International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 8, No. 6, June 2006 | pp. 41–48

DOI: 10.46882/2006/IJCST/080621

Research Article

Title: An Automated Arabic Text Categorization System Based on Support Vector Machines and Chi-Square Feature Selection

Names of Authors: Tariq Al-Farsi¹, Fatima Al-Zahra²

Authors’ Affiliations: ¹Department of Computer Science, Sultan Qaboos University, Muscat, Oman; ²Faculty of Sciences, University of Tunis El Manar, Tunis, Tunisia

Abstract:
Automated categorization of Arabic text documents faces specific challenges because of the language's complex morphology and rich root system. This paper introduces an Arabic text classification framework that uses a custom root-extraction stemming pipeline alongside a Chi-Square feature selection technique. High-dimensional vector models are categorized using a binary-tree Support Vector Machine (SVM) configuration. The platform was evaluated on an in-house corpus containing 7,430 news documents spread across six topical classifications. The system achieved a macro-averaged precision rate of P = 93.45% and an F1-measure index of 92.18%. Incorporating the Chi-Square feature reduction method compressed the vector space size by 82.3%, dropping classifier training times from t = 142 minutes down to t = 18.5 minutes. These metrics validate the framework's capability to process large textual streams for digital libraries and intelligence analysis systems.

Keywords: Text Categorization, Support Vector Machines, Arabic Natural Language Processing, Feature Selection, Stemming

Manuscript Timeline: Received March 10, 2006; Revised April 25, 2006; Accepted May 15, 2006; Published June 23, 2006

Citation: Al-Farsi, T., & Al-Zahra, F. (2006). An Automated Arabic Text Categorization System Based on Support Vector Machines and Chi-Square Feature Selection. International Journal of Computer Science and Technology, 8(6), 41–48.

International Journal of Computer Science and Technology | Vol. 8, No. 5, May 2006 | pp. 33–40

DOI: 10.46882/2006/IJCST/080520

Research Article

Title: Design and Implementation of a High-Performance Distributed Cluster File System for Linux Environments

Names of Authors: Mikhail Volkov¹, Alan Turing Fletcher²

Authors’ Affiliations: ¹Department of Computer Systems and Networks, Bauman Moscow State Technical University, Moscow, Russia; ²Department of Computer Science, University of Edinburgh, Edinburgh, UK

Abstract:
Large-scale computational tasks running on high-performance Linux clusters often experience processing delays due to standard storage systems. This paper presents a high-performance distributed cluster file system designed to remove these performance limits by handling metadata operations separately from parallel data transfers. Storage nodes run a custom block-allocation protocol that allows multiple clients to read and write to raw disk partitions at the same time using direct DMA transfers over InfiniBand networks. Performance benchmarks on a 32-node computing cluster achieved aggregate sequential write speeds of 4.25 GBps and read velocities of 5.12 GBps. The time required for metadata updates remained constant at O(1) through an active distributed lock-caching algorithm. The system recovered from a storage node failure in less than t = 4.5 seconds, proving its reliability for handling heavy data transfers in scientific computing environments.

Keywords: Cluster File System, Distributed Storage, Linux Clusters, High-Performance Computing, InfiniBand

Manuscript Timeline: Received February 12, 2006; Revised March 22, 2006; Accepted April 11, 2006; Published May 26, 2006

Citation: Volkov, M., & Fletcher, A. T. (2006). Design and Implementation of a High-Performance Distributed Cluster File System for Linux Environments. International Journal of Computer Science and Technology, 8(5), 33–40.

International Journal of Computer Science and Technology | Vol. 8, No. 4, April 2006 | pp. 25–32

DOI: 10.46882/2006/IJCST/080419

Research Article

Title: Quality-of-Service Routing Optimization in Mobile WiMAX Networks Using Cross-Layer Feedback Loops

Names of Authors: Ousmane Diop¹, Sarah Jenkins²

Authors’ Affiliations: ¹Département de Génie Électrique, Université Cheikh Anta Diop, Dakar, Senegal; ²Department of Computer Science and Electrical Engineering, University of Maryland, Baltimore, USA

Abstract:
Maintaining strict Quality-of-Service (QoS) guarantees for real-time video streaming over IEEE 802.16 mobile WiMAX architectures is difficult due to variable link quality. This study develops a QoS-aware routing protocol driven by cross-layer feedback loops between the Media Access Control (MAC) layer and the Network routing plane. The protocol monitors channel quality indicators (CQI) and slot allocation metrics to dynamically route high-priority data packets along stable paths. Network simulations with 100 mobile nodes moving at speeds up to 60 km/h show that the cross-layer protocol drops average end-to-end delay to less than t = 38 ms, which is a 31.4% improvement over standard single-layer ad-hoc routing. The jitter index fell by 24.5 ms ± 3.2 ms under heavy network loads, and the connection throughput rose by μ = 1.64 Mbps. These enhancements confirm the model can support steady video transmissions over wireless broadband connections.

Keywords: WiMAX Networks, Quality of Service, Cross-Layer Optimization, Routing Protocol, Mobile Broadband

Manuscript Timeline: Received January 08, 2006; Revised February 20, 2006; Accepted March 14, 2006; Published April 21, 2006

Citation: Diop, O., & Jenkins, S. (2006). Quality-of-Service Routing Optimization in Mobile WiMAX Networks Using Cross-Layer Feedback Loops. International Journal of Computer Science and Technology, 8(4), 25–32.

International Journal of Computer Science and Technology | Vol. 8, No. 3, March 2006 | pp. 17–24

DOI: 10.46882/2006/IJCST/080318

Research Article

Title: Synthetic Aperture Radar Image Segmentation Using Modified Fuzzy C-Means and Wavelet Texture Profiles

Names of Authors: Igor Ivanov¹, Chao-Chun Lin²

Authors’ Affiliations: ¹Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russia; ²Department of Electrical Engineering, National Cheng Kung University, Tainan, Taiwan

Abstract:
Speckle noise presents a persistent challenge in extracting distinct land-cover regions from Synthetic Aperture Radar (SAR) imagery. This study implements an un-supervised image segmentation architecture that incorporates multi-scale wavelet texture features into a modified Fuzzy C-Means (FCM) clustering algorithm. The algorithm extracts spatial context descriptors from spatial gray-level dependence matrices in the wavelet domain, while a local spatial penalty function balances gray-level variations during clustering. The algorithm was validated on radar imagery from the ERS-2 satellite. Experimental results demonstrate a segmentation classification accuracy of 94.12%, outperforming standard FCM techniques by 11.4%. The modified penalty framework reduced speckle noise distortion indexes by a factor of 2.6 without losing structural boundary details. The algorithm completed full image matrix cluster profiling within an average window of t = 1.84 seconds, establishing a reliable process for automated satellite mapping systems.

Keywords: Synthetic Aperture Radar, Image Segmentation, Fuzzy C-Means, Wavelet Transform, Texture Analysis

Manuscript Timeline: Received December 02, 2005; Revised January 15, 2006; Accepted February 08, 2006; Published March 24, 2006

Citation: Ivanov, I., & Lin, C. C. (2006). Synthetic Aperture Radar Image Segmentation Using Modified Fuzzy C-Means and Wavelet Texture Profiles. International Journal of Computer Science and Technology, 8(3), 17–24.

International Journal of Computer Science and Technology | Vol. 8, No. 2, February 2006 | pp. 9–16

DOI: 10.46882/2006/IJCST/080217

Research Article

Title: Detecting Distributed Denial of Service Attacks via Radial Basis Function Neural Networks

Names of Authors: Ahmed Al-Ghamdi¹, Linda Cartwright²

Authors’ Affiliations: ¹Department of Computer Science, King Abdulaziz University, Jeddah, Saudi Arabia; ²Department of Computer Science and Technology, University of Cambridge, Cambridge, UK

Abstract:
Distributed Denial of Service (DDoS) attacks pose a severe threat to enterprise infrastructure by overwhelming internet servers with high volumes of malicious network traffic. This paper proposes a network-based intrusion detection engine that uses a Radial Basis Function (RBF) neural network to isolate malicious traffic anomalies in real time. The model extracts packet entropy values, IP source variation rates, and protocol distribution ratios from active network streams. The RBF model was trained on the benchmark DARPA intrusion dataset using a k-means clustering approach to optimize hidden layer center positions. The neural classifier achieved a true positive detection rate of 98.76% for flooding attacks (TCP SYN and UDP floods) while keeping the false positive rate minimal at 0.08%. The classification processing pipeline handled peak incoming rates of 85,000 packets per second, demonstrating its readiness for deployment within high-volume network security infrastructures.

Keywords: DDoS Detection, Radial Basis Function, Neural Networks, Network Security, Packet Entropy

Manuscript Timeline: Received November 05, 2005; Revised December 14, 2005; Accepted January 10, 2006; Published February 23, 2006

Citation: Al-Ghamdi, A., & Cartwright, L. (2006). Detecting Distributed Denial of Service Attacks via Radial Basis Function Neural Networks. International Journal of Computer Science and Technology, 8(2), 9–16.

International Journal of Computer Science and Technology | Vol. 9, No. 1, January 2007 | pp. 1–8

DOI: 10.46882/2007/IJCST/090129

Research Article

Title: An Energy-Efficient Duty-Cycle MAC Protocol for Ultra-Dense Wireless Sensor Networks

Names of Authors: Hans-Werner Schmidt¹, Fatma Al-Jahani²

Authors’ Affiliations: ¹Institut für Informatik, Technische Universität München, Munich, Germany; ²Department of Computer Science, Qatar University, Doha, Qatar

Abstract:
Data link layer interference and channel idle listening are primary factors behind premature battery drain in dense wireless sensor networks (WSNs). This paper introduces an adaptive duty-cycle Media Access Control (MAC) protocol designed to reduce energy consumption. The protocol uses a localized slot-allocation strategy that adjusts active listening durations based on structural queue thresholds and neighbor transmission counts. Network simulations with 400 sensor nodes distributed across a 1,000-meter line setup showed that the protocol extended the total operational network lifetime by 38.2% compared to standard S-MAC implementations. The average node energy consumption stabilized at P = 11.4 μW, while the packet loss index fell by 22.4% during burst data periods, confirming that adaptive sleep scheduling can improve network lifespan without causing data delivery bottlenecks.

Keywords: Wireless Sensor Networks, MAC Protocol, Duty-Cycle, Energy Efficiency, Sleep Scheduling

Manuscript Timeline: Received October 11, 2006; Revised November 22, 2006; Accepted December 15, 2006; Published January 24, 2007

Citation: Schmidt, H. W., & Al-Jahani, F. (2007). An Energy-Efficient Duty-Cycle MAC Protocol for Ultra-Dense Wireless Sensor Networks. International Journal of Computer Science and Technology, 9(1), 1–8.