International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 1, No. 11, November 2020 | pp. 81–88

Research Article

Title: A Distributed Deep Learning Approach for Privacy-Preserving Collaborative Healthcare Diagnostics

Names of Authors: Aris Papadopoulos¹, Chloe Nicolaou², and Dimitris Angelopoulos³

Authors’ Affiliations:
¹Department of Informatics and Telecommunications, National and Kapodistrian University of Athens, Athens, Greece
²Department of Computer Science, University of Cyprus, Nicosia, Cyprus
³School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece

Abstract: Deep learning models have demonstrated clinical-grade performance in medical imaging diagnostics, yet training these networks requires massive, highly centralized datasets. In healthcare, centralizing electronic health records or medical scans faces severe regulatory bottlenecks due to strict data privacy legislations like GDPR and HIPAA. This paper proposes a privacy-preserving, distributed deep learning framework based on federated learning (FL) combined with differential privacy (DP) for collaborative disease classification. The proposed system enables multiple independent medical institutions to train a global Convolutional Neural Network (CNN) collaboratively without sharing raw patient data. Each institution computes local model weight updates using its internal dataset, and a central orchestrator aggregates these parameters using the Federated Averaging (FedAvg) algorithm. To thwart membership inference and gradient leakage attacks, a Gaussian noise injection mechanism is integrated into the local gradients based on a rigorous differential privacy budget. Simulations using chest X-ray images for pneumonia detection demonstrate that the proposed framework achieves a diagnostic accuracy of 93.4%, which is within a minor 1.2% margin of the baseline centralized training paradigm. Furthermore, the model successfully maintains data privacy under strong adversarial reconstruction models. The findings demonstrate that combining federated learning with differential privacy provides a secure pathway for large-scale medical AI collaboration without compromising patient confidentiality.

Keywords: Federated learning, Privacy preservation, Differential privacy, Healthcare diagnostics, Collaborative deep learning, Medical imaging

Manuscript Timeline: Received: August 14, 2020; Revised: September 20, 2020; Accepted: October 12, 2020; Published: November 1, 2020

Citation: Papadopoulos, A., Nicolaou, C., & Angelopoulos, D. (2020). A distributed deep learning approach for privacy-preserving collaborative healthcare diagnostics. International Journal of Computer Science and Technology, 1(11), 81–88. DOI: 10.46882/2020/IJCST/000011

International Journal of Computer Science and Technology | Vol. 1, No. 10, October 2020 | pp. 73–80

Research Article

Title: Optimizing Container Placement and Migration in Kubernetes Clusters Using Multi-Objective Particle Swarm Optimization

Names of Authors: Sven Lindqvist¹, Astrid Holm², and Frederik Jensen³

Authors’ Affiliations:
¹Department of Computer Science, KTH Royal Institute of Technology, Stockholm, Sweden
²Department of Informatics, University of Oslo, Oslo, Norway
³Department of Information Technology, Technical University of Denmark, Kongens Lyngby, Denmark

Abstract: Containerization technologies managed by Kubernetes orchestration platforms have become the industry standard for deploying microservices and cloud-native applications. However, efficient container scheduling—the assignment of container pods to heterogeneous physical worker nodes—remains a complex NP-hard optimization problem. Default Kubernetes schedulers rely on basic heuristic bin-packing or resource-spreading strategies that often result in server load imbalances, high thermal hotspots, and excessive cross-node network latency. This paper introduces a multi-objective container placement and live migration framework utilizing Multi-Objective Particle Swarm Optimization (MOPSO) to optimize resource utilization across large-scale enterprise clusters. The optimization formulation simultaneously minimizes CPU and memory allocation variance across physical nodes, reduces inter-node network communication overhead measured in megabytes per second (MB/s), and limits total pod migration counts during workload shifts. A custom Kubernetes controller intercepts scheduling requests and feeds cluster telemetry metrics into the MOPSO engine to compute Pareto-optimal deployment configurations. Experimental results conducted on a test cluster running 500 microservice pods across 50 nodes show that the proposed approach improves overall cluster resource utilization balance by 29.3% and reduces inter-node network latency spikes by 35.7% compared to the default Kubernetes kube-scheduler. The findings establish that metaheuristic swarm intelligence effectively handles complex scheduling trade-offs in dynamic cloud environments.

Keywords: Container orchestration, Kubernetes, Particle swarm optimization, Resource allocation, Cloud computing, Microservices migration

Manuscript Timeline: Received: July 12, 2020; Revised: August 18, 2020; Accepted: September 14, 2020; Published: October 1, 2020

Citation: Lindqvist, S., Holm, A., & Jensen, F. (2020). Optimizing container placement and migration in Kubernetes clusters using multi-objective particle swarm optimization. International Journal of Computer Science and Technology, 1(10), 73–80. DOI: 10.46882/2020/IJCST/000010

International Journal of Computer Science and Technology | Vol. 1, No. 9, September 2020 | pp. 65–72

Research Article

Title: An Intelligent Intrusion Detection System for Internet of Things Using Convolutional Neural Networks and Bidirectional LSTM

Names of Authors: Mehmet Yilmaz¹, Ayse Kaya², and Emre Demir³

Authors’ Affiliations:
¹Department of Computer Engineering, Middle East Technical University, Ankara, Turkey
²Department of Artificial Intelligence and Data Engineering, Istanbul Technical University, Istanbul, Turkey
³Department of Cyber Security, Bilkent University, Ankara, Turkey

Abstract: The exponential expansion of Internet of Things (IoT) ecosystems has introduced profound security vulnerabilities due to heterogeneous device hardware, weak default credentials, and lack of standardized firmware security protocols. IoT networks are exceptionally susceptible to botnet infections, distributed scanning, and man-in-the-middle exploits. Conventional signature-based intrusion detection systems fail to identify novel, polymorphic zero-day attacks targeting smart environments. This paper proposes a robust, hybrid deep learning intrusion detection system (IDS) combining 1D Convolutional Neural Networks (CNN) for spatial feature extraction and Bidirectional Long Short-Term Memory (BiLSTM) networks for capturing temporal sequential dependencies in network traffic. The network model processes raw packet streams categorized from benchmark IoT security datasets, including BoT-IoT and TON_IoT. The CNN layers automatically extract salient local correlations from traffic features, while the stacked BiLSTM layers analyze directional context across packet time sequences. Experimental evaluations demonstrate that the proposed CNN-BiLSTM hybrid architecture achieves an exceptional classification accuracy of 99.2%, precision of 98.9%, and recall of 99.1% across multi-class attack categories such as DDoS, data theft, and reconnaissance. Furthermore, the false alarm rate is suppressed to 0.008, confirming high operational reliability. The lightweight deployment analysis indicates that the model executes inference efficiently, making it well-suited for integration into resource-constrained edge gateways protecting smart home and industrial IoT installations.

Keywords: Intrusion detection system, Internet of things, Convolutional neural network, Bidirectional LSTM, Deep learning, Network security

Manuscript Timeline: Received: June 10, 2020; Revised: July 15, 2020; Accepted: August 12, 2020; Published: September 1, 2020

Citation: Yilmaz, M., Kaya, A., & Demir, E. (2020). An intelligent intrusion detection system for internet of things using convolutional neural networks and bidirectional LSTM. International Journal of Computer Science and Technology, 1(9), 65–72. DOI: 10.46882/2020/IJCST/000009

International Journal of Computer Science and Technology | Vol. 1, No. 8, August 2020 | pp. 57–64

Research Article

Title: Blockchain-Based Provable Data Possession Scheme for Secure and Auditable Cloud Data Storage

Names of Authors: Liam O'Connor¹, Sophie Martin², and Antoine Dubois³

Authors’ Affiliations:
¹School of Computer Science, University College Dublin, Dublin, Ireland
²Department of Computer Science, Sorbonne University, Paris, France
³Department of Informatics, Technical University of Munich, Munich, Germany

Abstract: Cloud data outsourcing relieves enterprises and individual users from local hardware maintenance overheads, but it introduces severe trust deficits regarding data integrity. Since cloud service providers (CSPs) may accidentally corrupt files or deliberately hide data loss incidents to protect their reputation, third-party public auditing schemes are necessary. Conventional public auditing architectures rely on a centralized Third Party Auditor (TPA), which can become a single point of failure, collude with dishonest CSPs, or compromise privacy guarantees. This paper proposes a decentralized, blockchain-based Provable Data Possession (PDP) scheme that eliminates the need for a trusted central auditor while ensuring strict data integrity verification. The proposed scheme uses homomorphic authenticators computed on client data blocks before cloud upload, allowing the verifier to check data integrity without downloading the original files. Verification challenges and cryptographic proof responses are recorded and validated through smart contracts deployed on a public-permissioned blockchain ledger, guaranteeing tamper-proof audit trails and automated micro-payment settlements upon successful verification. Performance benchmarks demonstrate that computation time for tag generation scales linearly at 1.8 ms per megabyte (MB), and on-chain verification gas costs remain economically viable. Security analysis confirms that the protocol ensures provable data security, withstands forge attacks, and protects data privacy against malicious auditors.

Keywords: Cloud storage, Provable data possession, Blockchain, Smart contracts, Data integrity, Public auditing

Manuscript Timeline: Received: May 14, 2020; Revised: June 18, 2020; Accepted: July 14, 2020; Published: August 1, 2020

Citation: O'Connor, L., Martin, S., & Dubois, A. (2020). Blockchain-based provable data possession scheme for secure and auditable cloud data storage. International Journal of Computer Science and Technology, 1(8), 57–64. DOI: 10.46882/2020/IJCST/000008

International Journal of Computer Science and Technology | Vol. 1, No. 7, July 2020 | pp. 49–56

Research Article

Title: Deep Reinforcement Learning for Dynamic Resource Allocation in Multi-Access Edge Computing Environments

Names of Authors: Kenji Sato¹, Hiroshi Tanaka², and Yuki Takahashi³

Authors’ Affiliations:
¹Department of Computer Science, University of Tokyo, Tokyo, Japan
²Department of Information Physics and Computing, Kyoto University, Kyoto, Japan
³Department of Communications Engineering, Osaka University, Osaka, Japan

Abstract: Multi-Access Edge Computing (MEC) brings computational and storage resources close to mobile users at the network edge, supporting delay-sensitive services like augmented reality, autonomous driving, and industrial automation. However, managing highly fluctuating user workloads and intermittent wireless channel conditions requires adaptive, real-time resource allocation strategies that avoid manual tuning or rigid static provisioning. Traditional optimization techniques struggle with the combinatorial complexity and non-linear dynamics of multi-user MEC systems. This paper investigates a deep reinforcement learning (DRL) framework based on the Deep Q-Network (DQN) algorithm to optimize dynamic CPU frequency allocation and transmission power control in multi-user edge environments. The system models the resource allocation problem as a Markov Decision Process (MDP) where the state space comprises queue backlogs, channel gains, and task execution deadlines, while the action space dictates CPU core allocation levels and uplink power scaling values. Through continuous interaction with a simulated dynamic wireless environment, the agent learns optimal allocation policies that minimize a composite cost function balancing task latency violation penalties and total energy consumption measured in joules (J). Simulation results indicate that the proposed DRL approach achieves a 31.8% reduction in overall system energy consumption and lowers task drop rates by 14.5% compared to greedy and heuristic baseline allocation strategies. The study demonstrates that reinforcement learning provides robust adaptation to non-stationary mobile traffic patterns.

Keywords: Multi-access edge computing, Deep reinforcement learning, Resource allocation, Edge intelligence, Markov decision process, Power control

Manuscript Timeline: Received: April 12, 2020; Revised: May 20, 2020; Accepted: June 15, 2020; Published: July 1, 2020

Citation: Sato, K., Tanaka, H., & Takahashi, Y. (2020). Deep reinforcement learning for dynamic resource allocation in multi-access edge computing environments. International Journal of Computer Science and Technology, 1(7), 49–56. DOI: 10.46882/2020/IJCST/000007

International Journal of Computer Science and Technology | Vol. 1, No. 6, June 2020 | pp. 41–48

Research Article

Title: Performance Evaluation of Software-Defined Networking Controllers Under Distributed Denial-of-Service Attacks

Names of Authors: Lucas Silva¹, Gabriel Santos², and Mariana Costa³

Authors’ Affiliations:
¹Department of Computer Science, University of São Paulo, São Paulo, Brazil
²Department of Computer Engineering, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil
³Department of Telecommunications, State University of Campinas, Campinas, Brazil

Abstract: Software-Defined Networking (SDN) decouples the control plane from the data plane, centralizing network intelligence and flow management within a programmable controller. While this architecture provides unprecedented flexibility and administrative control, the centralized controller represents a high-value target for distributed denial-of-service (DDoS) attacks, particularly packet-in flooding that saturates control channel bandwidth and processing queues. Evaluating the resilience and operational thresholds of different SDN controllers under high-intensity attack scenarios is essential for enterprise deployment safety. This research provides a comparative performance evaluation of three prominent open-source SDN controllers: ONOS, Ryu, and Floodlight, subjected to simulated TCP SYN flood and UDP flooding attacks. Testbed experiments measure performance degradation across multiple indicators, including maximum throughput measured in megabits per second (Mbps), control plane packet-in processing latency measured in milliseconds (ms), CPU utilization percentage, and packet drop ratio. Experimental findings reveal that ONOS exhibits superior clustering resilience, maintaining stable control packet handling under flood rates up to 50,000 requests/sec before experiencing severe queue congestion. Ryu demonstrates high agility in low-to-medium traffic profiles but encounters rapid resource exhaustion during sustained multi-vector saturation. Furthermore, the study analyzes the mitigation efficiency of an integrated sFlow-based anomaly detection module capable of dynamically installing drop rules on OpenFlow switches. The insights offer practical benchmarks for network architects designing robust, attack-resilient programmable network fabrics.

Keywords: Software-defined networking, SDN controllers, DDoS attacks, Network security, Performance evaluation, OpenFlow

Manuscript Timeline: Received: March 18, 2020; Revised: April 25, 2020; Accepted: May 18, 2020; Published: June 1, 2020

Citation: Silva, L., Santos, G., & Costa, M. (2020). Performance evaluation of software-defined networking controllers under distributed denial-of-service attacks. International Journal of Computer Science and Technology, 1(6), 41–48. DOI: 10.46882/2020/IJCST/000006