This post is also available in:
עברית (Hebrew)
As 5G networks continue expanding into critical infrastructure, autonomous systems, and industrial environments, the demands on network security are increasing rapidly. High-speed connectivity and massive device density create more opportunities for cyberattacks, while the low-latency nature of 5G leaves little room for delays in detecting and responding to threats.
Traditional network protection methods often treat encryption and intrusion detection as separate processes. Encryption secures data during transmission, while intrusion detection systems independently monitor traffic for suspicious activity. In fast-moving 5G environments, this separation can create processing delays and reduce the ability to respond to attacks quickly enough.
Researchers are now proposing a combined approach designed to address that limitation. According to TechXplore, the system merges encryption and anomaly detection into a single parallel architecture, allowing data to remain protected while traffic is continuously analyzed for signs of malicious behavior in real time.
The framework combines AES-GCM encryption with a Long Short-Term Memory (LSTM) neural network. AES-GCM secures data while also verifying that transmitted information has not been altered. The LSTM component, a form of deep learning designed to analyze sequences over time, monitors traffic patterns and identifies unusual activity that may indicate an attack.
According to the reported results, the integrated model achieved a detection accuracy of 98.1% with a false positive rate of 0.5%, meaning it identified threats with relatively few incorrect alerts. Encryption and decryption times remained within tens of milliseconds, performance levels considered suitable for real-time communications and low-latency applications.
Researchers also reported that the system adapts under different network loads. In high-bandwidth conditions, encryption delays reportedly decrease, suggesting the architecture can dynamically adjust to changing traffic demands. The model also reduced overall energy consumption compared to conventional encryption-focused approaches, which may be important for edge computing environments where processing power and battery life are limited.
From a defense and security perspective, resilient 5G protection is becoming increasingly important as military systems, smart infrastructure, and autonomous platforms rely more heavily on connected networks. Faster detection combined with integrated encryption could help secure systems operating in environments where cyberattacks must be identified and mitigated immediately.
The research highlights a broader shift toward combining AI-driven threat detection directly with core network security functions rather than treating them as isolated layers.
The research was published here.
