A newly developed dataset is used for efficient feature selection and efficient Bot-IoT attack identification in an IoT network context. The dataset includes information about botnet attacks, regular traffic flows, the Internet of Things, and countless more cyber attacks. The realistic simulation is employed for the production of this dataset with effective information features in order to track the precise traffic. Similar to this, more features were extracted and integrated with the extracted features set to improve the performance of machine learning models and effective prediction models. The research effort suggested employing deep learning algorithms like LSTM and CNN for effective botnet detection in IoT networks. By conducting in-depth tests with the most pertinent publically accessible dataset (Bot-IoT) in binary and multi-class classification scenarios, the efficacy of this strategy was confirmed. Because of its consistent updates, extensive attack diversity, and variety of network protocols, the Bot IoT was utilised as a dataset in this situation. We use the Bot-IoT dataset to assess our suggested strategy. Analysis of the simulation results revealed that, in comparison to the current BLSTM, our proposed method is effective and can, on average, produce better performance results