Development of Artificial Neural Network Based Model for Prediction of Radio Pathloss of a 4G-LTE-A Network in Ogbomoso, Nigeria
| dc.contributor.author | Aremu, O. A., Oyinkanola, L. O. A., Makinde, O. S., Fajemiroye, Joseph Ademola, Ajani, Sunday Adegbenro, Alade, Michel Olusope | |
| dc.date.accessioned | 2024-11-08T12:16:50Z | |
| dc.date.available | 2024-11-08T12:16:50Z | |
| dc.date.issued | 2022 | |
| dc.description.abstract | Pathloss propagation models are essential for optimal planning and deployment of wireless communication networks. However, the inherent complexity of radio signal propagation makes it difficult to predict received losses in different environments. Yet, there is a new paradigm shift towards the application of computational methods such as artificial neural networks (ANN) modeling to predict radio pathloss. In this study, a new ANN model was developed using measured pathloss data at four base stations in Ogbomoso south western Nigeria (Latitude 8°1227″N, Longitude 4°2436″E). Data collection, which spanned a period of twelve months took place in a free field at 900 MHz. The developed model was validated using independent pathloss data in different itineraries, distances and elevations and the results obtained were compared to existing empirical pathloss models such as: Cost-231, Okumura, Lee, and CCIR. The results obtained revealed that all the empirical pathloss models considered overestimated the measured pathloss with root mean square (RMSE) of 51.18, 44.22, 36.45 and 8.49 dB respectively. The developed ANN model validated using standard procedure had MAE, G, MBE and MAPE of 5.43 dB, 84.92%, 0.32 dB, and 2.28%, respectively, which is an indication of good model performance. Also, the developed ANN model performed better and was found to be optimal in terms of prediction accuracy with RMSE 5.59 dB at an R-value of 0.94, thus, out-peering existing empirical pathloss model. Based on these findings, the proposed ANN model can be deployed in all categories since the mean RMSE is within the acceptable standard value. | |
| dc.description.sponsorship | Self - Sponsored | |
| dc.identifier.issn | 2522-9400 | |
| dc.identifier.uri | https://kwasuspace.kwasu.edu.ng/handle/123456789/2708 | |
| dc.language.iso | en_US | |
| dc.publisher | European Modern Studies Journal | |
| dc.title | Development of Artificial Neural Network Based Model for Prediction of Radio Pathloss of a 4G-LTE-A Network in Ogbomoso, Nigeria | |
| dc.type | Article |