Awos Kh. Ali
Unknown Affiliation

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Collective Intelligence for Cybersecurity: Federated Learning under Non-IID Conditions for Intrusion Detection Mohammed, Hutheifa Anwar; Awos Kh. Ali
Sinkron : jurnal dan penelitian teknik informatika Vol. 9 No. 4 (2025): Articles Research October 2025
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/sinkron.v9i4.15017

Abstract

Cyber threats are becoming increasingly complex in cyberspace, which highlights the necessity for strong Intrusion Detection Systems (IDS). However, traditional centralized IDS methods have large problems with data privacy and scalability. Federated Learning (FL) is an intriguing new idea that lets multiple clients train a model together without sharing data directly, which keeps privacy intact. The proposed federated intrusion detection model develops and assesses FL models for detecting network intrusions, focusing on the important issue of non-independent and non-identically distributed (non-IID) data among clients. This work implements and compares two widely recognized FL algorithms, Federated Averaging (FedAvg) and Federated Proximal (FedProx), using a 1D Convolutional Neural Network (CNN) architecture specifically designed for tabular network traffic data. The authors utilize a Dirichlet distribution (α=0.1) to distribute the data among 10, 20, and 30 clients, thereby simulating non-IID conditions in the experiment. The authors thoroughly compare the performance of algorithms using two benchmark datasets: NSL-KDD and NF-Bot-Net-V2. The comparison reveals that while both FedAvg and FedProx achieve high detection rates on NSL-KDD, FedProx is more capable of maintaining stability and converging on the more complex NF-Bot-Net-V2 dataset, achieving an accuracy of 0.9953. The results highlight that FedProx is a more appropriate algorithm for implementing robust and privacy-preserving federated intrusion detection systems in statistically heterogeneous network environments found in the real world.
Reinforcement Learning Evaluation for Dependent Task Offloading in Mobile Edge Computing Systems Lubna Thair; Awos Kh. Ali
Sinkron : jurnal dan penelitian teknik informatika Vol. 10 No. 4 (2026): Research Articles October 2026
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/sinkron.v10i4.16636

Abstract

Dependent-task directed acyclic graphs require each task to execute locally or offload to a mobile edge computing server, coupling latency, communication cost, and device energy. This study reports a controlled evaluation of deep reinforcement learning for offloading under a common simulator, dataset, and train/validation/final-test protocol. Baselines comprise a recurrent policy with proximal policy optimization (PPO-LSTM) and a double deep Q-network; extensions comprise a Transformer policy, discrete soft actor-critic, and stabilized PPO with advantage normalization, scheduled optimization, divergence-based early stopping, and orthogonal initialization. Agents and five heuristics are evaluated on final-test graphs of ten to fifty tasks under latency-oriented and energy-aware quality of experience (QoE) objectives, with five seeds and Holm-corrected tests. Stabilized PPO attained mean QoE of 0.31–0.43 across task-size and bandwidth analyses, though not at every size or bandwidth mean, exceeding Heterogeneous Earliest Finish Time (HEFT) by 0.05 to 0.15 absolute, yet its gain over PPO-LSTM remained about 0.001 and was not Holm-significant at any bandwidth. Both PPO variants reached the validation threshold at median update five, against ten for the Transformer. Ablation identified no single driver: only removing advantage normalization produced a Holm-significant drop, and only under the energy-aware objective. The Transformer remained competitive, whereas discrete soft actor-critic reduced energy rather than latency. Matching the recurrent baseline in size and inference time, stabilization remains viable on resource-constrained devices but cannot replace retraining on much slower links.