Abstract
The past few years have witnessed dramatic growth in the number of wirelessly connected devices, which will continue to increase in the future. Following this trend, the capacity of the wireless networks has been enhanced to provide high-quality service to tens of billions of devices. At the same time, in response to the network enhancement, each device unashamedly requests more and more throughput to support high-data-consuming applications such as video calls, high-definition video streaming, and online multiplayer video games. This undoubtedly indicates that the demand for high wireless throughput and numerous new connections will keep increasing in the near future. In addition, the development of new technologies such as virtual/augmented reality, self-driving cars, remote surgery, and other latency-critical applications has caused concern regarding the network response latency. Thus, next-generation wireless networks have to satisfy three main requirements: i) high throughput; ii) simultaneous service to many users; and iii) low latency. Massive multiple-input multiple-output (MIMO) technology, where a base station (BS) equipped with a large antenna array is capable of serving many users simultaneously in the same time-frequency domain, has been developed to mitigate these requirements except the last. However, massive MIMO technology has to overcome the challenges related to the channel estimation (CE) overhead, which inevitably increases the communication latency, to become the absolute leader in the list of promising technologies for next-generation wireless communication. This dissertation focuses on developing solutions that are aimed to mitigate massive MIMO CE challenges. The dissertation consists of three main parts: massive MIMO channel modeling, user localization in massive MIMO networks, and full downlink channel reconstruction.
The first part (Chapter 3) discusses an approach for modeling spatially consistent channels in massive MIMO networks. The main focus is put on describing specular reflections of wireless signals from arbitrarily inclined surfaces by taking into account the signals' polarizations and the spatial distributions of massive MIMO antennas. The proposed approach has been validated through simulating signal transmissions in a realistic environment model based on Google Maps. Results show the importance of incorporating a spherical wave propagation model and the consideration of detailed 3D characteristics of the surroundings in the simulation of massive MIMO channels.
The second part (Chapter 4) introduces a solution for localizing users in massive MIMO networks. The main focus is on designing algorithms that are capable of estimating the positions of users using only uplink signals by exploring the advantages of the spherical wave propagation model proposed in the first section. The designed localization schemes have been evaluated through both simulation and proof-of-concept experiments. Simulation results show that the schemes can achieve decimeter-level localization accuracy using 64 and more antenna elements for distances up to 300 meters. The proof-of-concept experiment justifies the feasibility of user localization based on the estimation of the spherical shape of the incoming wavefront.
The third part (Chapter 5) investigates the problem of reconstructing the full downlink channel from incomplete uplink channel measurements in massive MIMO systems. This problem arises in the next-generation networks, where connected devices have multiple transmitting and non-transmitting antennas. To achieve high throughput, channels for non-transmitting antennas have to be reconstructed. This section presents ARDI, a scheme that builds a bridge between the radio channel and physical signal propagation environment to link spatial information about the non-transmitting antennas with their radio channels. By inferring locations and orientations of the non-transmitting antennas from an incomplete set of uplink channels, ARDI can reconstruct the downlink channels for non-transmitting antennas. The performance evaluation results demonstrate that ARDI is capable of accurately reconstructing full downlink channels when the signal-to-noise ratio is higher than 15dB, thereby expanding the channel capacity of massive MIMO networks.