Proceedings of the 1998 Winter Simulation Conference D.J. Medeiros, E.F. Watson, J.S. Carson and M.S. Manivannan, eds. INTERFERENCE RADIUS IN PCS RADIO RESOURCE MANAGEMENT SIMULATIONS Michael Liljenstam Rassul Ayani Department of Teleinformatics Royal Institute of Technology KTH-Electrum, S-164 40 Kista (Stockholm), SWEDEN ABSTRACT Simulation is widely used for performance analysis of Per- sonal Communication Systems due to their inherent com- plexity. However, resource demands (cpu-time and memory requirements) for this type of simulation are often high. Consequently, simulation fidelity demands must be care- fully weighed against available computer resources in mod- eling. In models including measurements of interference from other transmitters, one question that arises is up to what distance other transmitters must be included. This parameter has a direct impact on the amount of work per- formed in each interference calculation, and is also of great importance for parallel and distributed models since it influences partitioning. In this paper, the impact of the interference (or interaction) radius on fidelity and execution time is studied for a case model. 1 INTRODUCTION The number of subscribers in Personal Communication System (PCS) networks has grown rapidly over the last couple of years, and new systems are forthcoming with more bandwidth-intense services. Thus, PCS manufacturers and network operators need to tune existing networks and new designs for improved utilization of their given part of the frequency spectrum. Given the complexity of the sys- tems, simulation is widely used to study the effects of pro- posed new strategies and improvements. However, due to the often extremely resource demanding nature of these simulations, it is important to model the system as effi- ciently as possible and limit the calculation work involved whenever model fidelity permits. In this paper we study the effect of one parameter, the radius of interference calcula- tions, and its impact on model fidelity and execution time. Furthermore, if one wishes to pursue reduction in execution time and simulation of larger systems, this parameter could be important since it affects the possibility of parallelizing the model, i.e. applying Parallel Discrete Event Simulation (PDES) techniques. Parallelization of PCS models and models of related systems has attracted considerable atten- tion in the last couple of years, e.g. (Carothers et al., 1995), (Greenberg et al., 1994), (Harju et al., 1993), and (Liu et al., 1996). A Personal Communication System (PCS) provides wireless communication services to subscribers within its service area. The service area is populated by a number of Base Stations (BS), where the coverage of each BS is called a cell. Users carrying Mobile Terminals (MTs) roaming the service area may communicate through one of the BSs to reach other parts of the network or out through the fixed network. Since available radio frequency spectrum is lim- ited, frequencies used in one cell are reused in other cells that are sufficiently far away not to cause substantial inter- ference (frequency reuse). The problem of radio resource management is one important issue for good network per- formance (Jabbari et al., 1995). Zander (1996) defines the problem of Radio Resource Management as follows: for each call we should assign (i) a BS, (ii) a waveform, and (iii) a transmission power, so as to maximize the number of supported calls in the system while maintaining certain pre- defined Quality of Service criteria. Performing this assign- ment is the task of the Resource Allocation Algorithm (RAA). In a system based on Frequency Division Multiple Access (FDMA) and/or Time Division Multiple Access (TDMA), such as the GSM system, the selection of radio wave form equates to choosing a frequency and/or a time slot. In either case we term this a radio channel. In simulation studies the Signal-to-Interference Ratio (SIR) for a given radio channel is often calculated and used by either the RAA to make decisions or used as a measure to evaluate algorithms since it reflects obtainable speech quality. The size of the radius used for the interference cal- culations in the simulator thus affects the precision of the SIR value. If the algorithm makes use of this value it may in turn influence other measures such as dropping rate, han- dovers, and system load. 1629