Implementation Concepts for Distributed Cooperative Transmission V. Jungnickel, L. Thiele, M. Schellmann, T. Wirth, A. Forck Fraunhofer Institute Telecommunications, Heinrich-Hertz-Institut Einsteinufer 37, D-10587 Berlin, Germany W. Zirwas, T. Haustein, E. Schulz Nokia Siemens Networks GmbH & Co.KG St. Martinstraße 76, D-81617 Munich, Germany Abstract—Information theory predicts huge performance gains in terms of spectrum efficiency by using cooperative transmission from multiple base stations. Cooperation eliminates inter-cell interference and it enables a higher channel rank. The isolated cell capacity is an upper bound for unlimited backbone capacity, infinite signal processing power and infinite channel feedback. We discuss a combined physical, medium access control (MAC) and network layer approach targeting minimal implementation loss while relaxing assumptions. It is based on frequency-selective channel quality identifier (CQI) reports to the serving station selecting the right user(s) on a given resource with best signal to interference and noise ratio (SINR). Such scheduling is performed independently in each cell. In adjacent cells it forms a group of users whose mutual interference is smaller than on average. The task of cooperative transmission is to further reduce the mutual interference within this group. In a multi-user system, the group shares only part of the spectrum. Feedback can hence be limited to the granted resource blocks. Based on virtual pilot sequences we reduce the pilot overhead and use a suitable mirror feedback scheme in addition. Altogether the feedback overhead is reduced by 2 orders of magnitude. I. I NTRODUCTION Cooperative transmission refers to multiple base stations serving jointly a set of users scheduled onto the same resource in their respective cells. Using cooperative signal processing, the interference among the cells is reduced. For this reason, base stations exchange data signals. Channel information is measured at the terminal side from broadcast pilot signals and fed back to the base station. The principal concept has been proposed in [1, 2]. See also a performance analysis in [3] and a recent review article [4]. A measurement-based analysis is provided in [5]. Current cellular systems are interference-limited at least for small inter-site distances between 500 m and 1 km typical for the deployment of the Universal Mobile Telecommuni- cations System (UMTS) in urban areas. In the information- theoretic limit cooperative transmission could reach the bound of isolated cells [6] if we provide unlimited backbone capacity, infinite signal processing power and infinite channel feedback. The backbone may be realized using optical fibers between the base stations or replacements as microwave and free- space optics. Signal processing power becomes increasingly available at low cost [7]. However, feedback is ultimately bounded. The transmitter power of mobile terminals limits the achievable up-link capacity over cellular distances. Battery power is limited as well. In this paper, we develop a high-level implementation B1 T1 B2 exchange data and CSI over X2 local processing unit local processing unit channel feedback channel feedback T2 H 11 H 22 d 1 d 1 x 1 x 2 d 2 d 2 H 12 H 21 Figure 1. Principle of distributed cooperative transmission. concept for distributed cooperative transmission. We discuss essential ingredients such as synchronization of base stations, multi-cell channel estimation, efficient feedback of channel state information (CSI) and propose an integrated protocol ar- chitecture for the physical, MAC and network layers realizing a distributed cooperative cellular system. The paper is organized as follows. Section II introduces the distributed implementation concept. In Section III, syn- chronization of the base stations is considered. Section IV describes how to get the CSI efficiently to cooperative base stations. Section V investigates distributed cooperative trans- mission, the role of the effective channel and the benefit of user-specific pilots. Section VI summarizes the proposed cooperative transmission protocol. II. DISTRIBUTED I MPLEMENTATION One of the keys to introduce cooperative transmission into modern cellular systems is that the cooperative algorithm can be implemented in a distributed fashion, see [8]. In early work, e.g. [1–5] cooperative beam-forming has been performed at a central unit having all channel knowledge and where all data streams are available. Transmitted signals are jointly precoded and then distributed to the base stations acting merely as remote radio heads distributed across the service area. However, modern cellular networks as the Long Term Evolution (LTE) of the Third Generation Partnership Project (3GPP) have a flat and no traditional hierarchic architecture. Base stations act as autonomous network nodes. There is no central unit where cooperative processing could be done. 1035 978-1-4244-2941-7/08/$25.00 ©2008 IEEE Asilomar 2008