Abstract
In this paper, the propagation loss for 1800 MHz cellular network in a crowded market is studied and characterized using the Comit´e International des Radio-Communication, (CCIR) propagation loss model. Empirical measurement of the received signal strength in the market was conducted using CellMapper android app installed on Samsung Galaxy S4 phone. The CCIR model was configured with three different percentages of covered areas (PB). The model was optimized using the root means square error (RMSE) method and also by tuning the PB value. The un-tuned CCIR model gave an RMSE value of 9.23 dB which is above the acceptable upper limit of 6 dB for propagation loss prediction models. On the other hand, the PB-tuned CCIR model gave the best prediction result with an RMSE value of 2.177 dB and prediction accuracy of 98.11 % which is better than the performance of all the RMSE-tuned CCIR models. The results showed that apart from using an RMSE value to tune the CCIR propagation loss model, adjustment of some other key parameters of the model can as well provide a better prediction performance. However, the choice of the parameter to be tuned depends on the specific nature of the case study area.
Keywords: Propagation loss, Cellular network, CCIR model, RMSE-tuned model, Model tuning, Empirical model.
Received: 8 October 2018 / Revised: 13 November 2018 / Accepted: 19 December 2018/ Published: 14 January 2019
Contribution/ Originality
The paper's primary contribution is the development of an alternative approach for optimizing the CCIR model by adjusting the percentage of covered area rather than using the root mean square error (RMSE). The paper demonstrated that the proposed method can give better propagation prediction performance than the RMSE-based optimization approach.
1. INTRODUCTION
Propagation loss is one of the key elements that is usually required in the planning and design of cellular networks [1-7] . Generally, propagation loss models are developed and used to estimate the expected propagation loss a given wireless signal at a particular frequency will experience while propagating in a given area [8-14] . Studies have shown that no single propagation loss model can fit every situation [11, 15-17] . As such different propagation loss models have been developed over the years for different situations that take into consideration the nature of the network and the site.
In this paper, the focus is to study the propagation loss in a crowded marked using the propagation loss model developed by the CCIR (Comite' Consultatif International des Radio Communication, which is now known as International Telecommunication Union (ITU) [18-21] . The CCIR propagation model combined the effect of the free space propagation loss and the terrain-induced propagation loss. The specific focus of this paper is to derive the tuned CCIR model that will effectively predict the propagation loss in the market with the minimal prediction error and higher prediction accuracy.
2. STUDY SITE AND DATA COLLECTION
The study site is Itam market in Uyo , Akwa Ibom state Nigeria at a latitude of 5.047908 and a longitude of 7.898721, as shown in Figure 1. Though the market operates on a daily basis, it is fully occupied on its market day which comes up every 8 days. The study was conducted on a market when the market is fully crowded.
Figure-1. The Google Map Location of the study site
Source: Google maps [22]
The relevant data, namely, latitude, longitude , received signal strength and the base station information were collected using CellMapper android app which was installed on Samsung Galaxy S4 phone . The Samsung Galaxy S4 was held at about 1.5 meters above the ground and the CellMapper app was enabled to run while the phone was taken around the market at an average speed of 1.2 m/s. The CellMapper app captured and stored the listed requisite data in a comma-separated values (CSV) file. After the data collection, the CSV file was copied into the laptop and then processed for the propagation loss characterization based on CCIR model.
Table-1. The distance (d) from the base station to the mobile device and the received signal strength , (RSSI) captured within the case study site in October 2018.
d (km) |
RSSI (dB) |
d (km) |
RSSI (dB) |
0.3932 |
-77 |
0.6521 |
-85 |
0.4088 |
-78 |
0.6967 |
-84 |
0.4243 |
-77 |
0.7011 |
-85 |
0.4725 |
-81 |
0.7054 |
-85 |
0.4867 |
-85 |
0.7161 |
-86 |
0.5009 |
-81 |
0.7378 |
-87 |
0.5037 |
-83 |
0.7595 |
-86 |
0.5168 |
-82 |
0.7711 |
-86 |
0.5298 |
-84 |
0.7775 |
-88 |
0.5635 |
-82 |
0.7838 |
-89 |
0.573 |
-86 |
0.7884 |
-92 |
0.5824 |
-87 |
0.792 |
-90 |
0.6026 |
-88 |
0.7955 |
-92 |
0.6038 |
-87 |
0.8032 |
-88 |
0.6049 |
-86 |
0.8075 |
-89 |
0.6201 |
-87 |
0.8118 |
-90 |
0.6342 |
-87 |
0.8165 |
-96 |
0.6483 |
-88 |
0.8277 |
-97 |
0.6494 |
-87 |
0.8389 |
-97 |
0.6508 |
-86 |
0.8518 |
-98 |
Based on the data captured by the cell mapper, the prevailing circular network signal in the market area is the 3G network signal running at the frequency of about 18000 GHz. The key data obtained and used for the characterization of propagation loss in this paper is presented in Table 1 and Figure 2, where d is the distance from the base station to the mobile device and the RSSI is the received signal strength.
Figure-2. The graph plot of the received signal strength, (RSSI) versus distance (d) from the base station to the mobile device for the case study site in October 2018.
3. THE CCIR PROPAGATION LOSS MODEL
The Comit´e International des Radio-Communication, now ITU-R developed the CCIR propagation loss model which takes into account the varying degrees of urbanization in a given area. The model is given as Akinbolati, et al. [18]; Nnadi, et al. [20]; Oluwafemi and Femi-Jemilohun [21]:
The degree of urbanization is denoted as E which is defined in terms of (PB) the percentage of the covered area which in the case of urban area is the percentage area covered by buildings , hence
If the area is covered by about 16% buildings then E = 0.
PB ≥ 16% For urban area ; PB < 16% (typical PB =8%) for sub-urban area and PB < 16% (typical PB =3%) for sub-urban area.
Where f is frequency in MHz; d is distance in km; 150 MHz≤ f≤ 1000MHz; 30m ≤hb≤ 200m; 1m≤hm≤ 10 m and 1 km ≤ d ≤ 20km.
The prediction performance of the propagation loss models considered in this paper are assessed using RMSE given in equation 6 and prediction accuracy (PA) given in equation 7).
Where propagation loss (dB) data measured is point i,
is the CCIR predicted propagation loss (dB) for data point i and n is the total number of measured data points considered in the computation.
4. OPTIMISATION OF THE CCIR PROPAGATION LOSS MODEL
4.1. The CCIR Model is Optimised in Two Different Ways, Namely
- By using the root means square error
- By tuning (PB ) the percentage of covered area
4.2. Optimisation of the Ccir Propagation Loss Model by Using the Root Means Square Error
In this case, the root mean square error (RMSE) of equation 6 is used to tune each of the CCIR predicted propagation loss. The sum of errors (SoE) is given as;
4.3. Optimisation of the Ccir Propagation Loss Model By Tuning the Percentage of Covered Area
The tuning of PB, the percentage of covered area is conducted by using a tuning factor K such that the degree of urbanization is given as;
The value of K is repeatedly adjusted using Microsoft Excel Solver until the minimal value of RMSE is obtained. Then, the tuned value of E in equation 9 is used in equation 1 to determine the CCIR predicted propagation loss.
5. RESULTS AND DISCUSSION
The CCIR propagation loss model was configured with three different values of the percentage of covered area (PB%), namely; PB = 12 %, PB = 8% and PB = 3 %. The result of the CCIR model propagation loss prediction and performance measures are given in Table 2 for the case where PB = 12% . The un-tuned CCIR model gave An RMSE value of 9.23 dB which is above the acceptable upper limit of 6 dB for propagation loss prediction models. On the other hand, the RMSE–tuned CCIR model gave An RMSE value of 2.576 dB with a prediction accuracy of 98.70 %. The results of the un-tuned CCIR model propagation loss prediction for PB = 12 %, PB = 8 % and PB = 3 % are shown in Figure 4 while the prediction performance results of the un-tuned CCIR model and the RMSE- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 % are given in Table 3. Among the un-tuned CCIR model the one with PB = 12 % gave the best prediction with An RMSE value of 9.23 dB and prediction accuracy of 92.02% while among the RMSE- tuned CCIR model the tuned model with PB = 12 % gave the best prediction with An RMSE value of 2.771 dB and prediction accuracy of 98.10 %.
Table-2. The measured propagation loss, the un-tuned CCIR model propagation loss prediction and the RMSE-turned CCIR model propagation loss prediction for PB = 12%
d (km) |
Field Measured Propagation loss (dBm) |
CCIR Predicted Propagation loss (dBm) |
Predicted Propagation loss By The RMSE-Tuned CCIR Prediction |
d (km) |
Field Measured Propagation loss (dBm) |
CCIR Predicted Propagation loss (dBm) |
Predicted Propagation loss By The RMSE-Tuned CCIR Prediction |
0.3932 |
102.3 |
101.9397 |
101.9397 |
0.70105 |
110.3 |
110.5804 |
110.5804 |
0.40875 |
103.3 |
102.5193 |
102.5193 |
0.7054 |
110.3 |
110.6728 |
110.6728 |
0.4243 |
102.3 |
103.0772 |
103.0772 |
0.7161 |
111.3 |
110.8978 |
110.8978 |
0.4725 |
106.3 |
104.685 |
104.685 |
0.7378 |
112.3 |
111.3439 |
111.3439 |
0.4867 |
110.3 |
105.1274 |
105.1274 |
0.7595 |
111.3 |
111.777 |
111.777 |
0.5009 |
106.3 |
105.5571 |
105.5571 |
0.7711 |
111.3 |
112.0035 |
112.0035 |
0.5037 |
108.3 |
105.6404 |
105.6404 |
0.77745 |
113.3 |
112.1261 |
112.1261 |
0.51675 |
107.3 |
106.0226 |
106.0226 |
0.7838 |
114.3 |
112.2476 |
112.2476 |
0.5298 |
109.3 |
106.3953 |
106.3953 |
0.7884 |
117.3 |
112.3351 |
112.3351 |
0.5635 |
107.3 |
107.3168 |
107.3168 |
0.79195 |
115.3 |
112.4022 |
112.4022 |
0.57295 |
111.3 |
107.5653 |
107.5653 |
0.7955 |
117.3 |
112.469 |
112.469 |
0.5824 |
112.3 |
107.8097 |
107.8097 |
0.8032 |
113.3 |
112.613 |
112.613 |
0.6026 |
113.3 |
108.3192 |
108.3192 |
0.8075 |
114.3 |
112.6927 |
112.6927 |
0.60375 |
112.3 |
108.3477 |
108.3477 |
0.8118 |
115.3 |
112.7721 |
112.7721 |
0.6049 |
111.3 |
108.3761 |
108.3761 |
0.8165 |
121.3 |
112.8584 |
112.8584 |
0.6201 |
112.3 |
108.747 |
108.747 |
0.8277 |
122.3 |
113.0619 |
113.0619 |
0.6342 |
112.3 |
109.0829 |
109.0829 |
0.8389 |
122.3 |
113.2628 |
113.2628 |
0.6483 |
113.3 |
109.4115 |
109.4115 |
0.8518 |
123.3 |
113.4908 |
113.4908 |
0.6494 |
112.3 |
109.4368 |
109.4368 |
||||
0.65075 |
111.3 |
109.4679 |
109.4679 |
The Model Prediction Performance Measures |
|||
0.6521 |
110.3 |
109.4988 |
109.4988 |
RMSE |
9.23 |
2.771 |
|
0.6967 |
109.3 |
110.4874 |
110.4874 |
PA(%) |
92.02 |
98.1 |
|
Figure-3. The measured propagation loss, the un-tuned CCIR model propagation loss prediction and the RMSE-turned CCIR model propagation loss prediction for PB = 12 %.
Figure-4. The measured propagation loss, the un-tuned CCIR model propagation loss prediction for PB = 12 %, PB = 8 % and PB = 3 %
The prediction performance of the un-tuned CCIR model and the PB- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 % is given in Table 4. The graph of Figure 5 shows the measured propagation loss, the RMSE-tuned CCIR model and the PB- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 %. The three PB-tuned CCIR model gave the same result with An RMSE value of 2.177 dB and prediction accuracy of 98.11 % which is better than the performance of all the RMSE-tuned CCIR model.
In all, the results of the PB-tuned CCIR models show that the market has a effective covered area of 35.89 % which is more than the 16 % recommended in the original CCIR model for the urban area. So, for the market, the tuned the degree of urbanization is given as;
Table-3. The prediction performance of the un-tuned CCIR model and the RMSE- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 %
The Model Prediction Performance Measures |
|||
RMSE (dB) |
Prediction Accuracy, PA % |
Effective Value of PB % |
|
Un-tuned CCIR Model , PB =12% |
9.23 |
92.02 |
12% |
RMSE-Tuned CCIR Model, PB = 12% |
2.771 |
98.1 |
12% |
Un-tuned CCIR Model , PB =8% |
16.532 |
92.24 |
8% |
RMSE-Tuned CCIR Model , PB =8% |
2.782 |
98.08 |
8% |
Un-tuned CCIR Model , PB = 3 % |
27.089 |
78.73 |
3% |
RMSE-Tuned CCIR Model , PB = 3% |
2.782 |
98.08 |
3% |
Table-4.The prediction performance of the un-tuned CCIR model and the PB- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 %
The Model Prediction Performance Measures |
||||
RMSE (dB) |
Prediction Accuracy, PA % |
KPB |
Effective Value of PB % |
|
Un-tuned CCIR Model , PB =12% |
12.221 |
97.269 |
12% |
|
PB-Tuned CCIR Model, PB = 12% |
2.177 |
98.11 |
2.990905 |
35.89% |
Un-tuned CCIR Model , PB = 8 % |
16.531 |
92.247 |
8% |
|
PB-Tuned CCIR Model , PB = 8% |
2.177 |
98.11 |
4.486358 |
35.89% |
Un-tuned CCIR Model , PB = 3 % |
27.089 |
78.73 |
3% |
|
PB-Tuned CCIR Model , PB = 3% |
2.177 |
98.11 |
11.96362 |
35.89% |
Figure-5. The measured propagation loss, the RMSE-tuned CCIR model and the PB- tuned CCIR model for PB = 12 %, PB = 8 % and PB = 3 %.
6. CONCLUSION
Propagation loss for the 1800 MHz cellular network in a crowded market is studied and characterized using the CCIR propagation loss model. The model was studied under different degree of urbanisation values and also tuned using RMSE method as well as by tuning the percentage of covered area. The results showed that the market has a higher percentage of covered area than a typical urban area whose percentage covered area is given as 16 % in the original CCIR model. Furthermore, the tuning of the CCIR model using the percentage of covered area gave the best prediction performance, hence, the mathematical expression of the optimal CCIR model for the market was derived based on the percentage of covered area tuned-CCIR model.
| Funding: This study received no specific financial support. |
| Competing Interests: The authors declare that they have no competing interests. |
| Contributors/Acknowledgement: All authors contributed equally to the conception and design of the study. |
REFERENCES
[1] A. Taufique, M. Jaber, A. Imran, Z. Dawy, and E. Yaacoub, "Planning wireless cellular networks of future: Outlook, challenges and opportunities," IEEE Access, vol. 5, pp. 4821-4845, 2017.Available at: https://doi.org/10.1109/access.2017.2680318.
[2] D. González González, E. Mutafungwa, B. Haile, J. Hämäläinen, and H. Poveda, "A planning and optimization framework for ultra dense cellular deployments," Mobile Information Systems, pp. 1-17, 2017.Available at: https://doi.org/10.1155/2017/9242058.
[3] M. Jaber, Z. Dawy, N. Akl, and E. Yaacoub, "Tutorial on LTE/LTE-A cellular network dimensioning using iterative statistical analysis," IEEE Communications Surveys & Tutorials, vol. 18, pp. 1355-1383, 2016.Available at: https://doi.org/10.1109/comst.2015.2513440.
[4] R. O. ABA, "Path loss prediction for gsm mobile networks for urban Region Of Aba,South-East Nigeria. Nwalozie Gerald.Cet al," International Journal of Computer Science and Mobile Computing, vol. 3, pp. 267-280, 2014.
[5] S. Hurley, "Planning effective cellular mobile radio networks," IEEE Transactions on Vehicular Technology, vol. 51, pp. 243-253, 2002.Available at: https://doi.org/10.1109/25.994802.
[6] C. Ovon, "Models for greenfield and incremental cellular network planning," Doctoral Dissertation, University of Pittsburgh, 2012.
[7] C. Y. Lee and H. G. Kang, "Cell planning with capacity expansion in mobile communications: A tabu search approach," IEEE Transactions on Vehicular Technology, vol. 49, pp. 1678-1691, 2000.Available at: https://doi.org/10.1109/25.892573.
[8] I. Joseph and C. C. Konyeha, "Urban area path loss propagation prediction and optimisation using hata model at 800MHz," IOSR Journal of Applied Physics, vol. 3, pp. 8-18, 2013.Available at: https://doi.org/10.9790/4861-0340818.
[9] B. Jadhavar and T. Sontakke, "2.4 GHz propagation prediction models for indoor wireless communications within building," International Journal of Soft Computing and Engineering, vol. 2, pp. 108-113, 2012.
[10] M. Rani and M. Sonia, "Design & analysis of propagation models for WiMAX communication system at 3.9 GHz," International Journal of Computer Science and Mobile Computing, vol. 3, pp. 950-957, 2014.
[11] C. Phillips, D. Sicker, and D. Grunwald, "Bounding the practical error of path loss models," International Journal of Antennas & Propagation, pp. 1-21, 2012.Available at: https://doi.org/10.1155/2012/754158.
[12] S. Choudhary and D. K. Dhaka, "Path loss prediction models for wireless communication channels and its comparative analysis," International Journal of Engineering,Management and Sciences, vol. 2, pp. 38-43, 2015.
[13] U. M. Qureshi, F. K. Shaikh, Z. Aziz, S. M. Z. S. Shah, A. A. Sheikh, E. Felemban, and S. B. Qaisar, "RF path and absorption loss estimation for underwater wireless sensor networks in different water environments," Sensors, vol. 16, pp. 890-890, 2016.Available at: https://doi.org/10.3390/s16060890.
[14] I. Dove, "Analysis of radio propagation inside the human body for in-body localization purposes," Master's Thesis, University of Twente, 2014.
[15] S. Sun, T. A. Thomas, T. S. Rappaport, H. Nguyen, I. Z. Kovacs, and I. Rodriguez, "Path loss, shadow fading, and line-of-sight probability models for 5G urban macro-cellular scenarios," presented at the Globecom Workshops (GC Wkshps), 2015 IEEE, 2015.
[16] T. A. Thomas, M. Rybakowski, S. Sun, T. S. Rappaport, H. Nguyen, I. Z. Kovacs, and I. Rodriguez, "A prediction study of path loss models from 2-73.5 GHz in an urban-macro environment," presented at the 2016 IEEE 83rd Vehicular Technology Conference (VTC Spring) IEEE, 2016.
[17] A. I. Sulyman, A. Alwarafy, G. R. MacCartney, T. S. Rappaport, and A. Alsanie, "Directional radio propagation path loss models for millimeter-wave wireless networks in the 28-, 60-, and 73-GHz bands," IEEE Transactions on Wireless Communications, vol. 15, pp. 6939-6947, 2016.Available at: https://doi.org/10.1109/twc.2016.2594067.
[18] A. Akinbolati, M. O. Ajewole, and A. T. Adediji, "Path loss prediction modeling for digital terrestrial television (DTTV) in the tropical rain Forest Zone of Nigeria," Fudma Journal of Sciences, vol. 2, pp. 79-89, 2018.
[19] W. Debus and L. Axonn, "RF path loss & transmission distance calculations: Axonn, LLC Technical Paper 2006," pp. 1-12. Available at file:///C:/Users/TEMP.hp-PC.014/Downloads/path-loss-calculations.pdf. [Accessed 18th November 2018], 2006.
[20] N. C. Nnadi, I. C. Nnadi, and C. C. Nnadi, "Optimization of CCIR pathloss model using terrain roughness parameter," Mathematical and Software Engineering, vol. 3, pp. 156-163, 2017.
[21] I. Oluwafemi and O. Femi-Jemilohun, "Propagation profile and signal strength variation of VHF signal in Ekiti State Nigeria," I.J. Wireless and Microwave Technologies, vol. 3, pp. 9-24, 2017.
[22] Google maps, "Google map view of Itam market Uyo." Available at: https://www.google.com/maps/place/Itam+Market+Uyo/@5.0464166,7.8965323,17z/data=!3m1!4b1!4m5!3m4!1s0x105d565cf42bbe95:0xce3ed0b42e81c09b!8m2!3d5.0464113!4d7.898721. [Accessed 19th November 2018], 2018.
Views and opinions expressed in this article are the views and opinions of the author(s), Review of Computer Engineering Research shall not be responsible or answerable for any loss, damage or liability etc. caused in relation to/arising out of the use of the content. |