Hybrid Levenberg–Marquardt and LSBoosting Ensemble Algorithms for Optimal Signal Attenuation Modeling and Coverage Analysis

ABSTRACT This paper proposes and engages the Levenberg–Marquardt algorithm method via regression to optimally model and predict real‐time signal strength values acquired via telecom software investigation tools in LTE cellular networks. To further improve the Levenberg–Marquardt method, which is som...

Full description

Saved in:
Bibliographic Details
Published in:International journal of communication systems Vol. 38; no. 8
Main Authors: Isabona, Joseph, Imoize, Agbotiname Lucky, Lee, Cheng‐Chi
Format: Journal Article
Language:English
Published: Chichester Wiley Subscription Services, Inc 25.05.2025
Subjects:
Algorithms
cell coverage area
Cellular communication
fringe cell coverage
Levenberg–Marquardt algorithm
LSBoost ensemble
Optimization
path loss
Prediction models
Regression models
Signal analysis
signal path loss model
Signal strength
Software
Wireless networks
ISSN:1074-5351, 1099-1131
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:ABSTRACT This paper proposes and engages the Levenberg–Marquardt algorithm method via regression to optimally model and predict real‐time signal strength values acquired via telecom software investigation tools in LTE cellular networks. To further improve the Levenberg–Marquardt method, which is sometimes prone to parameter evaporation on high dimensional data with high bias or variance issues during the application, we explore the Least Square Boosting (LSBoosting) ensemble algorithms. The combined signal predictive modeling procedure is termed the hybrid LSBoost‐LM method. First, when the proposed hybrid LSBoost‐LM method was engaged for real‐time extrapolative signal analysis, the results displayed excellent root mean square error precision accuracies compared with two other standards, Bag‐LM and LM methods. As a case in point, the LSBoost‐LM method achieved 2.15, 3.44, 3.33, 1.31, and 2.19 dB RMSE values at different prediction study locations, which are relatively lower compared with the Bag‐LM and standard LM methods that achieved higher RMSE values of 3.38, 3.86, 4.07, 2.28, 3.98 dB and 5.57, 5.52, 5.14, 3.67, 4.56 dB, respectively. Secondly, applying the hybridized model produced up to 93.39% cell area coverage quality and 89.18% fringe cell area coverage quality across the eNodeB study locations. The proposed method can assist practicing RF network planners in realistic cell coverage quality estimation and analysis of related wireless networks.
Bibliography:The authors received no specific funding for this work.
Funding
ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1074-5351
1099-1131
DOI:10.1002/dac.70096