The Internet of Things (IoT) is one of the tainted technologies through the interconnection of smart devices, resulting in serious security threats. Intrusion detection system (IDS) based on deep learning is very significant in order to discover and avert cyberattacks on IoT devices. There are many issues related to traditional intrusion detection systems that make them ineffective in case of connected devices, thus forcing us to shift our attention toward deep learning-based methods. The purpose of the present paper is to thoroughly discuss IoT security using deep learning, with a focus on various algorithms, datasets, kinds of attacks, and evaluation metrics. Empirical findings demonstrate how effectively deep learning helps in identifying a variety of cyber threats related to IOT. The main aim of this study is to help the researchers in future directions.