Simplified Interference Modeling in Multi-Cell Multi-Antenna Radio Network Simulations Per Skillermark, Magnus Almgren, David Astély, Magnus Lundevall, and Magnus Olsson Ericsson Research Stockholm, Sweden {per.skillermark, magnus.almgren, david.astely, magnus.lundevall, magnus.a.olsson}@ericsson.com Abstract— This paper outlines and evaluates a simplified interference model applicable in multi-cell multi-antenna radio network simulations. Based on the path-loss, the model classifies interferers as either strong or weak and the channels of strong interferers together with the channel of the desired signal are accurately modeled using a spatial channel model (SCM). The SCM assures that the spatial signature of the signals is accounted for in the evaluations. The channels of weak interferers are simply characterized by the path-loss and interference is modeled as additive white Gaussian noise (AWGN). The model is verified by means of simulations of a 57 sector OFDM/TDMA network and by comparing results achieved using the simplified model to results from simulations with full interference modeling, i.e., to the case when all interferers are accurately modeled. The verification results demonstrate that the simplified model with at least eight links accurately modeled provides a high modeling accuracy and results are comparable to results achieved with full interference modeling. Moreover, in the employed simulation tool this reduces the simulation time by up to a factor of four. The model may hence be used as a means to speed up simulations of multi-cell multi-antenna radio networks. Keywords- radio network simulations, spatial channel model, multiple antennas, interference modeling I. INTRODUCTION Multi-antenna transmitters and receivers may be used to enhance the performance of radio communication networks and such multi-antenna transceivers are essential components in, e.g., E-UTRA [1] and WINNER [2] radio access. The antennas are for example used at the transmitter side for beam forming or spatial multiplexing and for diversity and interference suppression at the receiver side. The performance of this type of radio communication networks is often evaluated by means of computer simulations, which includes models of the user behavior, the network functionality and the radio propagation channel. When simulating a network comprising multi-antenna transceivers it is central that the employed channel model captures the spatial properties of the radio channel. Examples of channel models derived for evaluations of wideband multi-antenna networks include the 3GPP SCM [3] and the WINNER channel model [4]. In these ray-based channel models, which we here refer to as spatial channel models (SCMs), the radio channel is modeled by a multitude of rays (paths) each characterized by a delay, amplitude, phase and angle. A drawback of these spatial channel models is the associated computational complexity, which in many cases makes the channel modeling the bottleneck of a radio network simulation. For each transmit-receive antenna pair the number of rays in the models typically exceeds 100 and to model such a large number of rays is indeed tedious. Furthermore, even though it is possible to pursue alternative approximate implementations based on the channel correlation, a major challenge is that in radio network simulations the radio channel must be modeled not only for the desired links but also for all interfering links. For example, when simulating a cellular network with 57 cells and with ten users in each cell 57 x 570 = 32490 links must be modeled in every time step. In addition, the computational complexity increases with the number of antennas used at the different nodes. Like shown in [5] for the downlink the interference experienced at the receiving nodes is, however, often dominated by a single or a few interfering sources. Accordingly, if these dominating sources can be identified a possible means to reduce the computational complexity is to separate the modeling of strong and weak interferers. The channel of strong interferers may be modeled in detail using a SCM while weak interfering links are coarsely modeled or simply neglected. By modeling an appropriate number of links with a SCM one may arrive at a suitable trade-off between accuracy and computational complexity. In the description of the 3GPP SCM [3] the high computational complexity of the model is recognized and a (downlink) simplified interference model is described as a means to reduce the complexity. This downlink procedure is also mentioned in [6]. Compared to the model considered in this paper the difference lies mainly in the way channels of weak interferers are modeled and that the model proposed here is described such that it may be applied in both uplink and downlink. II. SIMPLIFIED INTERFERENCE MODELING A radio receiver in a cellular system is typically exposed to interference from a multitude of sources. Some interfering signals are strong and have a significant impact on the received signal quality while weak interfering signals do only marginally influence the performance. Accordingly, when evaluating the quality of the desired signal at a radio receiver strong interferers must be carefully modeled. Weak interferers, 978-1-4244-1645-5/08/$25.00 ©2008 IEEE 1886