Adaptive limited feedback for intercell interference cancelation in cooperative downlink multicell networks Berna ¨ Ozbek 1,2 , Didier Le Ruyet 1 1 Electronics and Communications Laboratory, CNAM, 292 rue Saint Martin, 75141 Paris, France 2 Electrical and Electronics Eng. Dep., Izmir Institute of Technology, Urla, 35430, Izmir, Turkey bernaozbek@iyte.edu.tr,didier.le_ruyet@cnam.fr Abstract—In this paper, we evaluate intercell interference can- celation techniques in cooperative downlink multicell systems. We compare two classical beamforming strategies for the design of the beamforming vectors: maximum ratio combining precoding and partial zero-forcing. Since full channel state information at the transmission is unfeasible, we consider a limited feedback link and compare different strategies for bit partitioning of the feedback information for serving and interferer base stations. We propose a simple adaptive bit partitioning strategy to select an optimum transmission scheme at the base station. We show that the adaptive bit portioning algorithm is improved the sum capacity performance of the multicell networks significantly. I. I NTRODUCTION Coordinated multicell transmission has recently received significant attention since it can increase data rates by mitigat- ing intercell interference (ICI) [1]. Full cooperation between the base stations (BSs) also called network multiple input multiple output (MIMO) scheme transforms the interference channel into a MIMO broadcast channel. Dirty paper coding and sub-optimal linear beamforming such as Zero Forcing (ZF) have been investigated in [2] [3] and upper bounds on the sum rates have been obtained for multicell transmission. Assuming that a linear transceiver processing is employed, a base station (BS) equipped with   antennas is able to accommodate up to   −1 interference signals in a coordinated multicell transmission. Coordinated multicell transmission requires channel state information (CSI) of all jointly processed links. A large amount of data needs to be exchanged between the BSs. For example, the requirement for CSI grows in proportion to the number of transmit antennas, the number of BSs and the number of users. In order to reduce the load on the backhaul links, the partial cooperative strategies have been considered in [4]. In coordinated single-cell transmission, the BSs share only CSI and no data exchange is required where a part of ICI is removed. Besides, it is possible to apply different beamforming strategies such as Maximum Ratio Combining (MRC) and Partial Zero-Forcing (PZF) beamforming [5] by removing some ICI. In [6], the authors have considered ICI cancelation for the mitigation of other cell interference. They have proposed an adaptive strategy where the BSs exchange the position of their users and then select the best transmission strategy between MRC and PZF. They have also shown that the adaptive strategy can provide significant gain compared to the schemes without ICI. In [7], a limited feedback scenario is considered for a multicell beamforming approach that approximately maximizes the sum-rates in a two-cell two-user network by quantizing the CSI of serving and interfering base station with the usage of two separate codebooks. In [8], a chordal distance-based compression scheme is introduced to reduce the feedback overhead and combat the intercell interference. In this paper, we propose a simplified bit allocation algo- rithm to transfer CSI of the users for limited feedback link to reduce the backhaul load. In the proposed algorithm, selection of the transmission strategies is only dependent of the position of the user in the multicell network rather than the position of the users in other cells. We demonstrate that the proposed algorithm outperforms the uniform bit allocation scheme for feedback link. This paper is organized as follows. In Section II, we describe the system model for the coordinated single cell transmission including limited feedback links. In Section III, we present the proposed algorithm to select transmission strategies and bit partitioning schemes for feedback links. In Section IV and V, we illustrate the performance results and draw the concluding remarks of this paper respectively. II. SYSTEM MODEL We consider a downlink multicell wireless network with  cells. As shown in Figure 1, the base station at each cell is composed of   transmit antennas and each user is equipped with one single antenna. Using a narrow band flat-fading model, the baseband re- ceived signal for the user in the  ℎ cell is written as,   = √  , h  , f    + ∑ ∕= √  , h  , f    +   (1) where  , is the received power at the  ℎ user from the  ℎ BS, h , is the channel between the  ℎ BS and the user of the  ℎ cell. f  is the beamforming vector at the  ℎ base station and   is the complex additive zero mean white Gaussian noise 978-1-4244-6317-6/10/$26.00  2010 IEEE ISWCS 2010 81