Other-Cell Interference Aware Precoding for the Downlink of Multi-User MIMO AF Communication

Abstract

Abstract-Amplify-and-forward (AF) is one of the most popular and simple approaches for transmitting information over a cooperative multi-input multi-output (MIMO) relay channel. In cooperative communication, relays are employed for improving the coverage or enhancing the spectral efficiency, especially of cell-edge users. However, in a multi-cell context, the use of relays will also lead to an increase in interferences that are experienced by cell-edge users of neighboring cells. In this paper, two novel precoding schemes for mitigating this adverse effect of cooperative communication are proposed. They are designed by taking into account the effect of interference coming from neighboring cells, i.e. other cell-interference (OCI), for maximizing the sumrate of cell-edge users. Our novel OCI-aware precoding schemes are compared against non OCI-aware precoding techniques and results show the large performance gain in terms of sum-rate that our schemes can achieved especially for large numbers of users and/or antennas in the multi-cell system. I. INTRODUCTION Cooperative communication has recently attracted considerable research interests [1]-[3]. Amplify-and-forward (AF) is a simple and practical approach for implementing cooperative multi-input multi-output (MIMO) communication. In AF, the relay node (RN) simply amplifies the received signal from the source node (SN) and forwards it to the destination node (DN). In the original AF scheme [3], the RN was first used as a simple equal gain amplifier. Since then, it has been shown in [4] and other many works that the RN can also be utilized as a smart precoder for improving the spectral efficiency of single and multi-user (MU) MIMO AF communication.As far as the downlink (DL) of MU MIMO AF communication is concerned, some methods have first been proposed in [5] and [6] to efficiently perform the precoding at the RN but only for the single antenna per user case. Recently in [7], a method for the MIMO case has been designed by assuming that the full channel state information (CSI) of the relay channel is available at the SN and that dirty paper coding is employed. Then in [8], we have developed three precoding methods for the DL of MU MIMO AF system, namely the AF-statistical knowledge of the relay-destination links, AF-channel block diagonalization (CBD) and AF-constrained gradient search for DL (CGSDL) methods, by considering two more realistic CSI assumptions than in [7], i.e. only the receive CSI or both receive and transmit CSI is available at the RN and, hence, without relying on DPC at the SN. All the previously cited precoding techniques only considered MU interference from different users within the same cell and, thus, did not take into account OCI. However, it has been shown

A low-complexity precoding scheme for the downlink of multi-cell multi-user MIMO AF system

“…The resulting scheme combines the advantages of our two previous schemes in [9], i.e. it provides similar sum-rate performance as the ECGSDL scheme and with the same order of computational complexity than the ECBD scheme.…”

“…by solving the gradient of E{(ŝ − s )(ŝ − s ) † } = 0, when each DN uses a linear Minimum Mean Squared Error (MMSE) receiver [9].…”

“…Then, numerous works have looked at extending this concept to various other scenarios such as single cell multi-carrier MIMO AF [6], single cell multi-user (MU)-MIMO AF [7], [8] or multi-cell MU-MIMO AF [9].…”

“…In cooperative multi-cell communication, we expect this problem to be exacerbated since one of the induced effects of relaying will be to increase the OCI due to neighboring cell-edge users. As a matter of fact, we have proposed in [9] two precoding schemes for tackling this problem, namely the AF-enhanced CBD (ECBD) and AF-enhanced constrained gradient search for DL (ECGSDL) schemes. The latter performs better than the former in terms of spectral efficiency but is more demanding in terms of computational complexity.…”

“…In Section II, we describe the system model for the DL of multi-cell MU MIMO AF and formulate the sum-rate equation in this scenario. Using this equation as an objective function, we design our novel multi-cell MU-MIMO AF precoding scheme in Section III by considering as in [9] that the SN-RN and RN-DN link channel state information (CSI) and the interference plus noise covariance matrix of each cell-edge user are known at the RN. In Section IV, we then present the sum-rate and computational complexity performances of our novel OCI-aware scheme for various settings.…”