Developing Cybersecurity Strategies for IP-Enabled Devices

Background Public services and critical infrastructures now face potential cybersecurity attacks due to their increasing reliance on interconnected digital networks. Such exposure includes ransomware and distributed denial-ofservice (DDoS) attacks, which can halt service delivery, lead to compromised data, and ultimately endanger public safety. The more we digitalize, the stronger the need becomes for well robust, intelligent intrusion detection systems. This paper proposes a Random Forest Classifier-based Long Short-Term Memory (RFCDLSTM). To enhance cyber protection of our public services and critical infrastructures. The hybrid model combines the ensemble learning capability of Random Forest with the sequential learning capabilities of LSTM in a fully integrated classifier to enhance the ability to detect cybersecurity threats. The proposed attained 99.95% accuracy, 99.95% precision, 99.95% recall, and 99.95%, F1-score outperforms the current ones in terms of efficiency. RFCDLSTM ensures trusted, real-time cyber threat detection, significantly improving public service infrastructure defence against modern security risks.