A Simple Algorithm To Distribute Optimally a Number of External Base Stations in a Microcellular Environment zyx Ramon Agusti Comes, Xavier Monfort Querol, Jordi Perez Romero Department of Signal Theory and Communications Technical University of Catalonia zyxw (UPC) C/ Gran Capita, s/n Campus Nord-Edifici D4 08034 Barcelona-Spain Corresponding author: Ramon Agusti Comes Phone:34-3-401 67 81 Fax: 34-3-401 72 00 E-mail: ramon@xaloc. upc.es, zyxwv -i ordip@xaloc. upc. es ABSTRACT A simple algorithm is proposed for the optimal distribution of a number of base stations in a microcellular environment. Criteria are defined in order to decide the combinations of antennas that should be tested, and a quality factor is proposed to compare them. Results are presented for a specXic environment. L INTRODUCTION One of the most important stages in the design of a mobile communications system consists in deciding a proper location for the base stations. There are many tools available to calculate the coverage of an antenna according to a propagation model, but it is also necessary to have some kind of tool making it possible to deal with tlus problem of location and so avoid a haphazard process of trial and error. However, little effort has been devoted to this issue in the literature published to date. The objective of this paper is to present a simple and fast algorithm that gives an initial approach to an optimal base station dstribution that can be used as a starting point in the process of obtaining a good real distribution for a microcellular outdoor scenario consisting of a regular distribution of buildings into parallel and perpendicular streets with antennas located below roof-top level, monopole antennas being the type to be used. zyxwvuts II. DEFINING AN OPTIMAL DISTRIBUTION OF BASE STATIONS In order to find the best distribution of base stations, given the number N of antennas to use and the characteristics of the environment, it is necessary to try different combinations of N positions of the antennas and compare them, zyxwvutsr so the first step is to define a quality parameter that allows tlus comparison process and which must be easy and fast to calculate. A first possible criterion for deciding whether a distribution of base stations is better than another is to compare them according to the power level received at the Merent points: a higher level means that the combination is more suitable. One way to measure this characteristic is to calculate the average power level of the different points considered. Another important aspect to focus on is the sensitivity level S defined by the system, in other words, the minimum power level needed by the receiver to work properly. As a result, we must penalize the quality factor of a distribution if it has a large number of points below t h s level. For example, in the DECT system, for whch the application was first designed, this level is -83 Bm, although a margin to prevent possible fading should also be considered, thus making this level higher. The above discussion leads us to define the quality factor as Q = P,Q+ P) where points considered, assumed to be negative, and is the average power level in dl3m received in all the Npoints_below-sensiti~~-level (2) zy P= “tal-point s Another issue to consider is the importance of reducing the number of tested combinations by assuming that many of them may be discarded without any test (for example those combinations with all the antennas placed together). The following aspects concerning this problem should be considered: 1 .- Reducing the overlapping coverage area between base stations. This can be reached by avoiding as much as possible those combinations where two or more antennas are in a LOS (Line Of Sight) situation. 0-7803-4872-9/98/$10.00 zyxwvut 0 1998 IEEE 1014