Minimization Of Energy Consumption For Mobile Edge Computing-Based Augmented Reality Applications

Recently,mobile Augmented Reality(AR)is becoming the most imperative topic.AR exploits computer graphics and builds a real world in response of user's input by combining the real world with the computer generated virtual image.With the advancement in mobile application user equipment such as mobile phones,laptops,tablets have become very popular and handy.But due to intensive computations and high battery drainage features of AR applications,their leverage advantage could not be taken.Mobile cloud computing concept can overcome both the problems of computation and battery drainage but it introduces delay because information can be sent from servers which are powerful in the form of network topology and by the far end users.Moreover,it doesn't provide high-quality transmission and can't be involved in real-time applications.AR application are time sensitive and require high computation power and battery capacity.Instead of using the whole application on a mobile,AR application can be offloaded on cloudlet server and to remove the delay problem for AR applications,the new concept was introduced that is Mobile Edge Computing(MEC).In this scenario,cloudlet was moved nearest to the user equipment that is at a mobile network edge.It deploys storage capacity among the network access of radio and provides direct traffic among end user providing high quality of service for AR applications.System failure can be avoided by MEC deploying in base station.The work in this thesis is summarized as follows:Firstly,AR applications are introduced and some background of AR applications,motivation,problem statement and our contribution are discussed.Machine learning,its types and Q learning is discussed briefly.The challenges faced by delay sensitive immersive AR applications,types of AR application and techniques used for these applications are explained in detail.Secondly shared resource allocation strategy is designed for MEC based collaborative AR applications by applying machine learning algorithm.We have used collaborative property of AR applications to offload some part of the application on MEC server.We have designed a Q learning algorithm to achieve our goal for collaborative offloading.Numerical results are presented to show the effective of the proposed algorithm.