فاطمه رئيسي امجد

The convergence of 5G communication networks an‎d non-terrestrial networks (NTNs), particularly leveraging Low Earth Orbit (LEO) satellites, is a key strategy for achieving global, ubiquitous, an‎d high-quality connectivity. However, the extremely high mobility of LEO satellites relative to ground terminals introduces a significant challenge: frequent han‎dovers. In satellite networks, unlike terrestrial ones, this process is primarily driven by the satellite’s rapid movement, an‎d its excessive repetition leads to increased signaling overhead, degraded Quality of Service (QoS), an‎d a poor user experience. This challenge is further aggravated by the simultaneous competition of multiple users vying for connection to satellites with the best signal conditions, leading to traffic bottlenecks an‎d increased blocking rates. Traditional han‎dover methods, which rely on local criteria (like signal strength), lose their effectiveness in this dynamic environment due to high propagation delays an‎d a lack of a comprehensive network view. To overcome these challenges, this research proposes an intelligent an‎d proactive framework for han‎dover management. The han‎dover problem is modeled as a Markov Decision Process (MDP), wherein the user, acting as an intelligent agent, learns an optimal policy for target satellite selec‎tion by interacting with the network environment an‎d leveraging a Deep Reinforcement Learning (DRL) algorithm. A key innovation of this research is the design of a multi-objective reward function. By penalizing unnecessary han‎dovers an‎d rewarding optimal selec‎tions, this function guides the agent to make decisions that simultaneously achieve three key objectives: minimizing the number of han‎dovers, balancing network traffic load, an‎d maintaining Quality of Service. Simulation results demonstrate that the proposed approach not only significantly reduces han‎dover frequency but also, through its ability to predict signal degradation an‎d make proactive decisions, considerably curbs the connection blocking rate. This effectively ensures the stability an‎d efficiency of the entire system in next-generation networks.

ﺷﺒﮑﻪ ﻫﺎي ﻣﺎﻫﻮاره اي ارﺗﻔﺎع ﭘﺎﯾﯿﻦ , ﺷﺒﮑﻪ ﻧﺴﻞ ﭘﻨﺠﻢ , واﮔﺬاري , ﯾﺎدﮔﯿﺮي ﺗﻘﻮﯾﺘﯽ

Low Earth Orbit Satellite Networks , 5G Network , Handover , Reinforcement Learning

Fateme Raisi amjad

Dr.nasser mozayani - Dr.Ahmad Akbari Azirani