Channel feedback optimization for network MIMO systems

Abstract

Network MIMO system is one of attractive techniques for combating with inter-cell interference. In the dissertation, we investigate channel feedback optimization for network multiple-input multiple-output (MIMO) systems. Among two types of network MIMO system, `Coordinated beamforming' and `Joint processing', we focus on latter type of network MIMO system and propose several feedback techniques enhancing the system performance.

In the first part of the dissertation, the selective channel feedback scheme is developed and the rate loss due to the selective channel feedback is derived. Unlike single cell multiuser system, network MIMO systems has unique characteristic called \textit{channel asymmetry} that the average channel gains from base stations to a certain user are different each other. Using \textit{channel asymmetry}, it is shown that reporting a few relatively strong channels with high accuracy is better than reporting all channels with low accuracy when the available feedback resource is limited. At first, the upper bound of rate loss when users have a single receive antenna is derived and then the upper bound of rate loss when users have multiple receive antenna is derived. Numerical results show that the derived upper bound of rate loss approaches the real rate loss within 2 bps/Hz in case of the number of receive antenna is one. In addition to the selective channel feedback, unequal feedback bit allocation is introduced. When total feedback resource is limited, allocating unequal feedback bits to the users can improve system performance and selective channel feedback further enhance the system performance with given feedback bits. And also, it is worth noting that selective channel feedback and feedback bit allocation affect each other. Thus, joint optimization is required to exploit the limited feedback resource efficiently. Through computer simulations, we show that the proposed algorithms provide significant amount of performance gain. The convergence proof of \textit{Iterative algorithm} is also provided. In the second part of the dissertation, threshold-based feedback user selection is developed. When total feedback resource is limited, it might be better to allow the users whose channel gain is relatively good and quantization error is small to report their channel state information. Therefore it is proposed that users whose efficient channel gain, the product of instantaneous channel gain and cosine of quantization error angle, exceed the threshold only can report their channel state information. We derive the proper number of feedback users in average sense that can balance the multiuser diversity gain and feedback accuracy. We also derive the threshold for each user that can guarantee the feedback possibility of users all the same. Numerical results reveal that the proposed threshold based algorithm gets a significant performance improvement. Finally, we propose a new feedback solution incorporating feedback user selection and joint optimization of selective feedback and unequal feedback bit allocation in the first part of the dissertation. Numerical results show that proposed feedback solution can enhance the average achievable rate further the individual schemes.