MIMO broadcast channels with finite rate feedback

Abstract

Multiple transmit antennas in a downlink channel can provide tremendous capacity (i.e. multiplexing) gains, even when receivers have only single antennas. However, receiver and transmitter channel state information is generally required. In this paper, a system where each receiver has perfect channel knowledge, but the transmitter only receives quantized information regarding the channel instantiation is analyzed. The well known zero forcing transmission technique is considered, and simple expressions for the throughput degradation due to finite rate feedback are derived. A key finding is that the feedback rate per mobile must be increased linearly with the SNR (in dB) in order to achieve the full multiplexing gain, which is in sharp contrast to point-to-point MIMO systems in which it is not necessary to increase the feedback rate as a function of the SNR.

Comparison of Partial CSI Encoding Methods in Multi-User MIMO Systems

“…It can be shown that for n T ≤ 4 the number of such matrices is then equal to Table 1 shows the number of different modulation matrices for both types of modulation and it is clear that the numbers quickly become unmanageable even for the linear modulation. There exist methods allowing to decrease the number of matrices, either by using additional antennas to decrease quantization noise [9] or by using nested quantization on slowly varying channels [10]. In general, however, it is safe to state that the gains of non-linear precoding may not be sufficient to warrant complexity, which is orders of magnitude higher than in the linear case (see the results of simulations in the next section).…”

“…The problem of CSI quantization in multi-user systems is, however, much more complex and relatively few results are available. The case of multi-user zero-forcing algorithm was discussed in [9], where the author presents theoretical analysis of the influence of quantization noise on the system throughput and shows methods for increasing the throughput ceiling in high power regions. In [10], we approached the problem of CSI quantization with block diagonal transmission and outlined possible codeword design algorithms.…”

An Expression for the Average Quantization Error

“…The first is to support the mathematical correctness of (11) and to show that it can be computed by a computer program. The second is to evaluate the robustness of the approximations in (3) and (4) due to Jindal [2][3][4]. We perform a simulation study to assess both these aspects.…”

“…In MIMO (multiple-input multiple-output) downlink channels, the quantization error is so critical. So much so that combining methods have been developed to minimize the channel quantization error at each receiver, and thereby minimize the multi-user interference, see Au-Yeung and Love [1] and Jindal [2][3][4]. These combining methods are known as quantization-based combining methods.…”

Dynamic Resource Allocation for Scalable H.264/AVC Video Transmission over MIMO Wireless Systems

“…In addition, MIMO systems are able to acquire spatial multiplexing gain as well; i.e. ; multiple streams can be transmitted simultaneously in order to achieve high data rate in wireless multimedia communications [7][8][9][10][11][12].…”