Investigating Automated Hyper-Parameter Optimization for a Generalized Path Loss Model

This work aims at developing a generalized and optimized path loss model that considers rural, suburban, urban, and urban high rise environments over different frequencies, for use in the Heterogenous Ultra Dense Networks in 5G. Five different machine learning algorithms were tested on four combined datasets, with a sum of 12369 samples in which their hyper-parameters were automatically optimized using Bayesian optimization, HyperBand and Asynchronous Successive Halving (ASHA). For the Bayesian optimization, three surrogate models (the Gaussian Process, Tree Structured Parzen Estimator and Random Forest) were considered. To the best of our knowledge, few works have been found on automatic hyper-parameter optimization for path loss prediction and none of the works used the aforementioned optimization techniques. Differentiation among the various environments was achieved by the assignment of the clutter height values based on International Telecommunication Union Recommendation (ITU-R) P.452-16. We also included the elevation of the transmitting antenna position as a feature so as to capture its effect on path loss. The best machine learning model observed is K Nearest Neighbor (KNN), achieving mean Coefficient of Determination (R2), average Mean Absolute Error (MAE) and mean Root Mean Squared Error (RMSE) values of 0.7713, 4.8860dB, and 6.8944dB, respectively, obtained from 100 different samplings of train set and test set. Results show that machine learning can also be used to develop path loss models that are valid for a certain range of distances, frequencies, antenna heights, and environment types. HyperBand produced hyper-parameter configurations with the highest accuracy in most of the algorithms.

Testing a 5G Communication System: Kriging-Aided O2I Path Loss Modeling Based on 3.5 GHz Measurement Analysis

Deep knowledge of how radio waves behave in a practical wireless channel is required for the effective planning and deployment of radio access networks in outdoor-to-indoor (O2I) environments. Using more than 400 non-line-of-sight (NLOS) radio measurements at 3.5 GHz, this study analyzes and validates a novel O2I measurement-based path loss prediction narrowband model that characterizes and estimates shadowing through Kriging techniques. The prediction results of the developed model are compared with those of the most traditional assumption of slow fading as a random variable: COST231, WINNER+, ITU-R, 3GPP urban microcell O2I models and field measured data. The results showed and guaranteed that the predicted path loss accuracy, expressed in terms of the mean error, standard deviation and root mean square error (RMSE) was significantly better with the proposed model; it considerably decreased the average error for both scenarios under evaluation.

Effect of Membership Functions and Data Size on the Performance of ANFIS-Based Model for Predicting Path Losses in the VHF and UHF Bands

This paper investigates the effect of number and shape of membership function (MF), and training data size on the performance of ANFIS model for predicting path losses in the VHF and UHF bands in built-up environments. Path loss propagation measurements were conducted in four cities of Nigeria over the cellular and broadcasting frequencies. A total of 17 broadcast transmission and cellular base stations were utilized for this study. From the results obtained, it can be concluded for the broadcasting bands that the generalized bell MF shows better performance with an average RMSE of 3.00 dB across all the routes, followed by gaussian, Pi, trapezoid and triangular MFs in that other with average RMSE values of 3.09 dB, 3.11 dB, 3.16 dB and 3.23 dB respectively. For the cellular systems, Triangular MF outperformed other MFs with the lowest average RMSE. The generalized bell MF was found to be suited for WCDMA band while triangular MF is most suited for GSM band. Furthermore, it can also be concluded that the higher the number of membership functions, the lower the RMSE, whereas, a decrease in the data size leads to a reduction in the RMSE values. The findings of this study would help researchers and network planners to make a more informed decision on choosing appropriate system parameters when modeling ANFIS models for path loss prediction.

Path Loss Characterization Using Machine Learning Models for GS-to-UAV-Enabled Communication in Smart Farming Scenarios

The purpose of this paper was to predict the path loss characterization of the ground-to-air (G2A) communication channel between the ground sensor (GS) and unmanned aerial vehicle (UAV) using machine learning (ML) models in smart farming (SF) scenarios. Two ML algorithms such as support vector regression (SVR) and artificial neural network (ANN) were studied to analyze the measured data in different scenarios with Napier and Ruzi grass farms as the measurement locations. The proposed empirical GS-to-UAV two-ray (GUT-R) model and the ML models were compared to characterize path loss prediction models. The performances of the path loss prediction models were evaluated using the statistical error indicators in different measurement locations and UAV trajectories. To obtain the statistical error indicators, the accuracy path loss results of UAV trajectory at 2 m altitudes showed the SVR model (MAE = 1.252 dB, RMSE = 3.067 dB, and R2 = 0.972) and the ANN model (MAE = 1.150 dB, RMSE = 2.502 dB, and R2 = 0.981) for the Napier scenario. In the Ruzi scenario, the SVR model (MAE = 1.202 dB, RMSE = 2.962 dB, and R2 = 0.965) and the ANN model (MAE = 1.146 dB, RMSE = 2.507 dB, and R2 = 0.983) were presented. For UAV trajectory at 5 m altitudes, the SVR model (MAE = 2.125 dB, RMSE = 4.782 dB, and R2 = 0.933) and the ANN model (MAE = 2.025 dB, RMSE = 4.439 dB, and R2 = 0.950) were resulted in the Napier scenario. In the Ruzi scenario, the SVR model (MAE = 2.112 dB, RMSE = 4.682 dB, and R2 = 0.935) and the ANN model (MAE = 2.016 dB, RMSE = 4.407 dB, and R2 = 0.954) were displayed. The proposed ML models using SVR and ANN can optimally predict the path loss characterization in SF scenarios, where the accuracy was 95% for the SVR and 97% for the ANN.

A Deep Neural Network-Based Multi-Frequency Path Loss Prediction Model from 0.8 GHz to 70 GHz

Large-scale fading models play an important role in estimating radio coverage, optimizing base station deployments and characterizing the radio environment to quantify the performance of wireless networks. In recent times, multi-frequency path loss models are attracting much interest due to their expected support for both sub-6 GHz and higher frequency bands in future wireless networks. Traditionally, linear multi-frequency path loss models like the ABG model have been considered, however such models lack accuracy. The path loss model based on a deep learning approach is an alternative method to traditional linear path loss models to overcome the time-consuming path loss parameters predictions based on the large dataset at new frequencies and new scenarios. In this paper, we proposed a feed-forward deep neural network (DNN) model to predict path loss of 13 different frequencies from 0.8 GHz to 70 GHz simultaneously in an urban and suburban environment in a non-line-of-sight (NLOS) scenario. We investigated a broad range of possible values for hyperparameters to search for the best set of ones to obtain the optimal architecture of the proposed DNN model. The results show that the proposed DNN-based path loss model improved mean square error (MSE) by about 6 dB and achieved higher prediction accuracy R2 compared to the multi-frequency ABG path loss model. The paper applies the XGBoost algorithm to evaluate the importance of the features for the proposed model and the related impact on the path loss prediction. In addition, the effect of hyperparameters, including activation function, number of hidden neurons in each layer, optimization algorithm, regularization factor, batch size, learning rate, and momentum, on the performance of the proposed model in terms of prediction error and prediction accuracy are also investigated.

Analisis Prediksi Path Loss Teknologi Seluler 5G Pada Sel Micro Urban Wilayah Kota Bandung

ABSTRAKImplementasi teknologi seluler 5G di Indonesia perlu dilakukan kajian dalam beberapa aspek. Analisis nilai path loss pada sistem seluler merupakan pendekatan dalam aspek large scale fading untuk menghitung cakupan layanan. Penelitian ini melakukan kajian nilai path loss dengan mengambil kondisi di Kota Bandung dengan karakter sel urban mikro outdoor. Model prediksi yang digunakan pada kajian ini meliputi model SUI, ABG, CI, dan NYUSIM simulator menggunakan frekuensi kerja 3,5 GHz dan 28 GHz dengan lebar pita 100 MHz dan 800 MHz. Hasil pengujian memperlihatkan simulator NYUSIM memberikan nilai prediksi path loss yang paling mendekati nilai rata-rata path loss dengan nilai margin sebesar 1,25 dB untuk frekuensi 3,5 GHz dan 1,8 dB untuk frekuensi 28 GHz. Frekuensi kerja 28 GHz memiliki nilai path loss lebih tinggi dibandingkan frekuensi 3,5 MHz sebesar 19 dB. Nilai path loss pada kondisi LOS dan NLOS berdampak pada penurunan nilai path loss sebesar 35% pada frekuensi 3,5 GHz dan 26% pada frekuensi 28 GHz.Kata kunci: path loss, micro cell, urban, NYUSIM, LOS, NLOS ABSTRACTThe implementation of 5G cellular technology in Indonesia needs to be studied in several aspects. Analysis of the path loss value on the cellular system is an approach in the aspect of large scale fading to calculate service coverage. This research studies the path loss value by taking conditions in the city of Bandung with the character of outdoor micro urban cells. The prediction models used in this study include the SUI, ABG, CI, and NYUSIM simulators using working frequencies of 3.5 GHz and 28 GHz with a bandwidth of 100 MHz and 800 MHz. The test results show that the NYUSIM simulator provides a path loss prediction value that is closest to the average path loss value with a margin value of 1.25 dB for the 3.5 GHz frequency and 1.8 dB for the 28 GHz frequency. The 28 GHz working frequency has a path loss value higher than the 3.5 MHz frequency of 19 dB. The path loss value in LOS and NLOS conditions has an impact on reducing the path loss value by 35% at a frequency of 3.5 GHz and 26% at a frequency of 28 GHz.Keywords: path loss, micro cell, urban, NYUSIM, LOS, NLOS