Q-learning based adaptive QoS routing protocol for MANETs

Mobile ad-hoc network (MANET) communication is challenging as they are dynamic in nature. We propose a MANET routing protocol considering the QoS (Quality of Service) parameters such as bandwidth efficiency, link stability and power metric. The idea of learning the network status information to enhance network routing is the key concept of our project. This helps to learn about the expiry of links and choose the path with more expiration time and the minimum power consumption. The learning is carried out instantly by a typical route request/reply packets containing the network status information. This reinforcement learning is called Q-Learning algorithm. The enhanced version of route determining process leads to tremendous improvement in QoS. We apply this learning algorithm in the existing MAODV protocol which will further enhance the QoS in MAODV. The simulation results show that the proposed system performs better than the existing system in terms of improving QoS.