Spatial mode switching in multi-user MIMO-OFDM systems: Performance in a real-world scenario Malte Schellmann, Lars Thiele, Thomas Wirth, Volker Jungnickel, Thomas Haustein Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute Einsteinufer 37, 10587 Berlin, Germany {malte.schellmann, lars.thiele, thomas.wirth, volker.jungnickel, thomas.haustein}@hhi.fraunhofer.de Abstract— Applying spatial mode switching in channel adaptive MIMO systems promises to achieve high spectral efficiencies in practical system setups. Several studies have pointed out significant performance gains obtained in realis- tically modeled channel scenarios. We have conducted chan- nel measurements with the Berlin LTE-Advanced Testbed operated in the city center of Berlin and used these for the performance evaluation of a spatial mode switching concept that has previously been evaluated in a link- and system- level simulation environment. Results from measurements and simulations are shown to be in close correspondence. I. I NTRODUCTION Channel adaptive transmission combined with multi- ple antenna technology (multiple-input multiple-output (MIMO)) is seen as a promising concept to achieve high spectral efficiencies in future radio networks [1]. While MIMO enables spatial multiplexing transmission of multiple data streams in the same frequency resource, the channel adaptive concept allows to form precoding vectors matched to the MIMO channel as well as to adapt the data rate to the current signal to interference and noise ratio (SINR) conditions. The basic idea of adaptivity is that the user terminals (UTs) provide feedback on their actual channel conditions to the base station (BS) [2]. This information is used at the BS within a resource allocation process to assign those transmission resources to the UTs that support high data rates. For channel adaptive MIMO transmission, there are two options that enable boosting the achievable spectral efficiencies further: The first one is termed spatial mode switching, where the system may select the number of independent data streams that are transmitted simul- taneously in the spatial domain of a given frequency resource [3], [4]. Reducing the number of spatial streams may be especially advantageous if the MIMO channel is ill-conditioned or if a user experiences low SNR conditions. The second option is the efficient applica- tion of multi-user MIMO (MU-MIMO), which allows to assign the simultaneously transmitted spatial streams to different users [5], in contrast to single-user MIMO (SU-MIMO), where all spatial streams are assigned to the same user. For an orthogonal frequency division multiplexing (OFDM) based MIMO system with fixed data user 1 user selection & stream generation spatial pre- coding 1 Nt mod mod user scheduling and mode selection signaling link signaling link beam 1 beam N t data user K Q 1 ... ... ... detected user data adaptive mux spatial equali- zation Nr 1 demod demod channel estimation and evaluation unit 1 ... ... detected user data adaptive mux spatial equali- zation Nr 1 demod demod channel estimation and evaluation unit 1 ... ... base station terminal 1 terminal K Nr Nr Fig. 1. System concept for channel-adaptive transmission in multi-user MIMO downlink. precoding vectors, it has been shown in [6], [7] that proper application of MU-MIMO enables the efficient use of spatial multiplexing transmission even at low SNR conditions. The performance of fixed precoding MIMO transmis- sion supporting spatial mode switching in an OFDM system with a configuration similar to the 3G Long Term Evolution (LTE) specification has recently been evaluated in a link-level [6] as well as in a system- level environment [7]. While link-level simulations were based on the WINNER channel model (WIM) [8], system- level simulations were based on the spatial channel model extended (SCME) [9]. For both channel models, co- polarized MIMO antennas have been assumed. In this paper, we carry out similar investigations for real-world channels measured with the Berlin LTE-A Testbed for a 2 × 2 MIMO setup with cross-polarized MIMO antennas. Comparison of the obtained results with those from link- and system-level investigations reveals close correspon- dence. II. SYSTEM CONCEPT We consider the downlink of a broadband multi-user MIMO-OFDM system, where a BS with N t antennas communicates with K UTs equipped with N r antennas each (refer to Fig. 1). The BS provides B ≥ N t fixed beams b u , which are used for spatial precoding of the transmission signals (grid of beams (GoB) concept, pre- sented in [10]). Up to N t beams may be simultaneously