Macrocell Path Loss Prediction Using Artificial Neural Networks

Document type: Journal Articles
Article type: Original article
Peer reviewed: Yes
Full text:
Author(s): Erik Ostlin, Hans-Jürgen Zepernick, Hajime Suzuki
Title: Macrocell Path Loss Prediction Using Artificial Neural Networks
Journal: IEEE Transactions on Vehicular Technology
Year: 2010
Volume: 59
Issue: 6
Pagination: 2735-2747
ISSN: 0018-9545
Publisher: IEEE
URI/DOI: 10.1109/TVT.2010.2050502
ISI number: 000282025900010
Organization: Blekinge Institute of Technology
Department: School of Engineering - Dept. of Electrical Engineering (Sektionen för ingenjörsvetenskap - Avd. för elektroteknik)
School of Engineering S-371 79 Karlskrona
+46 455 38 50 00
Authors e-mail:
Language: English
Abstract: This paper presents and evaluates artificial neural network models used for macrocell path loss prediction. Measurement data obtained by utilising the IS-95 pilot signal from a commercial code division multiple access mobile network in rural Australia is used to train and evaluate the models. A simple neuron model and feed-forward networks with different number of hidden layers and neurons are evaluated regarding their training time, prediction accuracy, and generalisation properties. Also, different backpropagation training algorithms, such as gradient descent and LevenbergMarquardt, are evaluated. The artificial neural network inputs are chosen to be distance to base station, parameters easily obtained from terrain path profiles, land usage and vegetation type and density near the receiving antenna. The path loss prediction results obtained by using the artificial neural network models are evaluated against different versions of the semi-terrain based propagation model Recommendation ITU-R P.1546 and the OkumuraHata model. The statistical analysis shows that a non-complex artificial neural network model performs very well compared to traditional propagation models in regards to prediction accuracy, complexity and prediction time. The average ANN prediction results were: 1) maximum error: 22 dB, 2) mean error: 0 dB and 3) standard deviation: 7 dB. A multi-layered feed-forward network trained using the standard backpropagation algorithm was compared with a neuron model trained using the LevenbergMarquardt algorithm. It was found that the training time decreases from 150 , 000 to 10 iterations whilst the prediction accuracy is maintained.
Subject: Telecommunications\General
Signal Processing\General
Keywords: Okumura Hata model, Path loss prediction, Recommendation ITU-R P.1546, artificial neural network, backpropagation, feed-forward network, field strength measurements, point-to-area