Optimized Q-Learning-Based Handover Decision Algorithm for Femtocells Using Load Balancing in LTE-A Networks

Optimized Q-Learning-Based Handover Decision Algorithm for Femtocells Using Load Balancing in LTE-A Networks

lte-a; q-learning; load balancing; macrocell; femtocell; cbr; voip

The rapid growth of mobile devices and demand for mobile data have made maintaining capacity, high coverage, and data speed challenging. With the emergence of small cell networks, the Long-Term Evolution (LTE) system helped to address these issues, Femtocell technology is being deployed to provide improved indoor coverage. However, a major challenge is the frequent handover and unequal distribution of cell loads, which lead to a reduction in call and data rates. Small cells have changing and unplanned load distribution over time, resulting in certain cells suffering high user density and strong resource competition, while others have low user density and wasteful resources due to low consumption. This imbalance in cell load distribution greatly influences overall network performance and prevents Femtocells from realizing their full potential. Despite several efforts by researchers to enhance network communication, handover is still a challenging issue, many related works have been done in the field but still it needs improvement. This research proposes an Optimized Q-learning-based Handover Decision Algorithm for Femtocells using Load Balancing in LTE-A Networks to improve overall network performance. The algorithm learns to prioritize and select cells with low load during target cell selection and not only provides good Quality of Service (QoS) but also has a low load, resulting in better traffic distribution across the cells. Several simulations were performed using LTE-Sim. Results proved the outperformance of the proposed algorithm over the existing algorithm in terms of QoS with a packet loss ratio for CBR packet transmission of 512 bytes with a rate of 8 packets/second intervals, 88.53%, and VoIP packet transmission of 32 bytes per 20 ms/time interval, 89.24% respectively.

Babangida Abubakar Albaba1,2, Aliyu Saidu1,2, Muhammad Sani Iliyasu, Abdulhamid Usman Nuruddeen, Kamal Sabo wada

www.ijcit.com

September 2024

peri irawan

pdf

english

text