Scholarlink

<p>Efficient 5G Radio Access Network (RAN) deployment requires mobile network operators to identify high-impact upgrade sites while adhering to budgetary and service quality constraints. We propose SSal, a machine learning-based system that prioritizes 5G upgrade candidates by analyzing the combined influence of network metrics, user behavior, and handset adoption on future 5G demand. Using supervised learning, SSal forecasts user migration from 4G to 5G and evaluates the resulting performance implications on 4G networks. Cells are first grouped by clustering to stabilize fits and improve prediction accuracy for per-cluster regressors. This is followed by a second-level clustering to rank groups using attributes. This attributes include bandwidth usage, device support, and key performance indicators (KPIs), including downlink throughput, traffic load, and PRB utilization. Cells from higher ranked clusters undergo regression analysis to estimate throughput improvements under reduced PRB utilization. The final ranking combines predicted throughput gain (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\Delta T\)</EquationSource> </InlineEquation>) and PRB relief (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\Delta \textrm{PRB}\)</EquationSource> </InlineEquation>) with operator-set weights (equal by default). This yields a prioritized upgrade list of 5G cell sites. Simulation results demonstrate that SSal effectively enhances 4G offloading, increases 5G traffic absorption, and improves site selection through scalable, data-driven decision-making.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SSal: A System Leveraging Machine Learning for Superior Site Selection of 5G Radio Access Network

  • 期刊论文

  • 发表日期：

    2026年2月24日

查看全文
  • Himadri Sikhar Khargharia,
  • Siddhartha Shakya

摘要

Efficient 5G Radio Access Network (RAN) deployment requires mobile network operators to identify high-impact upgrade sites while adhering to budgetary and service quality constraints. We propose SSal, a machine learning-based system that prioritizes 5G upgrade candidates by analyzing the combined influence of network metrics, user behavior, and handset adoption on future 5G demand. Using supervised learning, SSal forecasts user migration from 4G to 5G and evaluates the resulting performance implications on 4G networks. Cells are first grouped by clustering to stabilize fits and improve prediction accuracy for per-cluster regressors. This is followed by a second-level clustering to rank groups using attributes. This attributes include bandwidth usage, device support, and key performance indicators (KPIs), including downlink throughput, traffic load, and PRB utilization. Cells from higher ranked clusters undergo regression analysis to estimate throughput improvements under reduced PRB utilization. The final ranking combines predicted throughput gain ( \(\Delta T\) ) and PRB relief ( \(\Delta \textrm{PRB}\) ) with operator-set weights (equal by default). This yields a prioritized upgrade list of 5G cell sites. Simulation results demonstrate that SSal effectively enhances 4G offloading, increases 5G traffic absorption, and improves site selection through scalable, data-driven decision-making.