The Impact of Carrier Frequency at 8 MHz and 3.5 GHz in Urban and Rural Environments Using Large Antenna Arrays Blanca Ramos Elbal, Fjolla Ademaj, Stefan Schwarz and Markus Rupp Christian Doppler Laboratory for Dependable Wireless Connectivity for the Society in Motion TU Wien, Institute of Telecommunications Gusshausstrasse 25/389, A-4 Vienna, Austria Email: {brelbal, fademaj, sschwarz, mrupp}@nt.tuwien.ac.at Abstract The widespread use of wireless mobile communications in past years has resulted in a mass use of the 2 GHz band. As a consequence, the 3.5 GHz and 8 MHz bands have emerged as an excellent alternative for mobile operators. In this paper, we investigate such alternatives using the 3GPP 3-dimensional (3D) channel model. We show the impact of both 8 MHz and 3.5 GHz carrier frequencies not only in an urban but also in a rural environment. We consider indoor and s, as well as LOS and users. We evaluate and compare the performance of each scenario in terms of SINR, throughput, RSRP, RSSI and RSRQ. Index Terms 3GPP 3D channel model, system level simulation, rural environment, urban environment, antenna array. I. INTRODUCTION Spectrum scarcity in the 2 GHz band has led to the emergence of new bands as a potential alternative for the future 5th generation of mobile communications. Among such alternatives, mmwave wireless systems have erupted on the scene with considerable force and are foreseeably a key aspect of data rate enhancement. []. One of its main benefits is the underutilization of the spectrum in the [6 ] GHz range. Further positive aspects concomitant with high frequencies are a higher spatial resolution achieved by packing hundreds of antenna elements into an antenna array. But there are also some important drawbacks associated to this frequency range such as the high sensitivity to shadowing and blockage effects, large attenuation in long distances and high penetration losses. Furthermore, some challenges remain still open [2, 3]. Moreover a good candidate to complement the congested 2 GHz band is the 3.5 GHz band, commonly known as Citizens Broadband Radio Service. Already simulations and measurements have been carried out in this band [4, 5]. However, a new carrier frequency would force operators to replace or duplicate the actual antennas at the base stations (BSs). Dealing with that issue, the authors in [6] study the impact of re-using the antennas matched to 2 GHz to convey at 3.5 GHz. In [7], the authors develop a dual band printed dipole antenna to transmit at both 3.5 GHz and 2 GHz. Another possibility that offers more coverage but less capacity is the 8 MHz band. It is also used for short-range applications, including Radio Frequency IDentification [8] or wereables devices [9], as well as for long-range applications, such the television broadcasting []. In LTE it was usually employed to cover rural areas. Nevertheless, this carrier frequency has started to be used also for urban environments. In this paper we compare the performance in terms of macroscopic Signal-to-Interference-plus-Noise Ratio (SINR), Reference Signal Received Power (RSRP), Received Signal Strength Indicator (RSSI), Reference Signal Received Quality (RSRQ) and average user throughput by means of system level simulations. We consider both Rural Macro cell (RMa) and Urban Macro cell (UMa) cells at both 3.5 GHz and 8 MHz, for outdoor as well as for indoors users. II. SYSTEM MODEL Our scenario comprises a three-sectorized BS surrounded by six interfering three-sectorized BSs as depicted in Figure. We consider a dense scenario with users deployed according to a uniform distribution. LOS/ Fig. : System model comprising the observed sector and six interfering three-sectorized BSs. We consider that the users from our cell are always regarding to the interferings BSs. For each user position, the macroscopic SINR, throughput, RSRP, RSSI and RSRQ are computed. The RSRP expression that we use follows the definition for the RSRP p given by (), where p represents the antenna port at the transmitter. For each port, the expression contains a component for line-of-sight (LOS) and non line-of-sight () propagation conditions,
α,u,p and α n,m,u,p respectively. In the 3rd Generation Partnership Project (3GPP) 3-dimensional (3D) channel model, N clusters act as scattering objects and M rays depart from each cluster. Therefore, the component in LOS depends on the antenna port at the transmitter p and the antenna element at the receiver u, and the component in depends also on the cluster n and the ray m. Additionally, the RSRP p expression includes the pathloss P L and the shadow fading SF at the user position. The expression of α,u,p and α n,m,u,p are specified in [] and it takes into account variables such as the cross polarization power ratios for each ray of each cluster, the cluster power or the receive and transmit antenna element field patterns. ( U N M RSRP p = P L SF α,u,p 2 + u= n= m= α n,m,u,p 2 ) T Xpower U () After computing the RSRP p values, the computation of the RSRP given by [2] follows (2). Furthermore, the RSSI is defined as the total received power from both attached BS and interferers I and the noise Z, as expressed in the equation (3). The RSRQ expression is given by (4), where N Nrb is the total number of resource blocks and scales up the RSRQ value to the entire bandwidth. RSRP = max{rsrp, RSRP 2,...RSRP NT X } (2) RSSI = RSRP + I RSRP i + Z (3) RSRP RSRQ = N rb (4) RSSI In our scenario, each sector is endowed with a 2D antenna array as illustrated in Figure 2, where A equidistantly spaced antenna elements are mapped to port p. The total number of antenna ports is given by N T X. Therefore, N T X different values of RSRP p are computed and following (2) the maximum value is chosen to be the RSRP. From the two polarization modes defined in [], we consider cross-polarized antennas. In this mode the first column of antenna elements is mapped to the first port with a slant angle of +45, the second column of antenna elements is mapped to the second port with a slant angle of 45, and so forth. We consider that the BS is neither electrically nor mechanically tilted and consequently the BS beam points towards the horizon and no beamforming is performed. On the other hand, at the user side we consider two cross-polarized dipoles. III. SIMULATIONS This section presents simulation results obtained with the Vienna LTE-Advanced (LTE-A) system level simulator [3 5]. In each realization, the macroscopic SINR is computed for each user as well as the throughput and RSRP values. i m={, 2,..., } ω M (θ s ) ω 2 (θ s ) ω (θ s )...... p={, 2, 3,4} λ/2 Fig. 2: Antenna array structure at BS consisting of N T X antenna ports and A antenna elements in elevation mapped to a single antenna port. The RSSI and RSRQ parameters are computed according to (3) and (4). We consider a dense scenario with 5 users per cell in order to observe the average performance of the entire cell. We evaluate both RMa and UMa scenarios at 3.5 GHz and 8 MHz. The scenario parameters for each environment are specified in Table I. The inter site distance (ISI) and BS height are larger for the RMa environment, while the average building height h is smaller. The channel parameters employed are frequency dependent and are specified in [6, Table 7.5-6]. TABLE I: Reference scenario parameters 3D-UMa 3D-RMa Frequency {8 MHz, 3.5 GHz} ISD 5 m 2 m d cell 333 m 333 m h BS 25 m 35 m h 2 m 5 m We perform 3 simulation realizations with the parameters summarized in Table II. We do not include any feedback delay. First, we consider our attached BS and interfering links to be in. Figure 3 shows the empirical cumulative distribution TABLE II: Simulation parameters Parameter Value Carrier frequency 3.5 GHz, 8 MHz LTE bandwidth MHz BS transmit power 46 dbm N Tx N Rx 4 2 BS antenna polarization XPOL User antenna gain pattern omni-directional Antenna element gain 8 dbi Receiver type zero forcing Noise power density 74 dbm/hz Noise Figure 9 db LTE transmission mode 4 Number users per cell 5 User speed 5 km/h Scheduler Round robin Simulation length TTI Number of realizations 45
function (ECDF) of the macroscopic SINR evaluated in all user positions considering indoor as well as s, and Figure 4 illustrates the RSRP of the attached BS as well as the RSRP obtained from the interferers. The plotted results in Figure 4a for s shows that the signal coming from our attached BS is stronger at 8 MHz in both RMa and UMa environments. However, the results in Figure 3 show as well a minor difference in the SINR. This is due to the fact that the interference is also higher at 8 MHz, as depicted in Figure 4b. Regarding s, the worst performance is obtained for the users in a RMa environment at 3.5 GHz. This is caused by the higher attenuation at 3.5 GHz, and also because in a RMa cell the average building height is smaller than in a UMa environment. Therefore, the maximum 3D distance from the BS to the user location is in average higher in a RMa scenario. Figure 5 provides performance results in terms of RSSI and RSRQ and Figure 6 shows the ECDF of the average throughput. Though the results in macroscopic SINR do not show huge differences between scenarios as we explained above, to compute the throughput the channel is taken into account. Hence, there are differences in average throughput for s though there are not in the SINR. As the SINR, the RSRQ take into account both desired and interfering links and as a consequence no difference are appreciated between scenarios. The highest throughput is achieved with the RMa environment..9.8.7.6.5.4.3 UMa-8 MHz RMa-8 MHz.2. -6-4 -2 2 4 6 UE SINR (db) Fig. 3: ECDF of the SINR (db) at each user position in a UMa and RMa cell at 3.5 GHz and 8 MHz when the desired link in in. Outdoor users results are plotted in solid line, while the dashed lines are related to s..9.8.7.6.5 We perform further simulations considering the desired link in LOS for all users in the cell, and all interfering links in as depicted in Figure. Figure 8 shows the SINR for both indoor and s. In this case we can see differences in macroscopic SINR for the UMa cell. That is due to the pathloss model employed [6]. In LOS the pathloss model for a UMa environment is a two slope model which results into a slightly higher value at 3.5 GHz, but the difference with the value at 8 MHz is not very large. However, in the pathloss obtained at 3.5 GHz is quite large in comparison to the value at 8 MHz. In this way, the signal coming from the attached BS and the interferers is not proportional as in Figure 3, causing the differences in SINR. Though the best RSRP is the one obtained for s in a UMa environment at 8 MHz as Figure 8a illustrates, Figure 8b shows that the interference is also large, resulting the best SINR for users at 3.5 GHz..4 Figure 9 illustrates the ECDF of the RSSI and RSRQ. In this scenario, the largest total received power is achieved in a UMa environment at 8 MHz since the contribution of both attached BS and interferers is the largest one, as shown in Figure 8. Figure shows the average throughput when we consider our attached BS to be in LOS and the interfering BSs in. In a UMa environment the SINR is very high, and consequently the Channel Quality Indicator (CQI) achieved is the best. Since we consider a Round Robin scheduler, the resources are equally assigned to all users, and practically all users achieve the maximum throughput, which.3.3 UMa-8 MHz RMa-8 MHz.2. -4-2 - -8-6 -4-2 RSRP (dbm).9.8.7.6.5.4 UMa-8 MHz RMa-8 MHz.2. -2-8 -6-4 -2 RSRP - interferers -8-6 -4-2 (dbm) Fig. 4: ECDF of the RSRP (dbm) of the desired link and interfering links in a UMa and RMa cell at 3.5 GHz and 8 MHz for both indoor and s when the attached BS is considered to be in.
.9.9.8.8.7.7.6.6.5.5.4.4.3 UMa-8 MHz RMa-8 MHz.2. - -9-8 -7-6 -5-4 -3-2 -.3 UMa-8 MHz RMa-8 MHz.2. -6-4 -2 2 UMa-8 MHz RMa-8 MHz.7 8 2 RMa cell at 3.5 GHz and 8 MHz when the desired link is in LOS. Outdoor users results are plotted in solid line, while the dashed lines are related to s..8 6 Fig. 7: ECDF of the SINR (db) at each user position in a UMa and.9 4 UE SINR (db) RSSI (dbm).9.6.8.5.7 UMa-8 MHz RMa-8 MHz.4.6.3.5.2.4..3-7 -6-5 -4-3 -2 - RSRQ(dB).2. Fig. 5: ECDF of the RSSI (dbm) and RSRQ (db) in a UMa and RMa cell at 3.5 GHz and 8 MHz for both indoor and outdoor users when the attached BS is considered to be in. -4-2.7.8-6 -4-2 2 4.8.9-8 RSRP (dbm).9 - UMa-8 MHz RMa-8 MHz.6.7.5.6.4.5.3.4.2.3. UMa-8 MHz RMa-8 MHz.2. -2.2.4.6.8.2-6 -4-2 RSRP -8.4 avg. UE throughput (Mbps) Fig. 6: ECDF of the throughput (Mbit/s) of each user in a UMa and RMa cell at 3.5 GHz and 8 MHz when the desired link is in for both indoor and s. - interferers -8-6 -4-2 (dbm) Fig. 8: ECDF of the RSRP (dbm) of the desired link and interfering links in a UMa and RMa cell at 3.5 GHz and 8 MHz for both indoor and s when the attached BS is considered to be in LOS.
.9.8.7.6.5.4.3.2. UMa-8 MHz RMa-8 MHz - -8-6 -4-2 2 4 RSSI (dbm).9.8.7.6.5.4.3.2. UMa-8 MHz RMa-8 MHz -9-8 -7-6 -5-4 -3-2 - RSRQ(dB) Fig. 9: ECDF of the RSSI (dbm) and RSRQ (db) in a UMa and RMa cell at 3.5 GHz and 8 MHz for both indoor and outdoor users when the attached BS is in LOS..9.8.7.6.5.4.3.2. UMa-8 MHz RMa-8 MHz.2.4.6.8.2.4 avg. UE throughput (Mbit/s) Fig. : ECDF of the throughput (Mbit/s) of each user in a UMa and RMa cell at 3.5 GHz and 8 MHz when the desired link is in LOS for both indoor and s. with the parameters selected for the simulation is.4 Mbit/s. Due to the lower SINR of the s in the RMa environment s shown in Figure 7, not all users reach the maximum throughput. IV. CONCLUSION In this paper we investigate the macroscopic SINR, RSRP, RSSI, RSRQ and throughput performance at 3.5 GHz and 8 MHz for outdoors and indoors users. We evaluate a BS sector surrounded by six three-sectorized BSs in a hexagonal grid, in a UMa and RMa environments. We simulate a dense scenario to observe the entire cell performance. First, we consider the desired as well as the interfering links to be in, and we observe that the best performance is achieved for s in a RMa environment at 8 MHz. Furthermore, we isolate the interference by considering the interfering links to be in but assuming that the desired link is in LOS. In this scenario the highest performance is obtained for s in a UMa environment. In terms of SINR, the users at 3.5 GHz outperforms the users at 8 MHz, however in terms of throughput in a UMa cell there are no differences since at both carrier frequencies the highest throughput is achieved. ACKNOWLEDGEMENTS The financial support by the Austrian Federal Ministry for Digital and Economic Affairs and the National Foundation for Research, Technology and Development is gratefully acknowledged. REFERENCES [] S. Sharma and B. Singh, 5G networks: The next gen evolution, in 26 International Conference on Signal Processing and Communication (ICSC), Dec 26, pp. 55 6. [2] M. Kamel, W. Hamouda, and A. Youssef, Ultra-Dense Networks: A Survey, IEEE Communications Surveys Tutorials, vol. 8, no. 4, pp. 2522 2545, Fourthquarter 26. [3] S. Schwarz and M. Rupp, Society in motion: challenges for LTE and beyond mobile communications, IEEE Communications Magazine, vol. 54, no. 5, pp. 76 83, May 26. [4] C. Wang, J. Zhang, L. Tian, M. Liu, and Y. Wu, The Spatial Evolution of Clusters in Massive MIMO Mobile Measurement at 3.5 GHz, in 27 IEEE 85th Vehicular Technology Conference (VTC Spring), June 27, pp. 6. [5] A. S. Braga, R. L. F. Lopes, S. G. C. Fraiha, J. P. L. Araujo, H. S. Gomes, J. C. Rodrigues, H. R. O. Ferreira, and G. P. S. Cavalcante, Coverage area simulation for indoor 4G networks in 2.6 GHz and 3.5 GHz, in The 8th European Conference on Antennas and Propagation (EuCAP 24), April 24, pp. 225 229. [6] B. R. Elbal, F. Ademaj, S. Schwarz, and M. Rupp, Evaluating the throughput performance at 2 GHz and
3.5 GHz in a massive MIMO system, in WSA 27; 2th International ITG Workshop on Smart Antennas, March 27, pp. 6. [7] B. H. Ahmad and H. Nornikman, Dual band printed folded dipole antenna for wireless communication at 2.4 GHz and 3.5 GHz applications, in 25 Asia-Pacific Microwave Conference (APMC), vol. 3, Dec 25, pp. 3. [8] M. Mayer, B. R. Elbal, W. Gartner, R. Langwieser, and J. Kaitovic, A flexible setup to determine RFID tag requirements for multiple-response scenarios, in 26 IEEE International Conference on RFID (RFID), May 26, pp. 4. [9] R. Nagarjun, G. George, D. Thiripurasundari, R. Poonkuzhali, and Z. C. Alex, Design of a triple band planar bow-tie antenna for wearable applications, in 23 IEEE Conference on Information Communication Technologies, April 23, pp. 85 89. [] M. Ferrante, G. Fusco, E. Restuccia, M. Celidonio, P. G. Masullo, and L. Pulcini, Experimental results on the coexistence of TV broadcasting service with LTE mobile systems in the 8 MHz band, in 24 Euro Med Telco Conference (EMTC), Nov 24, pp. 6. [] 3rd Generation Partnership Project (3GPP), Study on 3D channel model for LTE, December 27. [2] 3rd Generation Partnership Project (3GPP), LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Physical layer; Measurements, 3GPP, TS 36.24, march 28. [3] M. Rupp, S. Schwarz, and M. Taranetz, The Vienna LTE-Advanced Simulators: Up and Downlink, Link and System Level Simulation, st ed., ser. Signals and Communication Technology. Springer Singapore, 26. [4] F. Ademaj, M. Taranetz, and M. Rupp, 3GPP 3D MIMO channel model: A holistic implementation guideline for open source simulation tools, EURASIP Journal on Wireless Communications and Networking, vol. 26, no., p., 26. [5] M. Taranetz, T. Blazek, T. Kropfreiter, M. K. Müller, S. Schwarz, and M. Rupp, Runtime precoding: enabling multipoint transmission in LTE-advanced system-level simulations, IEEE Access, vol. 3, pp. 725 736, 25. [6] 3rd Generation Partnership Project (3GPP), Technical Specification Group Radio Access Network; Study on channel model for frequencies from.5 to GHz, 3GPP, TR 38.9, Dec. 27.