Over the past years, expeditious evolution in communication and networks became ambiguous, heterogeneous to handle them in aspects like organizing and optimization of networks, so entail to introduce intelligence to networks where traditional networks are unable to withstand so introduced Software Defined Networks. Architecture is designed in such a way that machine learning techniques are able to deploy to control networks. In this paper on of the crucial task, congestion control in network that regulates transmission of data over the network and effective utilization of bandwidth or capacity of network is taken to consideration and opted Q learning and elucidate how it assists to perceive optimal path with aid of reward function and Q table maintains congestion less path from a source to desired node. Simulation results computed in network simulator 3 are results which are explained by considering three approaches Standard Q Learning and QCAR and proposed QLCCRP and compared factor of learning rate to end to end delay and packet delivery ratio. Proposed approach showed minimalistic improvement as compared to QCAR and drastic improvement to QL standard approach and many other factors yet to be compute.