Path Loss Modeling Based on Field Measurements Using Deployed 3.5 GHz WiMAX Network

Transcription

1 Wireless Pers Commun (2013) 69: DOI /s Path Loss Modeling Based on Field Measurements Using Deployed 3.5 GHz WiMAX Network Yazan A. Alqudah Published online: 8 April 2012 Springer Science+Business Media, LLC Abstract WiMAX technology carries the promise of broadband access and wireless coverage. Developing countries throughout the world have been fast at adopting and employing the new technology to bridge the digital divide. The deployment of WiMAX networks enables the validation and testing of the technology. It is imperative that the technology be tested in different environments and the results shared and compared. Jordan provides a unique environment in its architecture, building construction materials, usage model, topology and vegetation. This work considers a mobile WiMAX network operating at 3.5 GHz deployed in Amman, Jordan. The work presents a new model for predicting path loss based on the results of field measurements of signals power and it compares proposed model and measured data to different propagation models. Keywords Mobile WiMAX Path loss Propagation models Curve fitting 1 Introduction MOBILE WiMAX technology addresses the ever increasing demand for high data rate by users. Countries with no Internet access infrastructure have been quick to adopt the technology to service customers and help abridge the digital divide. Jordan has witnessed an unprecedented growth in the wireless broadband deployment in the recent years. Currently, over 5 operators offer Fixed WiMAX access throughout Jordan. WiMAX was introduced in the WiMAX forum in June 2001, and air interface was standardized by IEEE [1]. The IEEE known as Fixed WiMAX, specifies a fixed deployment mainly to homes and businesses. To address mobility, the IEEE e 2005, also known as Mobile WiMAX, was published. The industry has been quick to develop Y. A. Alqudah (B) Princess Sumaya University for Technology, Amman, Jordan y.alqudah@psut.edu.jo

2 794 Y. A. Alqudah and offer products both in the network and end-user equipments. Another improvement is Scalable OFDMA which enables scaling of the Fast Fourier transform (FFT) to the channel bandwidth in order to keep the carrier spacing constant across different channel bandwidths. Accurate propagation models are essential in RF planning and network deployment. They are used to predict performance, and ensure quality of service after deployment. Different models have been proposed to predict path loss and signal strength. These models are classified as empirical (stochastic) or deterministic. Empirical models rely on measured data. Hata Okumura and COST 231 are examples of empirical models for the macro cellular environment. Deterministic models predict path loss based on physical laws governing electromagnetic wave propagation. Deterministic models are known for their accuracy but require involved computations and accurate description of the environment and its objects. Several research studies have considered the performance of WiMAX network operating at 2.5, 3.5 and 4.9 GHz [2 12]. In [2], a fixed WiMAX network deployed in Oslo (Norway) operating at 3.5 GHz was studied. The results of measured signal power at 15 locations show that path loss falls between Free Space and Cost 231 Hata Models. Another study [3] using deployed WiMAX network in Katubedaa (Sri Lanka) reported a measured signal at 3 and 5 m CPE heights. Using least square regression of measured data to compare the model, the paper report Free Space and Erceg B model as best fit for 3 and 5 m CPE heights, respectively. This paper studies path loss model by collecting a large number of signal power field measurements using a deployed WiMAX network. The measured data are compared to different currently accepted propagation models. The fact that most existing models fails to accurately predict signal power motivated a new model. The model dependence on antenna height and distance is investigated and the results are compared to other existing models to determine their accuracy. This paper is organized as follows: Sect. 2 presents the propagation models used in the WiMAX network. Section 3 provides the methodology used to take the measurements. Section 4 provides the results of the field measurements, the results of the simulation of the experiment and the comparison between the results. Finally, Sect. 5 concludes the paper. 2 Propagation Models The performance of a wireless system depends on the channel it operates in. Understanding the channel is important to ensure coverage with minimum infrastructure cost. Path loss models are used during network planning, deployment and operation to predict signal power at receiver locations. The successful deployment of any wireless network relies on ensuring serviced area is covered by minimal infrastructure. One of the main important parameters required to ensure service is the received signal power. During planning, propagation models are utilized to predict path loss and received signal power. Several models are available to predict path loss [5]. These models have been traditionally applied to frequencies below 2 GHz. With the advent of WiMAX, it was necessary to extend these models to include the new operating frequencies and conditions. This work can be considered as an effort to provide actual field measurement and comparison to further tune and validate these models. In this section, we will introduce propagation models and explain the parameters used in their calculations.

3 Path Loss Modeling Based on Field Measurements Propagation Models Free Space Loss Model (FS) The free space path loss is an analytical model that predicts the strength of the signal received when a clear line of sight path exists between transmitter and receiver. The path loss depends on the distance between transmitter and receiver and frequency and is given by PL FS = log 10 (d) + 20 log 10 ( f ), (1) where d is the distance between transmitter and receiver in Km, and f is the operating frequency in MHz. The FS model does not account for multipath propagation and cannot be used for point-multipoint radio link [5]. We will however include it here as a reference Stanford University Interim Model (SUI) The SUI is an empirical model recommended by standardizing committee. The model is an extension of Hata model with correction for frequencies above 1,900 MHz [5]. The path loss is given by [13]: PL SUI = A + 10γ log 10 ( d d 0 ) + X f + X h + s, (2) where d is the distance between transmitter and receiver, d 0 is 100 m. A is given by ( ) 4πd0 A = 20 log 10, (3) λ where λ is the wavelength. The parameter γ is path loss exponent and is given by ( ) c γ = a bh b +, (4) h b where h b is the base station antenna height in meters. The constants a, b and c depend on the terrain and are given in the Table 1. Three different terrains are defined to represents hilly and flat areas with varying vegetations. Terrain A represents a hilly terrain with moderate or dense vegetation. Terrain B represents a flat terrain with moderate or dense vegetation. Terrain C represents a flat terrain with rare vegetations. The highest path loss results from Terrain A, while the lowest results from Terrain C. Table 1 SUI parameters for different terrains Parameter Terrain A Terrain B Terrain C a b c

4 796 Y. A. Alqudah The operating frequency correction factor X f and height correction factors X h are given by ( ) f X f = 6.0 log 10, (5) 2000 ( ) 10.8log hr , for terrian A and B X h = ( ) 20.0log hr , for terrian C, (6) and h r is the receiver antenna height. Finally the term s is the shadowing factor that depends on the terrain and can take on values between 8.2 and 10.6 db ECC-33 Model (ECC) The ECC-33 also known as Hata-Okumura extended model is based on Okumura model [14]. Recently, the International Telecommunications Union (ITU) extended original model to frequencies up to 3.5 GHz [15]. The proposed path loss in the ECC model is given by where A fs is the free space attenuation (db) and is given by A bm is median path loss given by PL = A fs + A bm G b G r, (7) A fs = log 10 (d) + 20 log 10 ( f ). (8) A bm = log 10 (d) log 10 ( f ) [log 10 ( f )] 2. (9) G b is transmitter antenna height ( ) hb ( G b = log (log (d)2), (10) and G r is receiver antenna height gain factor and is given by For medium city G r = [ log 10 ( f ) ][ log 10 (h r ) ]. (11) where h r is the receiver antenna height. G r = 0.759h r 1.862, (12) Cost-231 Hata Model The COST-231 model is an extension to the Hata-Okumura model that has correction for environment. The Hata-Okumura model was developed for 500 1,550 MHz. The Cost-231 model extends Hata-Okumura model to frequency range up to 2 GHz [14,16]. The model calculates path loss for urban, suburban and rural areas [17,19].Thepathlossisgivenby PL = log 10 ( f ) log 10 (h b ) ah m + ( log 10 (h b ) ) log 10 d + c m. (13)

5 Path Loss Modeling Based on Field Measurements 797 The parameter c m depends on the environment and is equal to 0 db for suburban and 3 db for urban area. The ah m parameter for urban area is equal to For suburban and rural area, ah m is given by ah m = 3.20(log 10 (11.75h r )) (14) ah m = ( 1.11 log 10 f 0.7 ) h r (1.5log 10 f 0.8). (15) 3 Measurements The field measurements were conducted in the Alkursi suburb of Amman. The area is covered by a base station that employs mobile WiMAX. The antenna is 30 m high with three sectors. The topography covered by the base station is hilly with the base station located at summit. The area covered by our measurements as well as the locations of collected data are plotted in Fig. 1. The altitude different between highest and lowest measurement location is approximately 280 m. A CPE is connected to a PC to access the CPE management page. The page provides measurements as well as link parameters. A total of 248 locations in the coverage area are considered as shown in Fig. 1. A Global Positioning System (GPS) and Google Earth R are used to record and plot the locations and to calculate the distance between the base station and the CPE. All measurements were performed at a height equal to approximately 2 m. The CPE web page management provides the Radio Strength Signal Indicator (RSSI) in dbm as well as information about the serving base station. Measured RSSI, Carrier to interference and noise ratio (CINR) and downlink and uplink modulation are tabulated along with location coordinates. The measured received power is plotted against the transmitter receiver separation distance in Fig. 2. In addition to scattered measured data, averaged data over 50 m separation is also plotted to better visualize the data trend. The predicted signal power is also plotted using P R = P T + G T + G R L T L R + PL. (16) where G T is transmitter gain, G R is receiver gain, L T is transmitter loss, L R is received feeder loss, and PL is the propagation model path loss. The plots of path loss using the equations in Sect. 2 and the measured path loss are shown in Fig. 3. The figure clearly shows that the measured signal power exceeds that predicted by all models. Similar results were also observed in other field measurement work [2,3]. In observing the measured power levels in the field and the model prediction, it is clear that current models fail to accurately predict measurement. In the next section, a mathematical model is devised based on field measurements to more accurately predict power. 4 Proposed Model Using a standard Macrocell model to describe path loss, we express path loss as [18]: PL (db) = k 1 + k 2 log (d) + k 3 log (h b ) + k 4 log (h b ) log (d) + k 5 log (h r ) log (d) + C, (17)

6 798 Y. A. Alqudah Fig. 1 A map showing locations of field measurements and BS where k i are constants, h b is base station height, h r is receiver height, d is the distance separation between transmitter and receiver, and C is a random variable that accounts for clutter and it is expressed in terms of its statistics. PL can be simplified as where PL = α + 10γ log (d) + C, (18) α = k 1 + k 3 log (h b ) + k 5 log(h r ), (19) γ = k 2 + k 4 log(h b ). (20) The parameter γ is the path loss exponent and α is a constant that accounts for losses. Using Matlab R linear least square curve fitting to minimize root mean square error and expressing path loss as a linear equation (with log(d) as independent variable), the path loss can be expressed as

7 Path Loss Modeling Based on Field Measurements 799 Received Signal Power (dbm) Measured Free Space SUI 1 (A) COST 231 (suburban) ECC Hata 2 (medium) Distance (m) Fig. 2 Measured and predicted received power and using path loss models versus distance from transmitter Path Loss (db) Measured Free Space SUI 1 (A) COST 231 (suburban) ECC Hata 2 (medium) log 10 (d) Distance (m) Fig. 3 Measured and predicted path loss (db) versus distance (m) PL = log(d). (21) The graph of (21) is shown in Fig. 3.

8 800 Y. A. Alqudah Path Loss (db) Measured data *log 10 (d) Distanc(m) Fig. 4 Path loss versus distance for base station height = 20 m Occurance Measured-Estimated Power (db) Fig. 5 Histogram of residuals. The residuals are modeled as a Gaussian random variable In order to further investigate the path loss exponent and its dependence on base station height and distance, we seek to estimate k 2,andk 4 in (20). This is done by seeking field measurements with different base station height. The field measurements with 20 m base station height are plotted in Fig. 4. Using curve fitting, the path loss can be expressed as

9 Path Loss Modeling Based on Field Measurements 801 Table 2 Using curve fitting to express path loss models Model α + 10γ log(d) α γ RMSE FS e 14 SUI e 14 COST e 14 ECC Proposed model Path Loss (db) Free Space SUI COST ECC Proposed Distance(m) Fig. 6 Comparing FS, SUI, COST, ECC and proposed model PL = log(d). (22) Solving Eqs. (21) and(22) fork 2 and k 4, we obtain k 2 = and k 4 = Thus, γ can be expressed as γ = log(h b ). (23) The statistics of the clutter is evaluated by considering the statistics of the residuals i.e. the difference between measured and predicted values of path loss. The graph in Fig. 5 shows the histogram of residuals. The clutter is modeled as a zero mean Gaussian random variable with a standard deviation equals db. Having expressed the measured data analytically, it is possible to compare the proposed model to other models. To do so, we use curve fitting to express Free Space, SUI, COST and ECC models using α + 10γ log(d).table2 summarizes the values of α and γ and a measure of the goodness of fitting. The plot in Fig. 6 shows the graphs of the proposed model as well as other models using the values in Table 2. With the aid of Fig. 6 and Table 2, it is possible to compare the accuracy of models in predicting path loss. The ECC model has the closest path loss exponent and overall performance.

10 802 Y. A. Alqudah COST is farthest from proposed model based on measurements. The SUI model diverges as distance increases. Overall, the models are pessimistic in their prediction with over 18 db difference at distances greater than 1 km. 5Conclusion This work considers the propagation modeling for a WiMAX network based on actual field measurements. The deployed network operates at 3.5 GHz. Measured signal strengths are collected in the base station coverage area. The measured values are compared against well accepted propagation models. This work introduces a new model for predicting path loss. The collected data are used to devise a propagation model based on curve fitting. The path loss constant and exponent are calculated from the linear modeling of path loss. The random variability in the measurements caused by clutter is used to estimate the variance of the clutter. References 1. IEEE Standard (2004). IEEE standard for local and metropolitan area networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems. 2. Grønsund, P., Grøndalen, O., Breivik, T., & Engelstad, P. (2007). Fixed WiMAX field trial measurements and the derivation of a path loss model, M-CSC. 3. Amarasinghe, K. C., Peiris, K. G. A. B., Thelisinghe, L. A. D. M. D., Warnakulasuriya, G. M., & Samarasinghe, A. T. L. K. (2009). Comparison of propagation models for Fixed WiMAX system based on IEEE , ICIIS. 4. Alqudah, Y. A., & Tahat, A. (2011). Path loss and propagation models at 3.5GHz using deployed WiMAX network. ICOIN. 5. Milanovic, J., et al. (2007). Comparison of propagation models accuracy for WiMAX on 3.5GHz. In 14th IEEE international conference on electronics, circuits and systems, 2007 (ICECS). 6. Imperatore, P., Salvadori, E., & Chlamtac, I. (2007). Path loss measurements at 3.5 GHz: A trial test WiMAX based in rural environment. In Proceeding of the 3rd international conference on testbeds and research infrastructures for the development of networks and communities. 7. Moraes, E., et al. (2009). WiMAX near LOS measurements and comparison with propagation models. In 3rd European conference on antennas and propagation, 2009 (EuCAP 2009), March 2009, pp Sarkar, T. K., et al. (2003). A survey of various propagation models for mobile communications. IEEE Antennas and Propagation Magazine, 45(3), Chehri, A., Fortier, P., & Tardif, P. M. (2010). Characterization of the ultra-wideband channel in confined environments with diffracting rough surfaces. Wireless Personal Communications, 62, Ahmed, B., Masa Campos, J., Lalueza Mayordomo, J. (2011). Propagation path loss and materials insertion loss in indoor environment at WiMAX band of 3.3 to 3.6 GHz. Wireless Personal Communications, doi: /s Sebastião, P., Velez, F. J., Costa, R., Robalo, D., & Rodrigues, A. (2010). Planning and deployment of WiMAX networks. Wireless Personal Communications, 55, Torres, R. P., Cobo, B., Mavares, D., Medina, F., & Loredo, S., et al. (2006). Measurement and statistical analysis of the temporal variations of a fixed wireless link at 3.5 GHz. Wireless Personal Communications, 37(1 2), Erceg, V., et al. (January 2001). Channel models for fixed wireless application, Tech. rep. IEEE Broadband Wireless Access Working Group. 14. Okumura, Y., Ohmori, E., Kawano, T., & Fukuda, K. (1968). Field strength and its variability in VHF and UHF land mobile radio services. Rev. Elec. Comm. Lab. 16, Electronic Communication Committee (ECC) with the European Conference of Postal and Telecommunication Administration (CEPT), The analysis of the coexistence of FWA cell in the GHz band, tech. rep., ECC Report 33, May 2003.

11 Path Loss Modeling Based on Field Measurements Hata, M. (1980). Empirical formula for propagation loss in land mobile radio services. IEEE Transactions on Vehicular Technology, VT-29, COST Action 231 (1999). Digital mobile radio towards future generation systems, final report, tech rep., European Communities, EUR Rimac-Drlje, S., et al. (2009). Receiving power level prediction for WiMAX systems on 3.5GHz. WCNC. 19. Shahjahan, M., & Abdulla Hes-Shafi, A. Q. (2009). Analysis of propagation models for WiMAX at 3.5GHz, M.S. thesis, Blekinge Institute of Technology, Karlskrona, Sweden. Author Biography Yazan A. Alqudah received his PhD degree from The Pennsylvania State University, USA in 2003 in electrical engineering. He joined Intel Corporation, Oregon in 2003 as senior technologist where he worked with Logic technology development (LTD) and Mobile wireless groups (MWG). In his capacity, he led the development of LTD yield analysis system and the development and integration of WiMAX technology. Dr. Alqudah received three Intel s recognition awards in 2005 and 2007 for his successful efforts. Since 2008, he has been with the communication engineering department at Princess Sumaya University for Technology as an assistant professor. Dr. Alqudah is a senior member of IEEE and is WiMAX RF certificated. His current research interests include broadband optical wireless communication, WiMAX deployment and performance, software architecture and development.