Enhancing Cybersecurity in Modern Networks: A Low-Complexity NIDS Framework using Lightweight SRNN Model Tuned with Coot and Lion Swarm Algorithms

The tremendous increase in internet-based traffic has put modern networks at risk from numerous threats. The effective functioning of network infrastructure is impeded by invasive traffic, which uses up resources and time. Increased output is the result of a well-planned approach to avoiding, identifying, and dealing with intrusion situations. When it comes to cybersecurity, intrusion detection systems (IDS) are important for keeping networks safe from all sorts of malicious cyberattacks. Finding highly adaptive, small-footprint Network Intrusion Detection Systems (NIDS) that can detect anomalies is the urgent need of this research. The research makes use of the comprehensive NSL-KDD dataset, which includes 43 variables labelled “attack” and “level.” It suggests a new method for intrusion detection that combines Deep Learning models based on feature selection. Coot Optimiser efficiently chooses the most important characteristics, while the Sliced Recurrent Neural Network (SRNN) model detects threats. Lion Swarm Optimisation Algorithm (LSOA) reduces computing complexity by properly tuning the SRNN’s fine-tuning parameters. In addition, the proposed intrusion detection approach may be more easily evaluated with this dataset. In addition, the main objective of this effort is to expand detection accuracy while retaining operating efficiency. A typical NIDS also uses a number of methods to look at both normal and dangerous behaviour. Our proposed strategy is shown to improve intrusion detection precision, which is a significant step forward in this area, according to the results. To combat increasing cyberthreats and adapt to evolving network conditions, our next steps will centre on fortifying and broadening our strategy.