More and more devices are exchanging data via the Internet of Things (IoT), which increases the quality of IoT facilities. There are various vulnerabilities, security weaknesses, and attack vectors in IoT systems that represent a long-term security issue. To realise the full potential of IoT applications, billions of linked devices must be secured. Information on IoT is vulnerable to threats, assaults, and flaws. To solve IoT-related security, privacy, and vulnerability challenges, a strong security solution is required. Many scientists have previously suggested solutions for security in IoT. Deep Learning is one of the most promising techniques for protecting IoT systems in recent years, and Reinforcement Learning is gaining popularity in this regard. Unlike other deep learning approaches, reinforcement learning may learn the environment with a little amount of data about the parameters to be learnt. Deep reinforcement learning (DRL) methods may be capable of dealing with the aforementioned challenges related with IoT devices in the near future. In this paper, the Deep Reinforcement Learning (DRL) approach is introduced to increase IoT security. This study presents quantum computing for feature selection, and to securely transmit data, the Ciphertext-Policy Attribute-Based Encryption (CP-ABE) method has been built for an IoT system, and its performance is compared with traditional approaches. The suggested solution allowed for safe and scalable data transmission. The accuracy obtained by the proposed system is higher when compared to the other approaches.