<p>Intrusion detection is the far most important thing in wireless sensor networks. In recent times Internet of Things (IoT) is gaining wide notice with the incorporation of the advanced form of wireless devices which can connect multiple hardware devices for different operations. Most of the existing IoT systems are the composition of many sensor nodes which are in wireless mode. Security concerns in terms of data security, secure communication, intrusion detection, authorization and authentication are the need of the hour. The main objective of this paper is to detect the intrusion based on the basis behavioural analysis. In the proposed work, the Mario game approach has been identified to classify the types of attack and provide a prevention methodology based on behavioural analysis. The attacks taken into consideration are the data mining-based attacks, tracking attacks, cyber espionage and eavesdropping attacks. Based on the fast attack detection and avoidance moves available in the Mario game, the real-time intrusions are detected and prevented. Once the attacks are identified, classification is performed using C4.5 (J48) to categorize the attacks which helps to recover the node from attack or avoid the node in future attack path formation. The proposed Mario game-based solution has provided an overall detection accuracy of 98 and accuracy obtained for preventing future attacks are 98.2% with less time consumption of 10% lesser than existing systems. In all the cases, the Mario based intrusion detection and prevention mechanism has improved the accuracy mechanism in intrusion detection and prevention.</p>