Multimode precoder design for STBC with limited feedback in MIMO based wireless communication system

Abstract

Abstract:In this paper, a multimode precoder design for spacetime block code (STBC) is investigated, which varies the number of streams depending on the channel condition. We develop a design criterion of minimizing the vector symbol error rate and derive an efficient offline algorithm to generate precoders. Simulation results show that the multimode precoded STBC (MM-STBC) provides substantial performance improvements compared with the single mode precoded orthogonal STBC (SM-STBC) for a fixed data-rate. It also outperforms the multimode precoded multiple-input multiple-output (MM-MIMO) system when feedback rate is very limited.

Precoder design with non-uniform power allocation for multimode precoded MIMO schemes with limited feedback

“…The number of feedback bits is assumed to be three, which results in N 1 = 4, N 2 = 4 precoder allocation of modes for the MM-SM by using the feedback bit allocation criterion in [2]. Since the MM-STBC can save one bit for full diversity compared to the MM-SM [3], at least two M t × 2 precoders are allocated to Mode 1 and the remainder M t × 4 precoders are allocated to Mode 2, i.e., N 1 = 2, N 2 = 6 for MM-STBC. Results show the exhaustive search design (III-B) achieves around 1 dB and 0.6 dB performance gain at the error rate 10 −3 compared to the unit-normalized precoder design for the MM-SM [2] and MM-STBC [3], respectively.…”

“…A set of input bits is demultiplexed in to M different bit streams for each channel. The value of M is referred to the mode of the precoder and varies between 1 and M t /2 J where J = 0 or 1 indicates multimode precoded spatial multiplexing transmission (MM-SM) [2] or multimode precoded STBC transmission (MM-STBC) [3], respectively. The reason why M t is divided by 2 for MM-STBC is because that each bit stream is transformed to a 2 × 2 matrix by a Alamouti STBC [5].…”

“…Taking approach in [2] and [3], we observe that the upper bound of the VSER for both MM-SM and MM-STBC with zero forcing (ZF) receiver is…”

“…However, the full CSI require significant feedback overhead, which is not practical. Therefore, various types of limited feedback systems have been designed [2,3], which achieve the water-filling (WF) gain from the perspective of designing a codebook. However, the codebook contains only a set of columns of quantized eigen beamforming and the total power is equally allocated to the eigenmodes.…”

“…Compared to [2] and [3], we relax equal power allocation constraint and use a generalize Lloyd algorithm [4] to design a codebook. The difficulty comes from the computation of centroid of Lloyd algorithm which is formulated as an optimization problem.…”