Azar, Y., Wong, G. N., Wang, K., Mayzus, R., Schulz, J. K., Zhao, H., Gutierrez, F., Hwang, D., Rappaport, T. S., 28 GHz Propagation Measurements for Outdoor Cellular Communications Using Steerable Beam Antennas in New York City, to appear in the 2013 IEEE International Conference on Communications (ICC), June 9 ∼ 13, 2013. 28 GHz Propagation Measurements for Outdoor Cellular Communications Using Steerable Beam Antennas in New York City Yaniv Azar, George N. Wong, Kevin Wang, Rimma Mayzus, Jocelyn K. Schulz, Hang Zhao, Felix Gutierrez, Jr., DuckDong Hwang, Theodore S. Rappaport NYU WIRELESS Polytechnic Institute of New York University, Brooklyn, NY 11201 tsr@nyu.edu Abstract —The millimeter wave frequency spectrum offers un- precedented bandwidths for future broadband cellular networks. This paper presents the world’s first empirical measurements for 28 GHz outdoor cellular propagation in New York City. Measurements were made in Manhattan for three different base station locations and 75 receiver locations over distances up to 500 meters. A 400 megachip-per-second channel sounder and directional horn antennas were used to measure propagation characteristics for future mm-wave cellular systems in urban environments. This paper presents measured path loss as a function of the transmitter - receiver separation distance, the angular distribution of received power using directional 24.5 dBi antennas, and power delay profiles observed in New York City. The measured data show that a large number of resolvable multipath components exist in both non line of sight and line of sight environments, with observed multipath excess delay spreads (20 dB) as great as 1388.4 ns and 753.5 ns, respectively. The widely diverse spatial channels observed at any particular location suggest that millimeter wave mobile communication sys- tems with electrically steerable antennas could exploit resolvable multipath components to create viable links for cell sizes on the order of 200 m. Index Terms —28 GHz, millimeter wave cellular, multipath de- lay spread, path loss exponent, millimeter wave communications, RF channel, channel sounder, sliding correlator, 5G I. I NTRODUCTION Cellular and wireless local area networks (WLAN) operate between 800 MHz and 5.8 GHz, and new 60 GHz WLAN products are beginning to proliferate. The growing market for broadband wireless services has led to a global bandwidth shortage for carriers [1][2][4]. Recent work suggests that mobile cellular is possible at carrier frequencies of tens of GHz, an order of magnitude greater than today’s cellular spec- trum bands, where high-gain miniaturized steerable antennas could be used to exploit the smaller wavelength [1][2][3][4], thus motivating researchers to develop new techniques for the rarely-used millimeter wave (mm-wave) frequency bands. At the mm-wave bands of 28 GHz and 38 GHz, unlike at 60 GHz or 380 GHz, atmospheric absorption does not significantly contribute to additional path loss, making it suitable for outdoor mobile communications [1]. Advances in the semiconductor industry allow for low-cost integrated mm-wave electronics in CMOS, and cost-efficient Fig. 1. Rain attenuation in dB/km across frequency and various rates of rain fall. The rain attenuation at 28 GHz has an attenuation of 7 dB/km for very heavy rainfall of 25 mm/hr. If cell sizes are about 200 m in radius, this attenuation will reduce by 80%. The red ring donates the attenuation at 28 GHz under very heavy rainfall [7]. small high-gain steerable antennas [1][2][3][4]. However, a myth in industry is that rain attenuation will challenge mm- wave cellular systems. Zhao et al studied the relationship between rain attenuation as a function of rain rate and carrier frequency [7], as shown in Fig. 1. The attenuation may be divided by five for cell sizes with a radius of 200 m (i.e. simply convert 1 km to 200 m distance). Doing this conversion, Fig. 1 shows that at a heavy rain rate of 7.6 mm/hour [6], the rain attenuation for a 200 m cell radius is only 0.6 dB at 28 GHz and 0.8 dB at 38 GHz. Even with very heavy rainfall of 25 mm/hour, the rain attenuation is only 1.4 dB at 28 GHz and 1.6 dB at 38 GHz. Thus, proper link design (with varying gain antennas, for example) could account for rain margin in future mobile mm-wave cellular systems. To study urban cellular propagation, it is customary to classify the physical environment as being either line of sight (LOS) or non line of sight (NLOS) between a transmitter (TX) and receiver (RX). NLOS may be further divided into mod- erately and heavily obstructed environments, where moderate NLOS conditions have small obstructions, such as trees or building edges that partially block the optical path between the TX and RX, while heavily obstructed NLOS conditions have large obstructions fully blocking the optical path. Previous research in Austin, TX conducted rooftop-to- ground measurements at 38 GHz, and peer-to-peer channels in an outdoor urban setting at both 38 GHz and 60 GHz for future generation cellular [4]. Previous measurements for 28 GHz LMDS systems [8] showed that steerable antennas could be used for fixed point-to-multipoint links, whereas another study showed that NLOS links could be made using steerable antennas for cellular coverage up to 200 m [9]. This paper presents the world’s first 28 GHz outdoor cellular propagation measurements in New York City for future fifth-generation (5G) mobile communications. To describe radio propagation path loss (PL) as a function of distance, the propagation path loss exponent (PLE) is a parameter that describes the attenuation of a signal as it propagates in the channel. Path loss at a close-in reference distance of d 0 is calculated and measured to be free space loss by the equation [10]: P L f s ( d 0 ) = 20 log 10 ( 4 πd 0 λ ) (1) where λ is the wavelength of the carrier frequency. In our measurements d 0 = 5 m and λ = 10.71 mm at 28 GHz. Path loss at a distance d , beyond d 0 can be described by the path loss exponent using the following equation: P L ( d )[ dB ] = P L f s ( d 0 )[ dB ] + 10 nlog 10 ( d d 0 ) (2) where P L ( d ) is the average path loss in dB for a given TX-RX separation of d , and n is the average path loss exponent over distance and all pointing angles. LOS links with the TX and RX antennas pointing towards each other (i.e. boresight) provide a path loss exponent of n = 2 with very small multipath delay spread [5], while the steerable antennas provide different path loss exponents (and different multipath power delay profiles (PDPs)) for different links made between a TX and RX, depending on the orientation of antennas and the surrounding LOS or NLOS environment. Path loss is important for determining coverage distances, system capacity, and link budgets for viable links in a cellular system. A higher PLE indicates greater attenuation of the propagating signal. Angle of Departure (AOD) and Angle of Arrival (AOA) data at the TX and RX locations, respectively, are needed to determine the angular spread and number of unique pointing angles of the mobile and base station antennas that yield viable links between TX and RX as given in [11]. While NLOS links may require equalization due to longer propagation delay time, Signal-to-Noise Ratio (SNR) and power efficiency are maximized by finding the optimal pointing angles at a particular location. II. 28 GH Z C HANNEL M EASUREMENT H ARDWARE A 400 megachip-per-second (Mcps) sliding correlator chan- nel sounder was used to conduct the propagation measure- ments. The probing signal consists of a pseudo-random noise sequence upconverted to 28 GHz with a maximum average output power of +30 dBm before TX antenna. The sliding correlator allows a multipath time resolution of 2.3 ns and 178 dB of total path loss (for the case of TX and RX using 24.5 dBi antennas). Note that the measured dynamic range of total path loss likely meets or exceeds that of future 5G cellular systems, thus ensuring measured results are meaningful. We took great care to not record noisy PDPs in the case of insufficient SNR, and operated at a 10 dB SNR (i.e. 168 dB of total path loss). Two types of antennas were used: a 15 dBi horn antenna with 30 ◦ beamwidth for both elevation and azimuth, and a 24.5 dBi horn antenna with 10.9 ◦ beamwidth. The antennas were vertically polarized. We used the same hardware that was used in previous measurement campaigns [9] with an intermediate frequency (IF) of 5.4 GHz and a separately supplied local oscillator (LO) of 22.6 GHz. III. 28 GH Z C HANNEL M EASUREMENT P ROCEDURE The 28 GHz channel propagation measurements were per- formed at the NYU campus in downtown Manhattan. Mea- surement sites included a wide range of urban environments, including parks, commercial districts, and general university areas with high rise buildings and dense pedestrian and ve- hicular traffic. To emulate future cellular base stations with relatively low heights, two TX sites were located on the Coles Sports Center building rooftop (7 m above ground level, with the TX located on the northwest and northeast corners of the roof), and one TX site was on the five-story balcony of Kaufman Business School (17 m above ground level) (see Fig. 2). All three TX sites used the same set of 25 RX sites, which were chosen randomly based on the availability of AC power, thus yielding 75 unique TX-RX location combinations. At each RX measurement location, the TX and RX direc- tional antennas were pointed in several different directions in elevation and the RX antenna was rotated exhausively over azimuth to find the strongest received power. The strongest link was usually made when the TX and RX antennas were directly pointed at each other. The 0 ◦ azimuth angle of the TX was set at the angle with the strongest link with 10 ◦ downtilt. Measurements were then taken for 3 different TX azimuth angles , -5 ◦ , 0 ◦ , and +5 ◦ , and for 3 different RX elevations, -20 ◦ , 0 ◦ , and +20 ◦ , with all possible combinations between the two (i.e. 9 total TX-RX antenna configurations). For each of the 9 TX-RX antenna configurations, the RX antenna was rotated 360 ◦ in the azimuth plane and a power delay profile measurement was recorded at every 10 ◦ where a link was made (PL < 168 dB). In all locations, both the TX and RX used 24.5 dBi antennas with 10 ◦ 3 dB beamwidth. Fig. 2. 28 GHz cellular measurement locations in Manhattan near the NYU campus. Three base station locations (yellow stars on the two-story rooftop of Coles Recreational Center and five-story balcony of Kaufman Business School) were used to transmit to each of the 25 receiver locations within 20 to 500 m. In total, 75 TX-RX location combinations were used for Manhattan measurements. Four RX locations on the west are not shown in the picture. Purple squares represent RX sites that are blocked by buildings. Signal could only be received at 26 of the 75 location combinations. IV. 28 GH Z U RBAN C HANNEL P ROPAGATION A NALYSIS Fig. 3 shows the path loss computed for each measurement acquired in New York City with the 24.5 dBi antennas. The smallest path loss is defined as the single best TX and RX antenna pointing combination at a given RX location, which corresponds to the strongest possible link created. The best LOS PLE was n = 1.68 (due to groud reflection), with a shadowing factor (standard deviation of shadowing, or SF) of only 0.2 dB. The aggregated LOS PLE, considering all the possible antenna configurations in a LOS environment, increased to 2.55 with a SF of 8.66 dB since antennas were often not optically aligned on boresight. In contrast, the NLOS PLE was 5.76, and reduced to 4.58 when only the best (i.e. strongest) NLOS link was considered at each RX location, yielding SFs of 9.02 dB and 8.83 dB, respectively. The overall PLE for all measurements, LOS and NLOS, was 5.73. The similarity of the overall PLE and NLOS PLE is due to the NYC environment providing significantly more NLOS situations, and very few LOS locations. Fig. 4 is a polar plot that shows received power in dBm as a function of receiver azimuth angle in a NLOS environment. The three values at each azimuth angle in Fig. 4 indicate the number of resolvable multipath components, path loss (relative to 5 m free space), and RMS delay spread. The plot shows for this RX location, 22 out of 36 possible azimuth angles in an urban NLOS environment are available. Received PDPs were thresholded to ∼ 168 dB path loss floor, and multipath components were detected by using a peak detecting algorithm. Fig. 4 counters previous data that stated no more than 5-10 links could be formed in a NLOS environment [5]. Rich multipath will allow future beam steering technologies to deduce algorithmically which azimuth angle would produce the greatest received power [12][13]. Angular analysis and ray tracing of the measured PDPs allow reconstruction of the paths taken by the RF signals for each AOA. Figs. 5 and 6 show PDPs of the largest observed multipath delay spread for a LOS and a NLOS environment. Fig. 5 Fig. 3. Measured path loss values relative to 5 m free space path loss for 28 GHz outdoor cellular channels. These path loss values were measured using the 24.5 dBi narrow beam antennas. The RX antenna was rotated in the azimuth plane in 10 ◦ steps. The values in the legend represent the PLE of each environment (LOS and NLOS). shows a LOS PDP with a maximum excess delay (20 dB) of 753.5 ns transmitted from Kaufman balcony to a receiver 52 m away. The transmitter was pointed -5 ◦ azimuth and - 10 ◦ below horizon. The receiver was pointed away from the TX with 0 ◦ elevation. Fig. 6 shows the NLOS maximum excess delay (20 dB) was 1388.4 ns, indicating that the farthest distance the RF wave traveled was approximately 423 meters beyond the propagation distance of the first arriving signal. A highly reflective environment with large radar cross sections for distant reflectors is the most likely cause of such a high excess delay. For most measurements, the RMS delay spread in LOS conditions did not exceed 100 to 200 ns over all locations (6 LOS TX-RX location combinations and 20 NLOS TX-RX Fig. 4. Polar plot showing the received power at a NLOS location. This plot shows an AOA measurement at the RX on Greene and Broadway from the TX on the five-story Kaufman building (78 m TX-RX separation). The polar plot shows received power in dBm with varying receiver antenna azimuth angle. The perimeter of the plot shows for each azimuth angle three values that correspond to the number of resolvable multipath components, path loss relative to 5 m free space, and RMS delay spread. Fig. 5. The largest observed multipath excess delay in a LOS urban environment at 28 GHz. It was observed with the TX on the five-story Kaufman balcony and the RX located 52 m away from the TX. The reference path loss, maximum excess delay (20 dB), RMS delay spread ( σ τ ), and TX and RX azimuth and elevation angles are shown on the right of the PDP. location combinations) and all beam configurations. Fig. 7 shows an AOA and AOD ray-tracing analysis for the LOS case in Fig. 5. Fig. 8 shows the average number of resolvable multipath components formed for various TX-RX distances. Under LOS conditions, an average of 7.2 unique resolvable multipath components will exist for each TX-RX link with a standard deviation of 2.2 paths between 35 m and 200 m. For NLOS conditions, an average of 6.8 unique resolvable multipath components will exist for each TX-RX link with a standard deviation of 2.2 paths. For both LOS and NLOS, the average number of paths increased with distance (up to 100 m), after which the average number of paths tended to slightly decrease with increasing distance beyond 100 m. Fig. 9 shows an outage map for all Manhattan measure- ments. These RX outage sites correspond to no signal detected from TX (i.e. PL > 178 dB) at any angle. For path loss between 168 dB and 178 dB, signal could be detected but Fig. 6. The largest observed multipath excess delay in a NLOS environment at 28 GHz. It was observed with the TX on the two-story Coles building rooftop and the RX located (behind a building) 97 m away from the TX. The reference path loss, maximum excess delay (20 dB), RMS delay spread ( σ τ ), and TX and RX azimuth and elevation angles are shown on the right of the PDP. Fig. 7. AOA measurements in outdoor downtown Manhattan for 28 GHz. The RX was located in front of WWH (Warren Weaver Hall) building and the TX was located on the balcony of Kaufman. The plot shows a potential path obtained by manual ray tracing to achieve a 753.5 ns excess delay shown in Fig. 5 in a LOS environment. The green dot represents the RX and the yellow star represents the TX. White arrows represent the direction of the horn antenna. The cyan path simulates the route taken by the wave from the TX to the RX that resulted in the first received power. The purple path represents the last received power 753.5 ns later. was not strong enough to always be acquired. The map is divided into different sectors which correspond to a TX site. The radii of these sectors are 200 m, which suggests that the maximum size of future mm-wave cellular networks in dense urban environments. It was found that an outage occurred in 57% of all the receiver locations. Within 200 m, the outage decreased to 16% as no signal could be acquired at four RX locations shadowed by the Bobst building. V. C ONCLUSION This article presented propagation measurements at 28 GHz in New York City. A total of 75 unique TX-RX combinations were measured with two different antenna gains and 36 pointing angles. The rate at which path loss increased as a function of distance varied in different NLOS environments. The overall path loss exponent, n , was found to be 5.73; however, through the use of beam steering, the angles cor- responding to the lowest path loss can be found and exploited Fig. 8. Average number of resolvable multipath components per TX-RX link for different distances in 28 GHz using two 24.5 dBi horn antennas. The distribution increases up to 71 m and then decreases until 193 m. Average number of resolvable multipath components was computed over all viable links made with all possible pointing angles at a particular T-R separation. Fig. 9. Map showing all Manhattan coverage cells with radii of 200 m and their different sectors. Measurements were recorded for each of the 25 RX sites from each of the three TX sites (yellow stars).”Signal Acquired” means that signal was detected and acquired (PL < 168 dB). ”Signal Detected” means that signal was detected, but low SNR prevented data acquisition (168 dB < PL < 178 dB). ”No Signal Detected” occurred with PL > 178 dB. To the west of Kaufman is the Bobst Library which blocked the signal from reaching the 4 sites marked with crosses, which resulted in no signals being detected at those locations. to decrease the path loss exponent [12][13]. Considering, at each location, only the best angle orientation with the highest received power, the path loss exponent dropped to 4.58 for NLOS and 4.47 over all locations. While performing 10 ◦ incremental azimuthal angular sweeps, PDPs were measured with less than 168 dB path loss typically at more than 20 out of 36 possible angles under dense urban NLOS conditions. For future development of a statistical channel model at 28 GHz, we computed the average number of resolvable multipath components, averaged over all pointing angles of the TX and RX, as a function of distance. Also, we found that 57% of all receiver locations, which exceeded a TX-RX separation of 200 m, were outages, where no signal could be detected; however, the outage decreased to 16% for distances within 200 m. We could find no links at distances greater than 200 m for Manhattan with 178 dB PL. This maximal cell size means that rain attenuation will not pose a significant problem due to the relatively shorter distances involved. VI. A CKNOWLEDGMENT This project was sponsored by Samsung DMC R&D Communications Research Team (CRT) through Samsung Telecommunications America, LLC. The authors wish to thank Shu Sun and Mathew Samimi of NYU WIRELESS for their contributions to this paper, and Shadi Abu-Surra of Samsung, the NYU Administration, NYU Public Safety and NYPD for their support of our measurements. Measurements recorded under U.S. FCC Experimental License 0040-EX-ML-2012. R EFERENCES [1] Rappaport, T. S., Murdock, J. N., Gutierrez, F., “State of the Art in 60 GHz Integrated Circuits & Systems for Wireless Communications,” Proceedings of the IEEE , August 2011, vol. 99, no. 8, pp. 1390-1436. [2] Zhouyue Pi; Khan, F.;, “An Introduction to Millimeter- Wave Mobile Broadband Systems,” IEEE Communica- tions Magazine , vol.49, no.6, pp.101-107, June 2011. [3] Gutierrez, F., Agarwal, S., Parrish, K., Rappaport, T.S., “On-Chip Integrated Antenna Structures in CMOS for 60 GHz WPAN Systems,” IEEE Journal on Selected Areas in Communications , vol. 27, no. 8, pp.1367-1378, Oct. 2009. [4] Rappaport, T. S., Ben-Dor, E., Murdock, J. N., Qiao, Y., “38 GHz and 60 GHz Angle-dependent Propagation for Cellular and Peer-to-Peer Wireless Communications,” IEEE Intl. Conf. on Comm. (ICC) , Ottawa, Canada, June 10-15, 2012. [5] Rappaport, T. S., Yijun Qiao. Jonathan I, Tamir, James N. Murdock, Eshar Ben-Dor., “Cellular Broadband Millime- ter Wave Propagation and Angle of Arrival for Adaptive Beam Steering Systems,” (invited) IEEE Radio and Wire- less Symposium (RWS 2012) , Santa Clara, CA, USA, Jan. 15-19 2012. [6] Dictionary.com,“Heavy rain,” in Dictionary.com’s 21st Century Lexicon. Source location: Dictionary.com, LLC. http://dictionary.reference.com/browse/heavy%20rain. [7] Q. Zhao; J. Li; “Rain Attenuation in Millimeter Wave Ranges,” Inter. Symp. on Antennas, Propagation & EM Theory, Oct. 26-29, 2006. [8] Seidel, S.Y.; Arnold, H.W.; “Propagation Measurements at 28 GHz to Investigate the Performance of Local Multipoint Distribution Service (LMDS),” IEEE Global Telecommunications Conference, 1995. GLOBECOM ’95 , vol.1, pp.754-757, 14-16 Nov. 1995. [9] Murdock, J.N., Ben-Dor, E., Qiao, Y., Tamir, J.I., Rap- paport, T.S., “A 38 GHz Cellular Outage Study for an Urban Outdoor Campus Environment,” 2012 IEEE Wire- less Communications and Networking Conference (WCNC 2012) , Paris, France, Apr. 1-4, 2012. [10] H. Xu, V. Kukshya, T.S. Rappaport, “Spatial and Tem- poral Characteristics of 60-GHz Indoor Channels,” IEEE Journal on Selected Areas in Communications , vol. 20, no. 3, 620-630, April 2002. [11] Samimi, M., Wang, K., Azar, Y., Wong, G. N., Mayzus, R., Zhao, H., Schulz, J. K., Sun, S., Gutierrez, F., Rap- paport, T. S., “28 GHz Angle of Arrival and Angle of Departure Analysis for Outdoor Cellular Communications using Steerable Beam Antennas in New York City,” to appear in the IEEE Vehicular Technology Conference, June 2013. [12] J.C. Liberti, T.S. Rappaport, “Analysis of CDMA Cel- lular Radio Systems Employing Adaptive Antennas in Multipath Environments,” 1996 IEEE Vehicular Technol- ogy Conference , vol.2, pp.1076-1080, Apr. 28 - May. 1, 1996. [13] G. Durgin, T.S. Rappaport, “Basic Relationship Be- tween Multipath Angular Spread and Narrowband Fading in Wireless Channels,” IEEE Electronics Letters , vol.34, no.25, pp.2431-2432, 10 Dec. 1998.