Towards Physics-Based Generalizable Convolutional Neural Network Models for Indoor Propagation

A fundamental challenge for machine learning models for electromagnetics is their ability to predict output quantities of interest (such as fields and scattering parameters) in geometries that the model has not been trained for. Addressing this challenge is a key to fulfilling one of the most appealing promises of machine learning for computational electromagnetics: the rapid solution of problems of interest just by processing the geometry and the sources involved. The impact of such models that can "generalize" to new geometries is more profound for large-scale computations, such as those encountered in wireless propagation scenarios. We present generalizable models for indoor propagation that can predict received signal strengths within new geometries, beyond those of the training set of the model, for transmitters and receivers of multiple positions, and for new frequencies. We show that a convolutional neural network can "learn" the physics of indoor radiowave propagation from ray-tracing solutions of a small set of training geometries, so that it can eventually deal with substantially different geometries. We emphasize the role of exploiting physical insights in the training of the network, by defining input parameters and cost functions that assist the network to efficiently learn basic and complex propagation mechanisms.

Towards Physics-Based Generalizable Convolutional Neural Network Models for Indoor Propagation

A fundamental challenge for machine learning models for electromagnetics is their ability to predict output quantities of interest (such as fields and scattering parameters) in geometries that the model has not been trained for. Addressing this challenge is a key to fulfilling one of the most appealing promises of machine learning for computational electromagnetics: the rapid solution of problems of interest just by processing the geometry and the sources involved. The impact of such models that can "generalize" to new geometries is more profound for large-scale computations, such as those encountered in wireless propagation scenarios. We present generalizable models for indoor propagation that can predict received signal strengths within new geometries, beyond those of the training set of the model, for transmitters and receivers of multiple positions, and for new frequencies. We show that a convolutional neural network can "learn" the physics of indoor radiowave propagation from ray-tracing solutions of a small set of training geometries, so that it can eventually deal with substantially different geometries. We emphasize the role of exploiting physical insights in the training of the network, by defining input parameters and cost functions that assist the network to efficiently learn basic and complex propagation mechanisms.

Impact of People’s Movement on Wi-Fi Link Throughput in Indoor Propagation Environments: An Empirical Study

There has been tremendous growth in the deployment of Wi-Fi 802.11-based networks in recent years. Many researchers have been investigating the performance of the Wi-Fi 802.11-based networks by exploring factors such as signal interference, radio propagation environments, and wireless protocols. However, exploring the effect of people's movement on the Wi-Fi link throughout the performance is still a potential area yet to be explored. This paper investigates the impact of people's movement on Wi-Fi link throughput. This is achieved by setting up experimental scenarios by using a pair of wireless laptops to file share where there is human movement between the two nodes. Wi-Fi link throughput is measured in an obstructed office block, laboratory, library, and suburban residential home environments. The collected data from the experimental study show that the performance difference between fixed and random human movement had an overall average of 2.21 ± 0.07 Mbps. Empirical results show that the impact of people's movement (fixed and random people movements) on Wi-Fi link throughput is insignificant. The findings reported in this paper provide some insights into the effect of human movement on Wi-Fi throughputs that can help network planners for the deployment of next generation Wi-Fi systems.

Channel Impulse Response at 60 GHz and Impact of Electrical Parameters Properties on Ray Tracing Validations

This paper outlines a study of the effect of changing the electrical properties of materials when applied in the Wireless InSite (WI) ray-tracing software. The study was performed at 60 GHz in an indoor propagation environment and supported by Line of Sight (LoS) and Non-LoS measurements data. The study also investigates other factors that may affect the WI sensitivity, including antenna dimensions, antenna pattern, and accuracy of the environment design. In the experiment, single and double reflections from concrete walls and wooden doors are analysed. Experimental results were compared to those obtained from simulation using the WI. It was found that materials selected from the literature should be similar to those of the environment under study in order to have accurate results. WI was found to have an acceptable performance provided certain conditions are met.

Improved Model for Indoor Propagation Loss in the 5G FR2 Frequency Band

One of the latest developments today is the 5G, or 5th generation mobile network. In addition to a number of innovations, the new system also includes millimeter-wavelength frequency ranges denoted with FR2, that formerly not applied for these specific purposes. Proper management of the transmitter and receiver antenna beams is required for efficient communication in this frequency range. For future use, the simplest implementation way is electronically shaping the antenna beams by an algorithm to orient the antennas in the best possible direction. The prerequisites for these algorithms are appropriate propagation models, which are currently lacking, and those that publicly available are not accurate enough for practical use.