Hybrid Single/Multi-User MIMO Transmission Based on Implicit Channel Feedback Katsutoshi Kusume ∗ , Karim Khashaba † , Guido Dietl ∗ , Wolfgang Utschick † ∗ DOCOMO Euro-labs, Landsbergerstr. 312, 80687 Munich, Germany. Email: kusume@docomolab-euro.com † Technische Universit¨ at M¨ unchen, Arcisstr. 21, 80290 Munich, Germany. Abstract—This paper investigates multiple input multiple out- put (MIMO) transmission techniques based on realistic assump- tions on feedback of channel state information. We consider three conventional techniques as the baseline: 3GPP long-term evolution (LTE) single user MIMO (SU-MIMO) based on im- plicit channel feedback, zero-forcing multiuser MIMO (ZF MU- MIMO) based on explicit channel feedback, and ZF MU-MIMO based on implicit channel feedback. SU-MIMO may not be able to exploit the full spatial dimension of the downlink MIMO channel. ZF MU-MIMO has the potential to improve the spectral efficiency, but the explicit channel feedback is not compatible with implicit feedback whereas implicit based ZF MU-MIMO is limited by performance and also the commonly assumed rank restriction makes it impossible to flexibly perform dynamic switching of SU/MU MIMO transmission. We propose a new hybrid scheme which enables such dynamic switching of SU/MU MIMO transmission by allowing user equipment to feed back the implicit channel information without any rank restriction. Computer simulation results show the benefits of the new hybrid scheme, which can properly switch to the better transmission mode in various spatial correlation scenarios. I. I NTRODUCTION In the last years standardization efforts were made that aim at the potential increase in spectral efficiency for multiple input multiple output (MIMO) wireless channels [1], [2] in commercial systems. This requires realistic assumptions on the channel knowledge at transmitter and thus, some kind of feedback mechanism is standardized to obtain some informa- tion about channel using a limited amount of feedback bits. In 3GPP Rel. 8 Long-Term Evolution (LTE), a single user MIMO (SU-MIMO) transmission mode is defined that is a point-to-point transmission scheme based on a predefined precoder codebook [3]. Given a MIMO channel, each user equipment (UE) finds the best preferred precoder from the codebook that maximizes the transmission rate. Once the best codebook entry is found, the UE then feeds back to the base station (BS) the information comprising of the so-called rank indicator (RI), precoding matrix indicator (PMI), and channel quality indicator (CQI), which respectively denote as the number of spatial streams, codebook index of precoding matrix, and SINR information. We note that UE does not feed back channel itself, but the best precoder so that the channel information is sent to the BS only implicitly. The BS receives the feedback information from all the active UEs and schedules a UE that has the best rate or has the best metric according to, e.g., proportional fair (PF) scheduling. A possible drawback of the SU-MIMO mode is that the transmitter may not be able to exploit the full spatial dimension of the downlink MIMO channel. Such situation occurs, for instance, when the number of antennas at the UEs is less than that of the BS. That is likely in practical systems. Even if both UE and BS have the same number of antennas, the spatial dimension is strictly limited when fading correlations among antenna elements are high. In such scenarios, multiuser MIMO (MU-MIMO) has the potential to better exploit the spatial dimension. Downlink (DL) MU-MIMO is a point-to-multi-point trans- mission. Due to the non-cooperative nature of the downlink channel, signal processing at the BS plays a more impor- tant role of interference suppression. Thus, certain kind of channel knowledge at the BS has a more significant impact on the performance as compared to SU-MIMO. The study initially started by assuming perfect channel knowledge at the transmitter and a number of references can be found in the literature, e.g., linear [4], [5], [6], [7] and non-linear processing [8], [9], [10], [11]. Again, realistic assumption on the channel knowledge at the transmitter is one of the key issues in commercial systems. Although Rel. 8 LTE defines a simple MU-MIMO mode that works with the same feedback mechanism as SU-MIMO, more advanced MU-MIMO techniques are discussed at 3GPP meetings targeting Rel. 10 LTE-Advanced to further improve the spectral efficiency. One such example is zero-forcing (ZF) MU-MIMO transmission based on the feedback mechanism called channel vector quantization (CVQ) [12], [13], [14]. However, CVQ is an explicit channel feedback mechanism which is different from the implicit mechanism of LTE. Therefore, some further studies attempted to replace CVQ with the implicit feedback of LTE for operating ZF MU- MIMO transmission, e.g., [15], [16]. While ZF MU-MIMO is demonstrated to work sufficiently well for highly correlated channels, the performance degrades as spatial correlation be- comes lower. When feedback amount is limited, SU-MIMO generally performs better than MU-MIMO since channel quan- tization becomes more difficult and so is the user separation. Thus, it makes sense to consider dynamic switching between SU-MIMO and MU-MIMO modes such that the system can flexibly operate in the better mode according to the present channel condition. Such dynamic switching feature has been agreed to be a part of LTE-Advanced and some mechanism has to be investigated. In this paper, we propose a new hybrid SU/MU MIMO transmission scheme that enables such flexible 978-1-61284-231-8/11/$26.00 ©2011 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE ICC 2011 proceedings