REVISITING RADIO PROPAGATION PREDICTIONS FOR A PROPOSED CELLULAR SYSTEM IN BERHAMPUR CITY Rowdra Ghatak, T.S.Ravi Kanth* and Subrat K.Dash* National Institute of Science and Technology Palur Hills, Berhampur, Orissa 761008 rowdraghatak@indiatimes.com *Final year BE students in Electronic and Communication Engineering ABSTRACT This paper presents a judicious method for planning the inclusion of radio propagation models in Berhampur. The conglomeration of variety of geographical features in the town urges the blending of empirical models. In the proposed model the area has been divided into three cells of radii 6.9 km with 60 0 sectoring. The sectors have been further analyzed by breaking into micro sectors of 10 0, this has been done in view of the demographic and service area. The empirical model COST 231[1] project has been used to give an increased RSL (Received Signal Level) at the MS (Mobile station). Case sensitive program has been written and the results generated have been used to plot the radiation pattern and scatter plots. The cellular model is based upon GSM-900 [2]. 1. INTRODUCTION The growing cellular market has left the existing system in Berhampur quite inefficient with low RSL. The proposed system architecture and the propagation prediction can benefit the service providers. Cell site selection can be based on signal coverage or on traffic. Signal coverage can be predicted by coverage prediction models and is usually applied to a start-up system. The task is to cover the whole area efficiently with the minimum number of cell sites. The main goal of all the propagation models is to show signal strength dependency on system parameters, such as frequency and antenna height, and on the geometry of built-up environment. In trying to understand and model propagation in the whole area of interest, it is first necessary to identify the features of the environment that are significant for wave propagation. Before fitting propagation models it is required to divide the area into cells, than the signal coverage area (cell area) is decided. The problem of estimating the cell area can be approached in two different ways. In the first approach, we may wish to determine the proportion of the locations where the received signal power is above a threshold value. In the second approach, we may determine the proportion of the circular area where the signal is above the threshold value. A given cell area consists of different environments, each of which contribute to signal strength loss. A given cell may consist of any one or more of the following environment [2] like the human made structures in an open area, in a sub urban area, in a dense urban area, in an urban area and in a rural area, natural terrain, flat terrain, hilly areas, forest, grasslands and water bodies. Ever increasing customers has resulted in reduced radii of cells to few kilometers. With decreasing size of cells the number of base stations has to be increased. The existing set up as shown in Fig.1 does not have any scope of economical expansion. The proposed system takes into account the growth of the urban settlement spanning around the highway, which is undergoing an expansion. The results generated from the prediction will depend upon the combination of the above factors, which make the cell geography. After the signal levels are obtained they are plotted against various parameters like angle, distance etc. 2. SYSTEM CONSIDERATIONS [2] Base station power : 40 dbm Base station antenna height Okumura-Hata (OH) : 30 meters Walfish-Ikegami (WI) : 20 meters Mobile antenna height: 2 meters Signal frequency :900 Mhz. Cell sectoring : 60 0 Mobile antenna gain :2.5 db
Base station antenna gain :10dB Cell radius : 6.923 kms. Total number of cells :03 3. OUTDOOR PROPAGATION MODEL CONSIDERATIONS [1][3][4] The following path loss models are used for the total loss calculations: Vegetation Loss Foliage Loss Plane Earth Loss Okumura-Hata Model Walfish-Ikegami Model Empirical models consider the same propagation characteristics and are generated by these assumptions. Our model has used the empirical models in a proper blend for a given region and eliminates the discrepancies in the measurement of RSL. It is so because in a given section or more preciously a 10 0 micro sector does not constitute of plane ground or vegetation or urban canyons only. A survey was made of each and every micro sector and market demography was studied to find the location of BS and the RSL was plotted. The results obtained predict that it can provide quality service to customers for the present areas and the developing areas. Amalgamation of COST 231 prediction models like Walfish- Ikegami & Okumura Hata gives reasonable values for urban environment. The intuitive approach is that, in an urban area we first incorporate OH model and then WI model. The best possible result is finally selected. In a micro sector some portion may contain building, some vegetation and some open space. Correspondingly, the different models are applied for best RSL. 4. CELL SITE SELECTION [5] The total area under consideration has been divided into three cells, each of radius 6.923 km. While Fig.1 shows the existing cellular system, Fig.2 shows how the area has been divided into cells for the proposed cellular system, each cell has been analyzed for different radial distances from the base station. Although a sixty-degree sectoring scheme has been followed, for the analysis all the cells are divided into thirty-six equal sectors i.e. ten-degree sectoring has been followed. Each ten-degree micro sector has been analyzed to calculate the received signal strength by fitting various path loss models according to the geographical conditions of the sector under consideration. While allocating the three cells it has been taken into consideration that, there might be an increase in urban area in the next ten years due to a new national highway coming up near by. Fig. 1. Existing cellular system. The intersection point of the cells is so selected because most of the mobile users are concentrated in this area and also in this way the dense area can be served by three base station transceivers. It has been proved from various experiments that the BTS antenna site can be changed up to R/4 (R= radius of the cell) from the theoretical site without affecting the radiation pattern. Hence to keep the signal strength high in the populated area the antenna site has not been kept at the middle of the cell, instead it has been placed 1.5 km inside (towards the populated region of the area under consideration shown in Fig.2 by shaded portion.). For each sector, a directional antenna is used for transmission and reception whose specifications are already mentioned in this paper. Since a sixty-degree sectoring has been followed, six antennas are required per cell, these antenna structures when fed with the signal
5.RESULTS AND ANALYSIS The received signal level purely depends upon the geographical structure of the sector, which the transmitted signal has to face while traveling from the base station the mobile unit. Hence a detail study of the area is necessary because the path loss models require details about the geometry of the various factors that effect free space transmission of the signal. All the factors are found out for this paper from detail maps and with the help of site surveys wherever necessary. Fig. 2. Proposed Cell Site Allocation. of same strength behave as an omni directional antenna as a whole. But in practical not all the antennas are not fed with the same signal strength. The signal strength is decided after the path loss studies are analyzed and the minimum signal received is found out. The signal power generally varies between 40 dbm to 43 dbm according to the requirements of that sector. The main purpose of using directional antenna is that all the cell sites do not require coverage due to absence of mobile subscribers in certain regions and also individual areas can be emphasized more due to sharing of burden between antennas. While framing this paper the area taken into consideration is although divided into three cells but analysis is done only in those sectors where the mobile subscribers are likely to be present and the antennas are also considered to be present in these sectors only. All the steps that are mentioned for allocation of cells can be done only if a detailed map (the map should be scaled so that measurements can be taken, like cell radius etc.) of the area is available which can either be obtained from the Internet or from the local authorities. Number of parameters that should be considered before deciding, where the cells will be placed and how many cells are required for the cellular service. The number of cells depend upon how large is the area and how densely the population to be served is. The number of cells should be minimized in order to reduce to cost but the grade of service should not be compromised. The following sets of figures are some of the results obtained from the path loss analysis. Fig. 3. Comparison of existing system and proposed system. Fig.3.shows a typical comparison of power levels found using both the existing and proposed cellular system. The pattern that is near to the periphery of the graph is of the proposed system while the other one is for the existing system. The radiation pattern has been found for that area of the city where most of the users are likely to stay. The results show that the proposed system is able to deliver more power than the existing system. These results also support the fact that the proposed system is fully able to accommodate future mobile users who are likely to increase in the near future due to a change in market demography as a consequence of development of urban area in the border of the city. Both the propagation models are also compared using scatter diagrams [6].
Fig.4 is the scatter plot of the city using WI while fig.5.is obtained using OH. pattern, which is nearer to the periphery of the graph, is obtained using OH while the other Fig. 4. Scatter Plot using WI. Fig. 5. Scatter plot using OH. A trend line is added to both the scatter plots using the least square method. The trend line shows the change in power level of the signal with respect to the TR separation (log. Scale). It was found out that both the scatter plots (WI&OH) gives trend lines which of almost of the same slope. Which supports the fact that both the models are quite comparable for the proposed cellular service. In addition, the minimum detectable signal level was found out to be satisfactory. There are some sectors in the system, which shows contradictory results, that is the power levels found out by using these models do not match each other. Fig.6 gives an example of such an area. This area of the cell mainly consists of an urban setup with some percentage of plane land and vegetation. The Fig. 6. Comparison of OH &WI for 5 km radius circle of cell 2. one is for WI. It shows that the signal level is quite acceptable when the modeling is based on WI but it is not so in case of OH. These types of results are obtained because the WI is highly dependent upon the urban environment details also which is not so in case of OH. Therefore in those areas where all the details like street orientation with respect to the signal etc. are present in those cases only WI is suitable to use. For obtaining the radiation plots, power levels are calculated for radial distances of 1km, 2km, 3km, and 4km&5km. But for the scatter plots 1.1 km, 2.1 km, 3.1 km, 4.1 km, 5.1 km distances are taken. The minimum distance for the scatter plot is not taken as 1 km because on the log scale log (1) becomes zero and TR separation cannot be taken as zero because signal power just below the base station antenna is zero. The results show that the signal power received at the mobile unit are highly random in nature and is a function of various parameters like signal frequency, signal power, base station antenna height, T-R separation, geographical conditions etc. The results obtained from the two models are quite comparable, but OH model is more suitable for rural and sub-urban areas because the parameters are easy to calculate due to requirement of less geographical information. The WI model is more applicable for urban and dense urban area. Fig.7 shows the irregularity in
the received signal of a particular sector (1.1) for a 5 km radius. Fig. 7. Radiation pattern of sector 1.1. The variation in received signal level at different places is mainly due to diverse geography of the sector. However, there are certain results, which shows high regularity in the radiation pattern. The following is an example of one such an area. 6.CONCLUSION We have predicted a propagation strategy for outdoor environment developed by combining empirical wave propagation models in context of the planning of mobile cellular radio network of GSM900 type. It shows that the result of area being split into for inclusion of various models has proved to give better RSL. Coverage predictions have a considerable impact in planning and deployment of cellular system. We hope this paper is going serve as reference for existing service providers who want to proliferate in to the developing areas and to the new service providers who want to enter this lucrative market. Our intention was to give a practical solution to the real problem that occurs when setting up a new cellular system by actual survey rather than using RF propagation simulation software. REFERENCES [1] E. Damosso, ed., Digital Mobile Radio: COST 231 View on the Evolution towards 3rd Generation Systems. Bruxelles: Final Report of the COST 231 Project, European Commission, 1998. [2] W.C.Y Lee, Mobile Cellular Telecommuniactions, 2 nd Edition, McGraw Hill International Editions. [3] Henry L. Bertoni, Radio Propagation for Modern Wireless Systems, Prentice Hall PTR, 2000. [4] Saleh Faruque, Cellular Mobile Systems Engineering, Artech House, 1996. Fig. 8. Comparison of WI&OH (3 km circle). In fig.7, irregularities were shown for the same sector (1.1) but fig.8.shows a uniform distribution of signal in the same sector but only up to a circle of radius 3-km. This type of pattern is obtained mainly because of uniform geography of the area and for such regions both OH and WI can be use efficiently. [5] M.D.Yacoub, Foundations of Mobile Radio Engineering, CRC Press, 1993. [6] Theodore S. Rappaport, Wireless Commiunicatons Principles and Practice, 2nd edition, Pearson Education Asia, 2002.