An Improving Intrusion Detection Model Based on Novel CNN Technique Using Recent CIC-IDS Datasets

The Internet of Things (IoT) concept past far utilized in the past years, resulting in the establishment of Smart Cities. IoT has been A key player in a variety of sectors, including healthcare, Industry 4.0, smart transportation, home automation. Smart Cities are a relatively new concept that efficiently uses adaptive optimization of the resources at hand to offer people outstanding facilities. IoT devices are designed to gather information from the privilege of surroundings and send it to other systems over the Internet. Cybersecurity risks including denial-of-service attacks, brute-force assaults, and unauthorized access might result from this. The safety and privacy of residents are the most pressing concerns that require attention. Deep learning-based IDS system. Data gathering from common datasets like CIC-IDS 2017 and CIC-IDS 2018 comprises multiple assault kinds. By using algorithms, pre-processing removes normalizations and missing values. Time stamps are selected and largely removed from an assault dataset using Random Forest (RF). Subsequently, essential traits are retrieved from those data using Deep Autoencoder (AE), after which among those many features. In this paper an appropriate comparison of the initial two datasets produced good findings for every deep learning system put to the test. Measurements in CIC IDS 2017 Precision, F1-score, and Recall were 99.5%, 98.7%, and 99.8%, respectively. For CIC IDS 2018, Precision, F1-score, and Recall were 99.5% for all algorithms. Our Novel CNN method yields the greatest outcomes.