status

Completed

year

2026

type

Literature review

Abstract

This literature review examines artificial intelligence-based techniques for optimizing horizontal handover in mobile and wireless networks. The study begins by explaining the role of handover in maintaining service continuity when a mobile user moves between base stations. It focuses specifically on horizontal handover, where the connection is transferred between base stations that use the same wireless technology, such as 4G-to-4G, 5G-to-5G, or Wi-Fi-to-Wi-Fi. Traditional handover approaches often rely mainly on signal strength indicators such as RSSI, RSRP, or RSRQ, where a device may switch to a neighboring base station when its signal becomes stronger than the current one. Although this method is simple and widely used, the review explains that signal strength alone is not enough for reliable decision-making because it can fluctuate due to movement, interference, buildings, obstacles, cell-border conditions, and temporary environmental changes. These limitations can lead to ping-pong handover, handover failure, interruption delay, packet loss, unnecessary handovers, and unstable quality of service, especially for real-time applications such as video calls, streaming, online gaming, and high-mobility scenarios. The paper reviews selected studies that apply AI methods to make handover decisions more adaptive and context-aware. Machine learning methods are discussed as practical approaches that can use multiple input parameters, including current base station RSSI, target base station RSSI, user speed, network load, delay, packet loss, and signal history. These models can support proactive handover decision-making and self-optimization by learning from network conditions rather than depending only on fixed signal-strength thresholds. Reinforcement learning studies are also reviewed because handover can be treated as a decision-making problem, where the system receives rewards for good handover behavior and penalties for harmful outcomes such as ping-pong events or handover failures. The review also covers LSTM and prediction-based approaches, which are useful because handover behavior is time-dependent and previous signal or mobility values can help predict future connection quality, especially in high-speed movement scenarios such as railway networks. The selected papers are compared according to method, focus, simulation or evaluation environment, key performance indicators, and limitations. Common metrics include ping-pong rate, handover failure rate, handover delay, packet loss, throughput, radio link failure, load balancing, and unnecessary handovers. The review shows that many studies are simulation-based, which supports the choice of a Python-based simulation as a realistic direction for a semester project. Based on the literature, the paper proposes a feasible direction that compares a traditional RSSI-based baseline with a simple AI-based method such as Decision Tree or Random Forest. The proposed simulation can model user movement, signal changes, network load, delay, and packet loss, then evaluate both methods using shared KPIs. The study concludes that AI-based handover optimization is a suitable and realistic research direction because it allows handover decisions to consider more context than RSSI alone, while still remaining implementable through a clear Python-based simulation plan.