EDGE-OPTIMIZED DEEP LEARNING FOR ADAPTIVE MALICIOUS ACTIVITY DETECTION IN MOBILE EDGE NETWORKS

Abstract

Mobile Edge Computing (MEC) is an emerging paradigm designed to enhance the performance of 
mobile applications by bringing computation and data storage closer to end-users. However, the 
adoption of MEC has introduced new security challenges, as these systems have become prime targets 
for malicious activities such as data breaches, denial-of-service (DoS) attacks, and intrusion attempts. 
These threats exploit infrastructure vulnerabilities, compromising system integrity, availability, and 
confidentiality. Traditionally, mobile edge security relied on signature-based and rule-based detection 
methods, which could only identify known threats through predefined patterns. Such methods were 
ineffective against evolving or novel attack techniques. As cyber threats have grown more complex, 
the need for advanced detection strategies has led to the integration of machine learning (ML) into 
security frameworks. ML-based approaches enable real-time identification and prevention of both 
known and unknown attacks, significantly enhancing the security posture of MEC environments. The 
limitations of earlier systems, including high false-positive rates, poor detection of novel threats, and 
dependence on manual updates, have driven the shift toward automated ML solutions. By employing 
classifiers such as Decision Trees, Random Forests, and Deep Neural Networks, these systems 
analyze vast datasets to detect and classify malicious behaviors, thereby improving accuracy, 
scalability, and adaptability. Ultimately, ML-powered security models aim to create intelligent, self
evolving mechanisms that continually learn from new data and effectively respond to emerging 
security threats in mobile edge environments.