Seamless handover optimization for LTE-A communication using Kalman filtering and DQN

Shakir, Abdullah Talaat (2024) Seamless handover optimization for LTE-A communication using Kalman filtering and DQN. Doctoral thesis, Universiti Teknikal Malaysia Melaka.

	Text (24 Pages) Seamless handover optimization for LTE-A communication using Kalman filtering and DQN.pdf - Submitted Version Download (486kB)
	Text (Full Text) Seamless handover optimization for LTE-A communication using Kalman filtering and DQN.pdf - Submitted Version Restricted to Registered users only Download (6MB)

Abstract

In LTE-A networks, achieving optimal Hard Handover (HHO) for uninterrupted connectivity and Quality of Service (QoS) adherence poses a persistent challenge, particularly in the context of Intelligent Transportation Systems (ITS) scenarios. Existing handover mechanisms often exhibit deficiencies in localization improvement, integration of process models and measurement data, and determining an optimal Time to Trigger (TTT) based on past learning experiences. These limitations are compounded by deterministic rules that fail to account for the dynamic nature of mobility and the multifaceted factors influencing handover decisions. To mitigate these challenges, this thesis proposes a novel hybrid model integrating a Map Sampling-based Kalman Filter (MA-KALMAN) and a Deep Q-learning Network (DQN) for HHO decision-making. The MA-KALMAN component improves the accuracy of mobility data by merging GPS measurements with process models, while the DQN framework optimizes decisions by learning from dynamic network conditions. Comparative evaluations against traditional models, including the Kalman filter for localization, Q-learning, and static approaches for handover decision-making, were conducted, focusing on key performance metrics such as RMSE, End-to-End delay (E2Edelay), Packet Delivery Ratio (PDR), Number of Hard Handovers (No. of HHO), and Hard Handover Ping-Pong (HHO Ping-Pong) instances. Empirical findings substantiate the superior performance of the MA-KALMAN and DQN-based handover decision-making models, which minimize latency, enhance reliability, and determine an optimal TTT, ensuring seamless connectivity and QoS adherence in LTE-A networks. This research advances wireless communications by addressing critical issues in localization and TTT optimization.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	LTE-A networks, Intelligent Transportation Systems (ITS), Kalman filter
Divisions:	Library > Tesis > FTKEK
Depositing User:	Muhamad Hafeez Zainudin
Date Deposited:	27 Dec 2024 15:55
Last Modified:	27 Dec 2024 15:55
URI:	http://eprints.utem.edu.my/id/eprint/28324
Statistic Details:	View Download Statistic

Actions (login required)

View Item