Pham Q. Viet

Raju Shrestha, Tien Ngoc Ha, Pham Q. Viet et al.

Radio maps provide metrics such as the received signal strength at every location in a geographical region of interest. Extensive research has been carried out in this context, but it relies almost exclusively on synthetic-data experiments. Thus, the practical aspects of the radio map estimation (RME) problem as well as the performance of existing estimators in the real world remain unknown. To fill this gap end, this paper puts forth the first comprehensive, rigorous, and reproducible study of RME with real data. The main contributions include (C1) an assessment of the viability of RME based on the estimation error that can be achieved, (C2) the analysis of the main phenomena and trade-offs involved in RME, including the experimental verification of theoretical findings in the literature, and (C3) a thorough evaluation of a wide range of estimators on realworld data. Remarkably, this reveals that the performance gain of existing deep estimators in their pure form may not compensate for their complexity. A simple enhancement (C4) is proposed to alleviate this issue. The vast amount of data collected for this study is published along with the developed simulator to enable research on new schemes, hopefully bringing RME one step closer to practical deployment.

Pham Q. Viet, Daniel Romero

Radio map estimation (RME) involves spatial interpolation of radio measurements to predict metrics such as the received signal strength at locations where no measurements were collected. The most popular estimators nowadays project the measurement locations to a regular grid and complete the resulting measurement tensor with a convolutional deep neural network. Unfortunately, these approaches suffer from poor spatial resolution and require a great number of parameters. The first contribution of this paper addresses these limitations by means of an attention-based estimator named Spatial TransfOrmer for Radio Map estimation (STORM). This scheme not only outperforms the existing estimators, but also exhibits lower computational complexity, translation equivariance, rotation equivariance, and full spatial resolution. The second contribution is an extended transformer architecture that allows STORM to perform active sensing, by which the next measurement location is selected based on the previous measurements. This is particularly useful for minimization of drive tests (MDT) in cellular networks, where operators request user equipment to collect measurements. Finally, STORM is extensively validated by experiments with one ray-tracing and two real-measurement datasets.

Pham Q. Viet, Daniel Romero

The deployment of Aerial Base Stations (ABSs) mounted on board Unmanned Aerial Vehicles (UAVs) is emerging as a promising technology to provide connectivity in areas where terrestrial infrastructure is insufficient or absent. This may occur for example in remote areas, large events, emergency situations, or areas affected by a natural disaster such as a wildfire or a tsunami. To successfully materialize this goal, it is required that ABSs are placed at locations in 3D space that ensure a high quality of service (QoS) to the ground terminals. This paper provides a tutorial introduction to this ABS placement problem where the fundamental challenges and trade-offs are first investigated by means of a toy application example. Next, the different approaches in the literature to address the aforementioned challenges in both 2D or 3D space will be introduced and a discussion on adaptive placement will be provided. The paper is concluded by discussing future research directions.