Wireless Mesh Networks (WMN) plays a vital role in modern day communication. It transfers the data to every node through hop by hop mechanism. It can act as a client as well as the server. The data transmitted to every node from source to destination and hence the next node selection plays a challenging task in WMN. The previous state of art models shows that the node selection is done with lot of time waste or a wrong node is selected. Hence, we propose a novel methodology in which enhanced bubble sorting is used to identify the next node and reinforcement learning is applied every time for the verification of exactness of the node. Traditional sorting algorithms are simple to implement but they lack in efficiency for handling a large number of mesh nodes. The enhanced bubble sort reduces the number of unnecessary comparisons based on the reward and loss of the reinforcement learning and hence choose the best node.