A Generalized Matched Filter Framework for Cellular Massive MIMO Networks

点击翻译按钮获取中文摘要

"Massive MIMO", also referred to as "Very-Large Scale MIMO Systems", stands for a recently introduced new paradigm in wireless communications considered to be an important candidate for the required technological progress necessary for the upcoming 5th generation of mobile communication systems. The huge number of antenna elements causes new benefits as well as challenges in wireless communication systems. One of the main challenges in Massive MIMO systems from the signal processing perspective is a consequence of the limited coherence time of the transmission channel. The desired number of users simultaneously served in neighboring cells of the network typically exceeds the number of channel accesses per coherence block dedicated for training. In other words, it appears to be impossible to train the desired number of channels without reusing training resources, which consequently introduces interference that is referred to as the so called "Pilot Contamination" phenomenon. Pilot Contamination ultimately jeopardizes the Massive MIMO paradigm as it degrades the potential performance gains even for an huge number of base station antennas. The focus of the proposed project is on finding novel concepts to break the "Dimensionality Bottleneck" in cellular Massive MIMO networks - imposed by the limited coherence time of the transmission channel. We will carry out a study of the full potential of beamforming and interference coordination in Massive MIMO networks based on Second-Order Statistics of the CSI and taking into account the requirement of numerically efficient solutions. The analysis will start from a single-cell perspective and continue with a study of multi-cell scenarios. The corresponding resource allocation strategies will have to be jointly designed with the targeted beamforming and interference coordination methods.Since the proposed work packages will be dominantly based on "Second-Order Statistics" of channel states, the acquisition of covariance matrices of channels will obviously play an essential role for the targeted solutions in the course of this project.