Swarm-based intelligent models for developing cybersecurity frameworks with IDS

The need for real-time and robust monitoring system has become most important with the exponential growth of networked physical and cyber threats. This paper focuses on the design and implementation of an intrusion detection System by using swarm-based intelligent model. This proposed system is capable of detecting the threats in real-time to prompt timely responses by leveraging temporal data analytics. The main objective of this paper is to minimize the potential damages with timely threat identification by developing scalable models so that these models can process and analyze the real-time data. To achieve this objective, we are proposing a multi-layered framework by identifying temporal patterns to improve detection accuracy with low-latency. The proposed approach focuses on the extraction of meaningful features from temporal time series data so that it will help us in enabling dynamic threat identification in multiple domains. From this work, the proposed system for anomaly detection in view of high-speed data, an adaptive threshold mechanism will be considered to reduce the false positives rate by 18%, and a lightweight strategy to ensure capability for low-latency applications. The Swarm-based LSTM achieved accuracy of 98.7 and 96.5% F1 Score with a precision 95.3% demonstrating optimal scalability and efficiency for real-time cybersecurity applications when compared with the vanilla LSTM, GRU, and Bi-LSTM. All these models were evaluated based on the data set KDDcup99.