Resumen de Serious game engineering and lighting models for the realistic emulation of 5g systems
The fifth generation of mobile communications, 5G, promises to be a technological revolution that goes beyond multiplying the data transmission speed of its predecessors. It aims to support a large number of devices and reach latencies very close to 1 millisecond. To meet these ambitious requirements, new enabling technologies have been researched. One of these is the use of millimetre-wave bands (mmW) in which a large amount of spectrum is available.
Complex channel models are required to predict radio channel characteristics and reliably evaluate 5G performance in the mmW bands. Specifically, the most accurate propagation models are those based on ray tracing, but their high computational cost makes them unfeasible for radio channel characterization in complex scenarios. On the other hand, in recent years, video game technology has developed powerful tools to model the propagation of light in super realistic scenarios. Given the spectral closeness between the visible spectrum and the mmW waves, the present Thesis has studied the application of light propagation modeling tools from game engines for radio channel modeling in mmW.
This Thesis proposes a model for estimating propagation losses in mmW called "Light Intensity Model'' (LIM). Using this model, based on the lighting processes performed by the game engines, the signal transmitters are replaced by light sources and the light intensity received at a point is translated into signal strength in mmW through a simple polynomial function.
One of the advantages of using the game engines is their great capacity and the ease with which the user can create super realistic scenarios that faithfully represent the geometry of scenarios where the radio channel is to be evaluated. In this way, accurate estimates of propagation losses can be obtained. The estimation of propagation losses with LIM has been compared with measurement campaigns in the 28 GHz and 73 GHz bands and with other propagation models. As a result, the LIM estimation error is smaller than the current stochastic models and is comparable with the ray tracing model. In addition, the computational cost of LIM compared to ray tracing is 130 times lower, allowing the use of LIM in highly complex scenarios for real-time radio channel estimation.
The game engines allow to characterize in a different way the interaction of the materials with the light configuring the normal map of their surfaces and their scattering and reflection functions. In this Thesis it has been determined the characterization of several materials that best fits to laboratory measurements made in a controlled scenario in the 28 GHz band. The LIM model using materials with this optimal characterization reduces by more than 50% its estimation error with respect to the application of LIM with default materials, while its computational cost remains 26 times lower than the ray tracing model.
Finally, a first version of a platform for the emulation of 5G systems has been developed on a game engine, which is the starting point for a complete 5G emulator. This platform not only contains the LIM model but also includes several 5G use cases in super realistic environments. The platform, which is based on the concept of "`Serious Game Engineering", breaks the limitations of mobile network simulators in terms of visualization capabilities and user interaction with network components in real time.
