A UNIFIED FEEDBACK SCHEME FOR DISTRIBUTED INTERFERENCE MANAGEMENT IN CELLULAR SYSTEMS: BENEFITS AND CHALLENGES FOR REAL-TIME IMPLEMENTATION Lars Thiele ⋆ , Thomas Wirth ⋆ , Thomas Haustein ⋆ , Volker Jungnickel ⋆ , Egon Schulz † , and Wolfgang Zirwas † ⋆ Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, Einsteinufer 37, D-10587 Berlin, Germany † Nokia Siemens Networks GmbH & Co.KG, St. Martinstrasse 76, D-81617 Munich, Germany email: {lars.thiele, thomas.wirth, thomas.haustein, volker.jungnickel}@hhi.fraunhofer.de ABSTRACT This paper contributes to the system concept of collaborative base stations from the perspective of distributed computing. Promising signal processing approaches based on terminal feedback are re- viewed. Their specific advantages and disadvantages are discussed and a framework for feedback provision is presented. All proposed schemes have in common, that a mobile terminal can choose its desired receive strategy independently from other mobile terminals according to its computational capabilities. The effective multi-cell channel after receiver processing is fed back and distributed within the collaboration area. This allows distributed processing at each base station and makes real-time implementation feasible. 1. INTRODUCTION Multiple antenna systems have been shown to allow an active ex- ploitation of the spatial degrees of freedom in order to increase the spectral efficiency and boost throughput in wireless communication systems. In particular, spatial separation of simultaneously trans- mitted data streams can be performed either at the transmitter or the receiver depending on the available channel state information (CSI) and the possibilities of joint signal processing. multiple- input multiple-output (MIMO) signal processing got high momen- tum in standardization on link level utilizing spatial multiplexing or space-division multiple access (SDMA) in standards like IEEE 802.11n, WiMAX and 3G Long Term Evolution (3G-LTE). If these so called point-to-point MIMO systems are operated isolated from each other, multi-antenna signal processing provides signal diver- sity reception in fading channels and multiplexing options in full rank channels at medium and high SNR. Early real-time measure- ments proved the expected gains in throughput and coverage in a field trial in downtown Berlin in 2007 and 2008 [1], [2]. If these MIMO systems connecting a base station and several terminals are deployed in a cellular environment with full frequency reuse cochannel interference (CCI) becomes a limiting factor. It has been discussed early that the concept of MIMO signal separation can be applied for active interference management. Suppression of interference from adjacent cells will become a key issue in cellular mobile communication systems. In principle, the MIMO concept has to be extended to a higher number of antennas involved in the joint signal processing. Higher order MIMO systems do not only scale the computational complexity but also face serious challenges regarding distributed channel data collection and coherent signal transmission or reception at antenna locations which might easily have distances of more than 1000 meters in between. Important is- sues like synchronization of collaborative base stations and signal flows in collaborative MIMO systems were addressed in [3]. In this paper we will focus on the active interference manage- ment of collaborative base stations (BSs) signal processing in the cellular downlink, see Fig. 1. Methods to efficiently pre-process and collect CSI from distributed mobile terminals (MTs) before feeding back this information to the serving BS are compared. We consider a cellular deployment with a decentralized signal processing archi- tecture as proposed for 3G-LTE-Advanced. We approach the solu- Figure 1: Coherent transmission of collaborative base stations tion from the view point of coordinated and distributed computing and show that this approach allows channel adaptive and coherent signal transmission from several BSs for active interference man- agement in a cellular collaboration area (CA). 2. SYSTEM MODEL The downlink MIMO-OFDM transmission system for an isolated sector with N T transmit and N R receive antennas per MT is de- scribed on a per sub-carrier basis y = HCx + n , (1) where H is the N R × N T channel matrix and C the unitary N T × N T pre-coding matrix; x denotes the N T × 1 vector of transmit symbols; y and n denote the N R × 1 vectors of the received signals and of the additive white Gaussian noise (AWGN) samples, respectively, with covariance E{nn H } = σ 2 I. In the following we consider the downlink channel of a cellular system where the frequency resources are reused in all neighboring cells. Depending on the deployment of BS sites and the actual posi- tion of the MT, the user will receive interfering signals sent to other users in addition to its desired signal. As an initial step, assume that all BSs provide Ω fixed unitary beam sets C ω , ω ∈{1, ..., Ω}. In general, each beam set contains α N T fixed pre-coding vectors (beams) b ω,u with u ∈{1, ..., α N T }, where α denotes the size of the CA. Each CA i independently se- lects one of these sets. In the following we assume that each user m is served with a single data stream, while the CA uses α N T active beams. The received downlink signal y m at the MT m in the cellular environment is given by y m = H m i b i,m    h i,m x i,m + αN T ∑ j=1 j=m H m i b i, j x i, j    ζ i,m + ∑ ∀l l =i N T ∑ j=1 H m l b l , j x l , j + n    z i,m (2) 17th European Signal Processing Conference (EUSIPCO 2009) Glasgow, Scotland, August 24-28, 2009 © EURASIP, 2009 1489