A Site-Specific MIMO Channel Simulator for Hilly and Mountainous Environments Jonathan S. Lu and Henry L. Bertoni NYU WIRELESS Center Polytechnic Institute of New York University Brooklyn, New York, U.S.A. lushiaoen@gmail.com and hbertoni@poly.edu Abstract— This paper presents a real-time site-specific MIMO channel simulator for communication links in rural environments. This simulator first predicts the delay, angle of arrival and departure, and amplitude of the individual multipath arrivals (direct, ground reflected, terrain diffracted, and terrain scattered) for a specified multiple antenna receiver and transmitter link. The predicted multipath characteristics are then used to compute the tapped delay line coefficients and/or frequency responses of the channel between each transmitter antenna and receiver antenna pair, which are the outputs of the simulator. To demonstrate the use of this simulator, Monte Carlo simulations of SISO and MIMO channel capacity for many databases are performed. Conclusions are drawn on the relationship between capacity, terrain roughness and other channel characteristics. Keywords— MIMO Channel Modeling, Mobile, Mountainous Terrain, Rural Propagation, Terrain Scattering I. INTRODUCTION Scattering from terrain can result in a rich multipath radio environment for radio links located in hilly or mountainous terrain [1]. To efficiently predict the multiple-input-multiple- output (MIMO) channel in rural terrain for UHF band (300 MHz – 3 GHz) military communications, a real-time site- specific MIMO channel simulator, which accounts for the scattering from terrain, was created from our previously developed site-specific propagation single-input-single-output (SISO) simulator. Previously proposed empirical MIMO channel models [2],[3] simulate the time domain channel in the form of tapped delay lines. The SUI channel model A [2] is the only one to consider hilly terrain. It was developed from 1.9 GHz measurements [4] taken in flat to hilly terrain with light to heavy tree density for fixed wireless applications. The transmit antenna heights in these measurements ranged from 12 to 79 m while the receive antenna height was 2 m. Because of the high antenna heights, relatively gentle terrain and fixed links, this model may only apply to specific radio scenarios in which the line-of-sight (LOS) is dominant. Thus, we have developed a real-time channel simulator which can apply to a greater range of radio scenarios and terrain, while accounting for the site- specific environment. For a MxN MIMO link, where the mobile transmitter (TX) has N antennas and the mobile receiver (RX) has M antennas, our proposed simulator utilizes the previously developed SISO simulator [1] to predict the local area amplitude, and angles of arrival and departure of each multipath arrival (direct, ground reflected, terrain scattered and terrain diffracted) between all TX and RX antenna pairs. The total phase of a terrain scattered arrival for an antenna pair, is determined by the propagation path length and additional phase incurred by scattering. Because available terrain databases have resolution (> 30m) larger than the UHF wavelengths (< 1m) considered in this work, the total phase cannot be deterministically known. Thus, the additional phase is modeled with a uniformly distributed random variable [5] and the path length is found from the geometry of the antennas with the center of the terrain element. Note that even though the total phase of this arrival cannot be deterministically known, the relative phases of this arrival between all antenna pairs are known from the different path lengths. After determining the multipath information of all arrivals, our simulator forms the MxN channel matrix in time or frequency domain. To demonstrate the use of the proposed simulator, we investigate the feasibility of MIMO communications in rural environments using Monte Carlo simulations of 2x2 MIMO radio links in many different types of terrain. An overview of the SISO channel simulator is presented in Section II. The methodology of our MIMO channel simulator is given in Section III, along with closed form expressions for the tapped delay line coefficients and channel frequency response in terms of the channel impulse response. The setup and procedure of the Monte Carlo simulations using our MIMO channel simulator is given in Section IV. The results of the Monte Carlo simulations are analyzed in Section V. II. RURAL ENVIORNMENT SISO CHANNEL SIMULATOR In this section we provide an overview of the previously developed SISO channel simulator for rural environments [1]. This simulator is used to predict the local area RMS voltage amplitudes a p , angles of arrival and departure, and initial propagation delays R p /c of the p = 0,1, .. P arrivals. Here c is the speed of light. A. Vertical Plane Model The Terrain-Integrated Rough Earth Model (TIREM) [6] is used to predict the received power from rays lying in the vertical plane (VP) containing the TX and RX. For line-of- sight (LOS) links, the dominant waves in the vertical plane travel along the direct and ground reflected paths. 2013 IEEE Military Communications Conference 978-0-7695-5124-1/13 $31.00 © 2013 IEEE DOI 10.1109/MILCOM.2013.135 764