Adaptive Task Scheduling Procedure By Using Improvised Q-Learning Technique In Internet Of Things

The arrival of the data packets fluctuates dynamically in an Internet of Things(IoT) ecosystem which consists of numerous devices. fault to operate them according to the QoS (Quality of Service) criterion could have a significant impact on the system's reliability and quality. As a result, in order to pay attention to the shifting traffic condition, data gathering gateway must quickly design a new scheduling policy. Traditional packet scheduling algorithms are inadequate in IoT as the communication of data arrangement is unidentified at earlier. Q-learning is a sort of reinforcement learning specially allows you to create a dynamic scheduling policy based on the status of individual queue beyond having any prior knowledge of the network's state. In IoT analytics, deep learning algorithms play a critical role. Machine data is sparse and contains a temporal component. The data from a certain devices, may react differently under other circumstances. As a result, it's impossible to capture all eventualities for the data preprocessing or training stage of an algorithm. Continuously monitoring sensor data is also inconvenient and costly. In this paper, Existing Q-Learning algorithm has been considered and by performing indexing in order to obtain structure similar matrix then data synthesis process is done through which proposed Q-Learning algorithm will be enhanced. The paper also provides an comparison between existing and proposed enhanced QL algorithm.