Doctoral Thesis: Massive MIMO in Real Propagation Environments

Sammanfattning: Mobile communications are now evolving towards the fifth generation (5G). In the near future, we expect an explosive increase in the number of connected devices, such as phones, tablets, sensors, connected vehicles and so on. Much higher data rates than in today's 4G systems are required. In the 5G visions, better coverage in remote regions is also included, aiming for bringing the current "4 billion unconnected" population into the online world. There is also a great interest in "green communications", for less energy consumption in the ICT (information and communication technology) industry. Massive MIMO is a potential technology to fulfill the requirements and visions. By equipping a base station with a large number, say tens to hundreds, of antennas, many terminals can be served in the same time-frequency resource without severe inter-user interference. Through "aggressive" spatial multiplexing, higher data rates can be achieved without increasing the required spectrum. Processing efforts can be made at the base station side, allowing terminals to have simple and cheap hardware. By exploiting the many spatial degrees of freedom, linear precoding/detection schemes can be used to achieve near-optimal performance. The large number of antennas also brings the advantage of large array gain, resulting in an increase in received signal strength. Better coverage is thus achieved. On the other hand, transmit power from base stations and terminals can be scaled down to pursue energy efficiency. In the last five years, a lot of theoretical studies have been done, showing the extraordinary advantages of massive MIMO. However, the investigations are mainly based on theoretical channels with independent and identically distributed (i.i.d.) Gaussian coefficients, and sometimes assuming unlimited number of antennas. When bringing this new technology from theory to practice, it is important to understand massive MIMO behavior in real propagation channels using practical antenna arrays. Not much has been known about real massive MIMO channels, and whether the claims about massive MIMO still hold there, until the studies in this thesis were done. The thesis study connects the "ideal" world of theory to the "non-ideal" reality. Channel measurements for massive MIMO in the 2.6 GHz band were performed, in different propagation environments and using different types of antenna arrays. Based on obtained real-life channel data, the studies include • channel characterization to identify important massive MIMO properties, • evaluation of propagation conditions in real channels and corresponding massive MIMO performance, • channel modeling for massive MIMO to capture the identified channel properties, and • reduction of massive MIMO hardware complexity through antenna selection. The investigations in the thesis conclude that massive MIMO works efficiently in real propagation environments. The theoretical advantages, as observed in i.i.d. Rayleigh channels, can also be harvested in real channels. Important propagation effects are identified for massive MIMO scenarios, including channel variations over large arrays, multipath-component (MPC) lifetime, and 3D propagation. These propagation properties are modeled and included into the COST 2100 MIMO channel model as an extension for massive MIMO. The study on antenna selection shows that characteristics in real channels allow for significant reductions of massive MIMO complexity without significant performance loss. As one of the world's first research work on massive MIMO behavior in real propagation channels, the studies in this thesis promote massive MIMO as a practical technology for future communication systems.