Lighting Up a Hotel with mmWave FWA
By Brett Mills-Meiner, Head of Product for Cloud Software
How to Make mmWave Operationally Feasible illustrated the accuracy of Pivotal’s RF planning tool, WaveScape™, in predicting indoor FWA coverage at an apartment building in Atlanta, and how that coverage was expanded with Pivot 5G repeaters. This blog describes another WaveScape analysis, this time for a hotel in Bellevue, WA, and how that coverage was expanded with one Pivot 5G repeater.
As noted in the previous blog about an apartment building in Atlanta, the high-resolution data (15cm) used by WaveScape was required to meaningfully predict adequate coverage and throughput, both of which can vary dramatically among living units, even those adjacent to each other. mmWave characterization depends on precise measurements of foliage, angle of incidence, and line of sight, all of which require high-resolution geodata to accurately quantify. Predicting coverage among guest rooms in Bellevue is no different.
Figure 1, borrowed from the blog, Cyclops Knows Where Your Signal is Coming From, shows the 240-guest room, four-sided hotel acquiring mmWave coverage from an existing gNB. Two sides of the hotel acquire mmWave signal from the gNB alone — one via line-of-sight (LOS), and the other via reflection. The third side is covered by a Pivot repeater which can be turned on or off. The fourth side, in the upper left of Figure 1, receives no mmWave coverage from the gNB. The hotel glass penetration loss is 10dB, which resulted in better coverage and performance via reflection than the reflection in an earlier blog, Case Study: A Deterministic Approach to mmWave FWA Qualification. For that MDU, the glass loss for out-to-in penetration was 35dB.
Like the Atlanta MDU analysis, this one combined floor-by-floor signal penetration analysis with readily available floor plans showing every guest room. WaveScape leverages AI to align the room locations, building outlines, clutter, and signal strength predictions so that they are aligned with the physical X, Y, and Z locations in the real world.
Figures 2 through 5 show indoor coverage by guest room, windows closed, from the gNB directly and from the reflection. No guest rooms occupy the first floor of the hotel. Guest rooms colored in Green qualified for FWA service. Those colored in Red were disqualified. WaveScape qualifies individual rooms for FWA service if mmWave can establish an FWA link (>-97dBm RSRP indoors) along >50% of the face of the exterior of the room.
WaveScape predicted that 67 guest rooms served by the gNB alone would meet the signal and throughput requirements to qualify for FWA service. Predictions were validated by on-site measurements. Besides signal predictions (see Figures 2 to 5), WaveScape predicted qualified downlink throughput, by room, ranging from 300 Mbps to beyond 1,700 Mbps.
Reflections
The south side of the hotel acquired coverage not from the gNB directly but by a reflection from the office building shown in the lower right corner of Figure 6. Guest room coverage on this side of the building varies by floor, either red or green, in Figures 2 to 5. This variability is caused by the foliage between the office building and the hotel. One can see how the two green reflection paths utilizing WaveScape’s RayViewer™ feature in Figure 6, which illustrates how the signal reached the building through two “keyholes” in the foliage. The second floor coverage shown in Figure 2 is most distinctly affected; coverage is qualified in rooms 206 to 210, corresponding to the left keyhole, and in rooms 250 to 253, corresponding to the right. All remaining second floor rooms were disqualified.
Repeater Expands Qualified Coverage
WaveScape predicted that another 60 rooms would qualify for FWA service by redirecting the gNB signal to the north side of the building using the Pivot 5G repeater shown in Figure 1. These predictions were borne out by on-site measurements. Figures 7 to 10 show affected rooms, by floor, change in color from Red to Green. Inset images in the upper left correspond to the results in Figures 2 to 5 when the Pivot was turned off.
Put to the Test
To test WaveScape’s predictions, Pivotal was given access to the 25 vacant units in the hotel to conduct on-site measurements of RSRP, throughput, and other metrics. Table 1 compares WaveScape RSRP predictions to on-site RSRP measurements made with an off-the-shelf consumer CPE. WaveScape predictions were within 10dB of the measured values for 21/25 measurements, and WaveScape room qualifications (which account for other factors like angle of incidence and line-of-sight status) were correct in 22/25 units.
Auditing Mis-Qualified Units
A few predictions, even after on-site measurements, were hard to explain. One of them was room 423. WaveScape predicted inadequate RSRP to room 423 on the north side of building when, not surprisingly, the Pivot was turned off. But surprisingly, on-site measurements inside 423 showed adequate RSRP, enough to generate 616 Mbps of downlink throughput. Uplink throughput, at 28 Mbps, still disqualified the room, and 616 Mbps must have been generated by a reflection, but from where? Field techs used Pivotal’s RF scanner, Cyclops, to identify the reflection source as nondurable — a guest vehicle parked in the north lot, Figure 11. Only when the parking space was occupied, and when the Pivot repeater was turned off, did the mysterious reflection appear, resulting in adequate RSRP and downlink throughput.
RF network performance will always have a statistical element to it, but WaveScape makes conservative assumptions and only considers reliable reflectors in its FWA qualification predictions. If WaveScape says a hotel guest room or residential unit is qualified for FWA, then carriers can be confident that the subscriber will receive a solid mmWave FWA link. Table 2 provides a summary of the points with significant variance between prediction and reality.
| Unit / Pivot | Pivot Status | Prediction vs. Reality | Audit Findings |
|---|---|---|---|
| 423 | Off | Underpredict 11.5dB, WaveScape False Negative | Signal was coming from an intermittent reflection, so even though measurements supported FWA, service should not be extended to this unit |
| 218 | On | Underpredict 10dB WaveScape correctly qualified for FWA | Pivot provides stronger side-lobe performance than expected, variation did not cause difference in prediction status. |
| 232 | On | RSRP measurement within .1 dB, WaveScape disqualified unit due to angle of incidence issues | Side-lobe performance of CPE created more-robust link than expected. Correct RSRP prediction means the unit could be qualified with less bias towards reliability. |
| 302 | N/A | Underpredict 16.8 dB, WaveScape correctly qualified for FWA | Cyclops determined signal traveled through sparse area of tree- WaveScape correctly determined this link should be qualified anyways though |
| 454 | N/A | Underpredicted 11.7dB, WaveScape False Negative | Cyclops determined signal bounced off a AC unit on top of the reflecting building, which WaveScape deemed not reliable enough |
Table 2: Significant variance between prediction and reality among these hotel rooms
Notably, Pivotal’s audit of this building’s coverage found no ‘false positives’ and very few ‘overpredictions’ - which can be attributed to WaveScape’s bias towards reliability when qualifying units for FWA.
Conclusion
Coverage and throughput for a hotel, like any MDU, can vary dramatically from room to room, even those close to each other. Carriers need unit-level FWA qualification to make mmWave FWA operationally feasible. As a high resolution, cloud-based modelling tool for mmWave FWA address qualification, WaveScape offers a unique solution that makes conservative but realistic unit qualification predictions, thus allowing carriers to make full use of their mmWave spectrum. Pivotal extends its gratitude to Embassy Suites for renting one of its suites to serve as Pivotal’s 5G Test Lab, and for granting Pivotal quick access to its guest rooms to conduct on-site measurements.