Reinforcement Based Learning Approach for Handoff in LTE Mobile Communication Technology

Abstract

Long Term Evolution is getting popular now a days as number of mobile user are increasing at very fast rate. Not only in terms of data connectivity, Long Term Evolution is getting popularity for voice transmission also using Voice over Long Term Evolution protocol which is an extension to Long Term Evolution protocol. As the number of users are increasing day by day along with their dependence on communication system, quality of Service is becoming an important factor. User equipment is not always stationary and moves while communication is going on and during this process user equipment moves from one eNodeB to other. Handover takes place when device changes eNodeB. Handover is an important parameter which effects the performance of communication system. In this paper, handover protocol in Long Term Evolution has been modified by implementing tabular temporal difference Learning concept, a reinforcement learning algorithm which guides network in selecting next eNodeB to which User Equipment will get connected when signal strength of current eNodeB reduces and there are more then one option available for handover by first calculating the reward R and then taking decision as per signal strength and recommendations of tabular Temporal Distancing based reinforcement algorithm. Performance of this modified handover mechanism has been evaluated for parameters: Forced Termination Probability, Outage Probability and Blocking Probability. It is found that performance in terms of above parameters improve considerably when tabular temporal distancing based algorithm is implemented in handover process of Long Term Evolution. When traffic under any eNodeB is increases, impact of modified algorithm further increases.

Keywords: LTE; Handover; Reinforcement Learning; eNodeB; Mobility