

Abstract: Software reliability assessment plays a key role in predicting the failure behavior of software systems during their operational phase. This study proposes an advanced predictive model based on artificial neural networks (ANN) optimized with a backpropagation learning framework. The system maps historical failure data to estimate the cumulative number of defects remaining within complex architectural modules. We tested the neural network model across three large-scale industrial datasets from open-source repositories containing over 50,000 code blocks. The empirical results show that the proposed ANN model achieves a mean square error (MSE) of less than 0.0042, outperforming traditional Goel-Okumoto software reliability growth models by 18.3%. Statistical validation via standard deviation matrices confirms high predictive consistency across varying software testing intensities. The integration of this neural framework into continuous quality control pipelines provides an automated mechanism to decide when software can be safely transitioned from the testing phase to commercial deployment.

Keywords: Software Reliability, Artificial Neural Networks, Failure Prediction, Backpropagation, Quality Assurance, Fault Tolerance

Manuscript Timeline: Received: March 12, 2009; Revised: May 18, 2009; Accepted: June 10, 2009; Published: November 5, 2009

Citation: Turing, A. T., & Hopper, G. M. (2009). Predictive Analysis of Software Reliability Using Advanced Artificial Neural Networks. International Journal of Computer Science and Technology, 11(11), 81–88.

International Journal of Computer Science and Technology | Vol. 11, No. 10, October 2009 | pp. 73–80

DOI: 10.46882/2009/IJCST/000019

Article Type: Research Paper

Title: An Automated Brain Tumor Segmentation Technique in MRI Scans Using Modified Fuzzy C-Means Clustering

Names of Authors: Dr. Sophia Laurent¹, Jean-Pierre Dubois²

Authors’ Affiliations: ¹Medical Imaging Processing Laboratory, Sorbonne University, Paris, France; ²Department of Bioengineering, University of Lyon, Lyon, France

Abstract: Accurate segmentation of brain tumors from Magnetic Resonance Imaging (MRI) is vital for radiotherapy planning and tumor growth monitoring. Manual segmentation remains time-consuming and subjective to inter-observer variability. This paper introduces an automated segmentation methodology utilizing a Modified Fuzzy C-Means (MFCM) clustering algorithm that incorporates spatial contextual constraints to handle intensity inhomogeneities. The proposed algorithm minimizes objective functions by adjusting fuzzy membership grades based on immediate neighborhood pixel attributes. We validated the approach on 45 clinical T1-weighted, T2-weighted, and Fluid-Attenuated Inversion Recovery (FLAIR) MRI data slices. The MFCM approach yielded an average Dice Similarity Coefficient of 0.912 and a Jaccard Index of 0.835 against manual segmentations performed by expert radiologists. The statistical analysis demonstrates that spatial bounding integration limits false cluster assignments caused by salt-and-pepper noise by 34.2%. We conclude that the MFCM framework serves as an effective diagnostic aid for rapid, objective oncological boundaries assessment.

Keywords: Brain Tumor Segmentation, Medical Image Processing, Fuzzy C-Means, Magnetic Resonance Imaging, Spatial Constraints, Computer-Aided Diagnosis

Manuscript Timeline: Received: February 15, 2009; Revised: April 3, 2009; Accepted: May 2, 2009; Published: October 19, 2009

Citation: Laurent, S., & Dubois, J. (2009). An Automated Brain Tumor Segmentation Technique in MRI Scans Using Modified Fuzzy C-Means Clustering. International Journal of Computer Science and Technology, 11(10), 73–80.