Deep Learning-Based Cybersecurity Framework for Hybrid Cloud Computing Using LSTM and Image Processing
AbstractHybrid cloud computing has become a preferred infrastructure for enterprises due to itsscalability, flexibility, and cost efficiency. However, integrating private and public cloudenvironments introduces significant cybersecurity challenges, including ransomware attacks,malware infiltration, insider threats, data breaches, and Distributed Denial-of-Service (DDoS)attacks. Conventional security solutions often fail to detect sophisticated zero-day andpolymorphic attacks because they rely heavily on signature-based detection. This paperproposes a Deep Learning-Based Cybersecurity Framework that combines Long Short-TermMemory (LSTM) networks with Image Processing techniques for intelligent threat detectionin hybrid cloud environments. The framework transforms network traffic into visualrepresentations while simultaneously analyzing sequential traffic behavior using LSTM.Image-based feature extraction is performed through Convolutional Neural Networks (CNN),and both models are integrated using a feature fusion mechanism to improve detectionaccuracy. Experimental evaluation on benchmark cybersecurity datasets demonstratesimproved detection accuracy, reduced false positive rates, and faster response timescompared to conventional machine learning techniques. The proposed framework is suitablefor real-time cloud security monitoring and advanced intrusion detection systems.
Authors
- Rajani, Jyoti
Institutions
- Baba Mastnath University (IN)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24411540
- Primary Topic
- Network Security and Intrusion Detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00