ICT-MobileSummit 2009 Conference Proceedings Paul Cunningham and Miriam Cunningham (Eds) IIMC International Information Management Corporation, 2009 ISBN: 978-1-905824-12-0 Poster Paper Cell-ID positioning in WiMAX networks Analysis of the Clearwire network in Belgium Mussa BSHARA 1, Rudi VERSCHRAEGEN 2 and Leo VAN BIESEN 1 1 Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium Tel: +32 (0)2 629.29.46, Fax: +32 (0)2 629.28.50 Email: {mbshara,lvbiesen}@vub.ac.be 2 Clearwire, Avenue Louise 326, 1050 Brussels, Belgium Tel: +32 (0)2 790.89.10 Email: rudi.verschraegen@clearwire.be Abstract: Base station identification, or cell identification (Cell-ID) is the first method used by operators to provide location-based Services (LBS). The advantage of using this method is the simplicity and the low cost implementation and processing. However, this method suffers from the low accuracy as the user will use the best (the strongest) signal available which considered to be generated by the closest BS. Due to the propagation environment and multipath components, this assumption is not always true. The transmitted signal could reach far places and the cells could become very big especially in open areas. In this paper, we study Cell-ID localization in WiMAX networks in Belgium by studying the already deployed pre-wimax network in Brussels capitol city, Aalst and Gent regions. Keywords: Cell-ID, cell size, GSM, LBS, positioning, WiMAX. 1. Introduction Cell-ID has been used by GSM operators to provide the first generation of locationbased services which did t gain much attention; thus the ability of GSM network to position a user depending on the Cell-ID remained unknown for most users. The second generation of LBS or what called the pro-active LBS is upcoming and expected to be known by all or at least most of the users. And it will be widely used as the service will automatically be trigged depending on some pre-defined spatial conditions [1]. For example, if the user comes close to a certain place (or to a certain person) he/she will receive -automatically- an alert. The new LBS generation has to be provided seemingly, free of charge and without imposing extra hardware on the user s terminal. Therefore, user positioning has to be obtained in a simple and cheap way depending on the network resources. That is, the Cell-ID is one of the best candidates and actually the first choice one can think of despite its low accuracy. The low accuracy property has been stuck to Cell-ID positioning during the last years; but is this assumption is always true? why low accuracy? because the cell size is too big; then what if the network become denser and the cells become smaller? shall the positioning accuracy improvement is enough to provide LBS especially in urban environments where the LBS are needed and used the most? Copyright c The authors www.ict-mobilesummit.eu/2009 1 of 6
2. Cell-ID positioning Cell-ID positioning is classified under proximity sensing mechanism, because the position of a terminal is deduced from the closest (the serving) base station position which is known to the terminal [2]. There are several ways to deduce the terminal position from the serving BS position; the simplest way is to consider the terminal position is the same as the BS position. But this way doesn t provide the lowest error because the probability of being at the same location or very close to the BS is not that high, and this probability goes down and becomes smaller when the cell size becomes bigger. Some other approach has been used to obtain better positioning accuracy, like the center of gravity (CoG) and Center of the circumscribing circle (CCC). 3. Analysis of the Clearwire network in Belgium At first, WiMAX networks introduced as a solution for the last mile problem with a reach of about 30 miles in rural areas [3]. But the actual deployment introduced smaller cells than in GSM networks due to the use of much lower transmission power, denser networks and clover-leaf deployment which allows having sector cells. Most of the WiMAX sites in the network under study have four cells (four base stations) or three cells [4]. The coverage area of a BS depends on its reach which depends on the transmission power and the operation environment. Figure 1 shows an example of a typical WiMAX cell (BS4 in site2). The coverage area of the BS4 is obtained by direct measurements. Figure 1: Typical WiMAX BS coverage area (Brussels environment) The coverage area is the area where the transmission of a considered BS can be received. In Cell-ID positioning we are concerned about the area where the transmission of a considered BS station is the strongest received transmission. This area is equal or smaller than the coverage area. Also, by taking into account demographic and geographic factors, higher accuracy can be achieved; for example the possibility of being on the bus stop is higher than being in the middle of the nearby lake!. Population Copyright c The authors www.ict-mobilesummit.eu/2009 2 of 6
densities, the most visited places and public places information and also application specific information (being on the road or using the public transport), can help in tuning our estimation by assigning more weight to places where the user is most likely to be. The CoG will be shifted according to these weights and our position estimate will be the new weighted CoG. Figure 2 depicts the actual cell size of a typical WiMAX BS in an urban environment. Figure 2: The actual cell size of one of the Clearwire BSs in Brussels Many measurement campaigns were conducted in the areas mentioned in figure 1 and 2. The collected data was used to obtain the first conclusions about the cell size and coverage area. But, as it is rather impossible for us (at least due to time constraint) to conduct a nation wide measurements, the only solution was to use the data provided by Clearwire. The provided data has a resolution of 20mx20m, i.e. the considered cell consists of sub-areas of 400 m 2. The sub-areas are the red triangles in figure 2. The multipath effect produced some relatively far and isolated sub-areas. If these areas are ignored, the possibility of improving the positioning accuracy is higher than adding an extra error during positioning process, because the majority of the subareas are located in one area close to the BS. The decision of excluding the far isolated sub-areas or not, can be made depending on the application and user requirements with the help of geographical and/or demographical information. However, all the sub-areas were considered in this study. 3.1 Computing the cell size for the pre-wimax network in Belgium The Clearwire pre-wimax network in Belgium is deployed in Brussels city (urban area), east of Brussels (suburban area), south of Brussels (suburban area), Gent (urban and suburban area) and Aalst (urban and suburban area).the cell size was computed in all the mentioned areas by considering the strongest received signal strength (which is the case in Cell-ID positioning).the cell size in Brussels city is found to be much smaller (for 67 % of the cells is less than 0.5 km 2 ) than the sub-urban areas. The biggest cell size (the worst accuracy) is found in the south of Brussels region which in fact could be considered as rural area more than sub-urban one. The comparison between the achieved results with the ones provided in [1] and obtained for Proximus GSM network in Belgium, shows that the pre-wimax network has smaller cells than Copyright c The authors www.ict-mobilesummit.eu/2009 3 of 6
the GSM network does. The ratio between the pre-wimax cell size and the GSM network cell size depends on the considered area. In general and considering all the areas together, the ratio is about 0.25. i.e: the pre-wimax network cell size is about 25 % of the GSM network cell size. Therefore, a higher positioning accuracy will be achieved by using Cell-ID in the pre-wimax network comparing to the studied GSM network. Figure 3 shows the obtained cell size cumulative distribution function (CDF) for each of the considered areas and figure 4 shows the obtained results for Proximus GSM network. Figure 3: Cell size analysis for pre-wimax network in Belgium Figure 4: Cell size analysis for GSM Proximus network in Belgium 4. Enhanced Cell-ID The cell size can be reduced remarkably by considering more than one cell-id, because the probability of receiving more than one BS in a certain area is much less than the probability of receiving one BS. Instead of considering only the strongest BS, let us consider also the second strongest and the third strongest BSs and calculate the new cell size. Copyright c The authors www.ict-mobilesummit.eu/2009 4 of 6
Figure 5: The enhanced cell size (blue) and the normal one (red), case 1 Figure 6: The enhanced cell size (green) and the normal one (red), case 2 Figure 7: The enhanced cell size for Brussels city Figure 5 and figure 6 show two examples of the enhanced cell-id. The red area is the cell size depending on one BS (a quite big cell was chosen ) and the blue and Copyright c The authors www.ict-mobilesummit.eu/2009 5 of 6
green areas represent the cell size using three BSs. In the two provided examples, the cell size is significantly reduced. This solution requires an already built database contains the required information about the strongest, the second strongest and the third strongest BSs for the considered area; however, this database can be built easily using radio planning tools. The cell size for Brussels city using the enhanced approach was calculated and plotted in figure 7. For 67 % of the cases the cell size is about 4900 m 2 which represent a square area of (70mx70m), which is indeed a very good accuracy for a system depends on Cell-ID positioning. 5. Conclusions Cell-ID is very interesting positioning solution for wireless operators. It is cheap and easy to implement. WiMAX networks started to be widely deployed and LBSs started to be seen in the horizon including new ones. The mobile WiMAX standard which expected to be finalized in the near future (the end of 2010) has noticed providing LBSs to WiMAX users. In this paper, we discussed cell-id positioning and provided an analysis for the pre-wimax network in Belgium operated by Clearwire. The study shows higher positioning accuracy than in GSM networks due to much smaller cells. An enhanced cell-id positioning was proposed, it reduces the cell size and -by consequence- improves the positioning accuracy in considerable values. All the used data is a realistic data obtained by measurements or provided by the operator. Therefore, realistic solutions can be implemented, and most of the known LBSs can be provided by WiMAX operators in Belgium depending on cell-id positioning. References [1] N. Deblauwe, PhD thesis: GSM-based Positioning: Techniques and Application. VUB, 2008. [2] Cello consortium report:http://www.telecom.ntua.g/cello/documents/cello-wp2- vtt-do3-007-int.pdf. [3] Local and metropolitan area networks part 16: Air interface for fixed broadband wireless access systems, 2004. [4] Bipt belgian institute for postal services and telecommunications website: http://www.sites.bipt.be. Copyright c The authors www.ict-mobilesummit.eu/2009 6 of 6