Authors : Mr. Pankaj S. Wankhede , Ganesh Khekare , Dr. Amitabh Wahi
DOI : 10.46335/IJIES.2025.10.8.20
Abstract: The current reality is very fast-paced edge computing that shortly will be demanding robust security and scaling solutions for its necessarily distributed and resource-constrained nature. Addressing this need, this paper critically assesses machine learning and deep learning models tailored to enhance both security and scalability in edge computing and points at improvement areas so further work in these regards may be fruitful. On the other hand, most of the surveys about this topic usually suffer from the narrow scope, since most of them miss either a wide spectrum of ML and DL techniques or their concrete applications in the context of edge computing. This work gives a comprehensive review of a good number of ML and DL models, ranging from supervised learning algorithms like Support Vector Machines and Random Forests to unsupervised learning methods like K-means clustering, and advanced deep learning architectures such as Convolutional Neural Networks, Recurrent Neural Networks, and Generative Adversarial Networks. Thus, these models are primarily rated based on their capability to enhance security measures like intrusion detection, data encryption, and anomaly detection, and scalability in handling edge device dynamics and heterogeneity. As noted in the review, among all the models, very few can take such complex patterns of data and realize high accuracy for security applications like the DL models, especially CNNs and GANs. In particular, SVM and RF are known to be robust and efficient for processing small- to medium-sized datasets typical in edge environments. However, some limitations include high computational costs and large training datasets required by the DL models are also discussed. Hybrid approaches combining multiple models have been reviewed to leverage their strengths and make up for the weaknesses of individual models. The comprehensive review has developed several insights that can support future research in strengthening more efficient, scalable, and more secure edge computing frameworks. In particular, gaps and potential improvements related to ML and DL applications stand as greatly important in the progress of edge computing technologies toward better reliability and enhanced performance in real-world scenarios. This thus has a very significant impact on the domain because it provides an in-depth understanding of the current state of capabilities and future directions for ML and DL models at the edge.
Innovative Scientific Publication,
Nagpur, 440036, India
Email:
ijiesjournal@gmail.com
journalijies@gmail.com
© Copyright 2025 IJIES
Developed By WOW Net Technology