Enhancing the quality of human's daily life in respects of comfort is the chief objective of smart environments (SE). The Internet of Things (IoT) is basically an increasing network of smart objects. It commences diverse services in human's routine life relying on its available and also dependable activities. The chief problems in any real-globe SE centered upon the IoT model are the security in addition to privacy. The security susceptibility in IoT-centered systems creates security risks that affect SE applications. So, an Intrusion Detection System (IDS) based Modified Adaptive Neuro-Fuzzy Inference System is proposed aimed at detecting the attacks on IoT Smart Cities (SM). The proposed method comprises '2' phases. They are training and testing. First, the IDS are trained by performing three processes: preprocessing, feature selection (FS) and classification. For training, the proposed technique utilizes the data from the NSL_KDD dataset. Then, IoT sensor values are tested employing the same steps of training. The result of testing comprises '2' models. They are the attacked data and non-attacked data. The non-attacked data is sent to the user securely with the help of Improved Rivest Shamir Adleman method. After that, the user receives and decrypts the data. Then, the decrypted data is forecasted for further analysis. The proposed techniques' experimental outcomes used in FS, classification, and also secure data transmission are contrasted with the existent methods.
KeywordsAttack detection • Internet of Things (IoT) • Smart cities • Crow search optimization (CSO) • Chaotic mapping (CM) • CM based CSO (CM-CSO) • Adaptive neuro fuzzy inference system (ANFIS) • Modified ANFIS (MANFIS) • Intrusion detection system (IDS) • Rivest Shamir Adleman (RSA) encryption • Improved RSA (IRSA)