As 5G networks become a bigger part of everyday life, scientists are racing to solve one major problem: security.
A new study suggests that combining encryption with artificial intelligence could dramatically improve the safety of 5G systems while keeping them fast enough for real-time use.
The research, published in the International Journal of Information and Communication Technology, focuses on protecting 5G networks from cyberattacks.
These networks are designed to support extremely fast internet speeds, low delays, and huge numbers of connected devices at the same time.
They are expected to power technologies such as self-driving cars, smart cities, remote healthcare, industrial robots, and advanced mobile apps.
However, the same features that make 5G powerful also make it more vulnerable to cyber threats.
With so many devices connected and so much data moving quickly through the network, attackers may have more opportunities to spread malware, steal information, or disrupt services.
Traditionally, network security has relied on two separate systems. One system encrypts data so unauthorized people cannot read it. Another system, known as an intrusion detection system, watches network activity for suspicious behavior.
The problem is that these systems usually work independently, which can slow down responses to attacks in fast-moving 5G environments.
To solve this issue, the research team developed a new system that combines both functions into one unified model. Their approach uses AES-GCM encryption together with an artificial intelligence model called a Long Short-Term Memory network, or LSTM.
AES-GCM is a well-known encryption method that protects data and checks whether it has been changed during transmission. The LSTM model is a form of deep learning that can study patterns over time. In this case, it analyzes network traffic and learns to recognize unusual behavior that may signal an attack.
Instead of encrypting data first and checking for attacks later, the new system performs both tasks at the same time. This allows the network to remain protected while continuously monitoring for threats.
The results were impressive. The researchers reported an attack detection accuracy of 98.1%, while the false positive rate was only 0.5%. In other words, the system was very good at identifying real threats without wrongly flagging normal activity.
The system was also fast enough for real-time communication. Encryption took about 18.4 milliseconds, while decryption required around 21.7 milliseconds. According to the researchers, these speeds are suitable for modern 5G applications where delays must remain extremely low.
Another important finding was that the system adapted well to different network conditions.
During heavy traffic, encryption delays actually became lower, suggesting the model can adjust dynamically as network demand changes.
The researchers also found that the system used less energy than some traditional encryption-only approaches, which could be especially useful for edge devices and mobile systems with limited power.
The study highlights how combining artificial intelligence with cybersecurity tools may help build safer and smarter 5G networks for the future.
