Capacity of Measured MIMO Channels in Dependence of Array Element Spacing and Distance between Antennas D. Yacoub * , C. Schneider † , S. Warzuegel ‡ , W. G. Teich * , R. Thom¨ a † , J. Lindner * * University of Ulm, Department of Information Technology, Ulm, Germany Email: {doris.yacoub, werner.teich, juergen.lindner}@uni-ulm.de † University of Ilmenau, Dept. of Electrical Engineering and Information Technology, Ilmenau, Germany Email: {Christian.Schneider, Reiner.Thomae}@tu-ilmenau.de ‡ MEDAV GmbH, Uttenreuth, Germany Email: steffen.warzuegel@medav.de Abstract—In this paper, we investigate the effect of antenna spacing within the transmitter and receiver arrays as well as the distance between the transmitter and receiver arrays on the capacity of a measured multiple-input multiple-output (MIMO) channel. During the measurement, the transmitter was mounted on a vehicle moving at a constant speed towards the receiver which was mounted on a bridge. The capacity, as expected, was found to increase with increased antenna element spacing within the transmitter or receiver arrays. However, the resulting capacity curves versus distance were found to fluctuate as the transmitter approached the receiver. Reflections from the street surface are assumed to be the cause of those fluctuations. In order to test this theory, we simulated a simple two-path channel model. I. I NTRODUCTION In the last few years, the need for reliable transmission at high rates has shifted the attention towards multiple-input multiple-output (MIMO) systems. The capacities of MIMO channels have been calculated for Rayleigh fading channel models (e. g. [1]) as well as for frequency selective channel models (e. g. [2]). In this paper, we look at the capacity of a real measured channel with a strong line of sight (LOS) component and investigate the effect of array element spacing and the seperation between the transmitter and receiver arrays on the MIMO channel capacity. The transmitter was mounted on a vehicle moving at a low speed towards the receiver, which was mounted on a bridge. The channel was measured for several transmitter as well as receiver array element spacings and was continuously measured as the transmitter approached the receiver. The results have shown that despite the strong LOS component, the capacity was found to fluctuate in a periodic manner as the vehicle moved. Reflections from the street surface are assumed to be responsible for such an effect. To test this assumption, a simple two-path model was simulated and compared to the measured channels. Throughout this paper, vectors and matrices will be denoted by underlined and doubly-underlined letters, respectively. Low- ercase letters will be used for the time domain whereas the uppercase letters for the frequency domain. Scalars will be simply designated by letters. II. MEASURMENT SETUP In this section, we briefly describe the measurement setup at both the transmitter and the receiver. The transmitter (Tx) was mounted on a vehicle moving at a constant low speed (around 10 km/h, doppler shift thus neglected) towards a stationary receiver (Rx), which was mounted on a bridge (Fig. 1). The channel measurements were started as the vehicle was about 217 m in front of the bridge and stopped after it passed the bridge by around 62 m (Fig. 1). The channel measurements were carried at a center frequency, f o , of 5.2 GHz and for a bandwidth, BW , of 120 MHz. For more details about channel sounder measurements, please refer to [3]. The transmitter and receiver array setups are shown in Fig.2. The receive elements were tilted down by 45 o . The MIMO channel was measured for three different setups at the receiver and two setups at the transmitter. The setups correspond to different antenna element spacings within either the transmitter or receiver arrays. The total span, W , and separation between the antennas, d (also given as a function of the wavelength, λ =0.0577 m), for the three setups at the receiver are given in the following table. Receiver Setup W [m] d[m] Large (L) 17.5 2.5 ≈ 43λ Medium (M) 6.16 0.88 ≈ 15λ Small (S) 1.022 0.146 ≈ 2.5λ At the transmitter, the distances d 1 and d 2 for the two different setups are as follows: Transmitter Setup d 1 [m] d 2 [m] Large (L) 0.862 ≈ 15λ 0.944 ≈ 16λ Small (S) 0.185 ≈ 3λ 0.185 ≈ 3λ All combinations of the transmitter and receiver setups (6 in total) were used for channel measurement as well as for the capacity calculations. In all what follows, we shall denote the setups by the receiver setup (L, M or S) followed by the transmitter setup (L or S). For example, LS denotes the results of the MIMO channel measured using large receive and small transmit array element spacings.