Performance Enhancement for Microcell Planning Using Simple Genetic Algorithm

Transcription

1 Performance Enhancement for Microcell Planning Using Simple Genetic Algorithm Hsin-Piao Lin, Ding-Bing Lin, Rong-Terng Juang Institute of Computer, Communication and Control, National Taipei University of Technology, Taipei, Taiwan, R.O.C., EXT. 2248, Abstrate-Microcell planning is quite difference from the cell planning used for macrocells. For achieving the efficient cell planning, it is necessary to understanding the propagation characteristics of microcell systems. In this paper, we present the propagation measurement results at the campus area located at urban Taipei city. We verify Walfisch-Ikegami model is an accurate model for predicting path loss in microcell systems. Based on Walfisch-Ikegami model and digitized building information, cell planning using simple genetic algorithm is presented. The optimum solution for base station numbers, locations, antennas heights, and transmitting power can be achieved. On other hand, due to capacity demand and coverage requirement., the existing system must be enhanced. We also present the simulation results for system enhancement using simple genetic algorithm with the help of digitized buiding information and Walfisch-Ikegami model. 1. Introduction The demand for higher capacity in mobile communication system leads the operating companies to adopt strategies based on microcell planning [1,2]. It is important and necessary to develop an accurate prediction model for reducing procedure and improving efficiency of cell planning in microcell environment. The Walfisch-Ikegami propagation model [2, 3] has been proposed for predicting path loss in an urban area with smaller cells. In this paper, we carried out a series of propagation measurements at the campus of NTUT (National Taipei University of Technology) to observe the propagation characteristics due to different antenna heights. And then, we extract the model parameters for Walfisch-Ikegami model by comparing the results between simulation and measurements to verify the accuracy of Walfisch-Ikegami model. Further more, using Walfisch-Ikegami model and with the help of the digitized building information, we can analyze the relationship between coverage and economy efficiency based on simple genetic algorithm [4]. The proposed method can find out how many sectors are needed to substitute for omni-directional antenna base stations while system migration from macrocell to microcell. Besides, the better locations and their transmitting power of these sectors

2 can be determined. Thus, system modification or re-planning will be achieved easily. 2. Comparison Between Measurement And Prediction Model The measurement system is much like those ones in [5, 6]. The base station antenna was located at 1st floor to the 6th floor in the building of EE department. The transmitted power is 30 dbmw operating at MHz. During the measurement, the received power and position, latitude and longitude, were recorded. Fig. 1 shows a typical result of our measurements. Fig. 1. The received signal level along the selected route when the base station antenna was mounted on the 1 st (upper), 3 rd (center), and 5 th floor (lower). The digital maps with databases of building information, such as average building height, average road width, and environment type, are used to extract model parameters for Walfisch-Ikegami model. A typical case of comparison between prediction and measurement results is shown in Fig. 2. In that case, the base station antenna was mounted on the first floor. The prediction model simulated the propagation behavior much similar to the measured data for the whole route. Table I shows the root mean square (RMS) errors between simulated and measured data. The minimum RMS error is 3.9 db at 1 st floor, while the maximum error is 9.1 db at the 4 rd floor. As the results shown in Table I, the Walfisch-Ikegami model can be used for coverage prediction in microcells planning.

3 3. Simple Genetic Algorithm A genetic algorithm is a population-based algorithm. A population is a set of individuals that represent candidate solutions for a given problem. The genetic algorithm here is much like the one used in [4]. Based on standard operations as selection, crossover and mutation, it will optimize the base station locations and transmitting power based on the fitness evaluation. In the following, we briefly describe the main parameters used in our method. The location and transmitting power were encoded in DNA string. The bit strings we consider represent the whole set of possible BS locations. The different value of set entries represent different transmitting power. The fitness function is as flower: where CovergeRate is the percentage of locations covered by the selected BSs and Overlapping is the percentage of locations covered by 2 BSs or more. CoverageThreshold is the desired coverage rate. According to the fitness function, the fitness volume of every individual could be evaluated and sorted in order. The higher 50% of individuals were survived and reproduced offspring by uniform crossover. The final operation, mutation, was assigned to a constant mutation rate of 0.3. In the analysis there were 1500 individuals in a generation and iterating stop at 3000 th generation. 4. Simulation Result Fig. 3 shows an example of microcell planning in an area of 14km x 15km digitized building map, on which the optimized cell planning was achieved by the proposed method. The average building height is 20.5 meter and the standard deviation is 15.1 meter. There are 7 sectors used and each has transmitting power of 2.3W. The dotted area is the area covered by the active BSs and the coverage is 90%.

4 Fig. 3. An optimum solution achieved by the proposed approach Fig. 4 shows an example of system enhancement. The original macrocell system has 2 omni-directional antenna BSs with the transmitting power 5w for each, and total coverage is 81%. Using the proposed approach, we can achieve the system enhancement by adding 3 sectors with 1W transmitting power. The coverage is achieved to 99% by 18% increasing, while the transmitting power of the 2 omni-directional antennas can be reduced to 1 W for each.

5 Fig. 4. A System enhancement example with 2 original omni-directional antennas ( ) and 3 added sector antennas ( ). 5. Conclusions Microcell planning is a complex procession. To achieve the optimization of cells planning, knowing the characteristics of prerequisite EM wave propagation is a prerequisite. We have presented the propagation measurement of different base station antenna heights at the campus of NTUT in the urban Taipei city. We found that the errors between measured data and prediction using Walfisch-Ikegami model is quit less, and verified that using this model for microcell coverage prediction is accurate and reliable in our selected area. Besides, optimization method for cell planning and system enhancement using simple genetic algorithm with Walfisch-Ikegami model and digitized building information are developed. Thus, an easy and efficient cell planning and system enhancement method for microcell systems can be delivered. References [1] M. Madfors, K. Wallstedt et al., High capacity with limited spectrum in cellular

6 systems, IEEE Communication Mag., vol. 35, no. 8, pp , [2] S.R. Saunders, Antennas and propagation for wireless communication systems, John Wiley & Sons, pp , [3] D.Har, A.M. Watson, and A.G. Chadeny, Comment on diffraction loss of rooftop-to-street in cost 231 Walfisch-Ikegami model, IEEE Trans. Veh. Tech., vol. 48, pp , [4] Calegari P., Guidec F., Kuonen P., and Wagner D., Genetic Approach to Radio Network Optimization for Mobile Systems, IEEE Veh. Tech. Conf., vol. 2, pp , [5] G.L. Siqueira, E.A. Vasquez, R.A. Gomes, and C.B. Samprio, Comparison between propagation measurement and coverage prediction models for less urban cells, IEEE Telecom. Conf., vol.2 pp , [6] G.L. Siqueira, E.A. Vasquez, C.B. Sampaio, M.A. Socorro, Optimization of base station antenna position based on propagation measurements on dense urban microcells, IEEE Veh. Tech. Conf., pp , 1997.