Optimized location prediction handover algorithm for Long Term Evolution Advanced (LTE-A) network

Abstract

The demand for seamless internet connectivity driving attempts to provide broadband mobile wireless communication even in a fast moving vehicle. One of the solutions to overcome the growth of connected wireless mobile devices is the deployment of small cells in dense heterogeneity network. In an attempt to provide quality of service (QoS) for vehicular users, the handover is one of the essential elements in the wireless network. However, the uncontrolled deployment of the access point (AP) in a small cell network that increases rapidly especially in the urban areas challenges handover management among vehicles. Furthermore, the vehicle only has limited time to spend when it passes through overlapping regions in small cell size. It may cause the incident of frequent and unnecessary handover due to the vehicle’s movement and cause wasted resources and overhead signaling. The main objective of this thesis is to develop an efficient handover algorithm that can allocate the appropriate amount of handover resources within the shortest time. The work proposed handover prediction algorithm that ensures to provide high QoS and reserved the handover resources in advanced. Vehicular location prediction (VLP) using Markov chain is developed to predict the user’s movement based on real user data traces. Vehicular location prediction handover algorithm (VLP-HA) is developed based on the prediction result from VLP. While optimization vehicular location prediction handover algorithm (OVLP-HA) is an enhancement of VLP-HA with decision strategy based on optimal forwarding (OF) weight. The performance is evaluated in terms of the rate of prediction accuracy for VLP. While handover performance for VLPHA and OVLP-HA is evaluated based on the number of ping-pong effect and data throughput. The result for prediction accuracy shows that VLP has notably improved the accuracy rate by 32% and 5% compared to human behavior-based prediction technique (HBP) and location prediction using Kalman filter (LPKF) r espectively. Then, the prediction from VLP is used in VLP-HA. The simulation is done within three level density traffic for reflecting r eal s cenarios which a re an urban and r ural a rea. The result shows further improvement for VLP-HA which is no ping-pong effect when VLP-HA is applied compared to A2A4 handover algorithm (A2A4-HA) and human behavior-based prediction handover algorithm (HBP-HA). In order to find the optimal handover point so that VLP-HA could provide higher QoS and at the same time reduced the ping-pong effect, the optimized VLP-HA by using OF that is developed based on ant colony optimization (ACO) algorithm. Two parameters considering the packet delivery ratio (PDR) and the number of unnecessary handover are determined. The best OF value is applied in the OVLP-HA. It is found that the handover performance for OVLP-HA has 7% improved data throughput and 33% less ping-pong effect compared to A2A4-HA and HBP-HA. The proposed handover algorithm has significantly enhanced the handover performance through the number of ping-pong effect, data throughput and optimized resource allocation. The proposed handover algorithm is adaptable to variation of AP’s level density and can be used in any network area such as urban area or rural area.