NetSim NTN, Verified Against 3GPP TR 38.821

Set 1 · LEO-600 and LEO-1200 · S-band · Handheld UE

In Release 16, the companies in 3GPP RAN1 ran common calibration cases to align their NTN system-level simulators. We ran four of those cases in NetSim. The NetSim CDFs of coupling loss and Geometry SINR lie on the curves that 11 companies reported to 3GPP.

Cases SC9, SC10, SC14, SC15 Coupling loss Geometry SINR R1-1913404 company curves
SC10 · beam layout
SC10 beam layout: 19 metric beams surrounded by 108 outer interference beams, with 190 UE positions

The six reference scenarios in TR 38.821

TR 38.821 defines six reference scenarios by orbit, payload type and beam behaviour (Table 4.2-1 of the report).

Transparent satellite Regenerative satellite
GEO based non-terrestrial access network Scenario A Scenario B
LEO based non-terrestrial access network: steerable beams Scenario C1 Scenario D1
LEO based non-terrestrial access network: the beams move with the satellite Scenario C2 Scenario D2

The calibration assumptions of Table 6.1.1.1-5 apply to scenarios A, C2 and D2. The four cases on this page are LEO cases. NetSim runs them with a transparent payload, which is Scenario C2.

What we compare against

The 3GPP Release 16 study on NR over non-terrestrial networks, and the calibration data behind it.

The study

3GPP studied NR over non-terrestrial networks in Release 16. TR 38.821 reports the study.

The common cases

Table 6.1.1.1-9 of TR 38.821 lists 30 calibration study cases. The companies ran them with the assumptions of Table 6.1.1.1-5, to align their system-level simulators.

The data

R1-1913404 (Thales, RAN1#99) collects the downlink CDFs of each company. Table 6.1.1.2-1 of TR 38.821 reports the average of these results.

Companies with downlink results in R1-1913404

  • Thales
  • ESA
  • Ericsson
  • ZTE
  • Panasonic
  • Huawei
  • MediaTek
  • Nokia
  • Nomor
  • Samsung
  • Sony
  • NetSim (this comparison)
Source Title Used here for
TR 38.821, Table 6.1.1.1-1 Set-1 satellite parameters for system level simulator calibration EIRP density, maximum gain and beamwidth per orbit
TR 38.821, Table 6.1.1.1-3 UE characteristics for system level simulations Handheld UE: 23 dBm, 0 dBi, 7 dB noise figure, 290 K
TR 38.821, Table 6.1.1.1-5 System Level Simulation assumptions for calibration Bandwidth, re-use options, channel, UE distribution, metrics
TR 38.821, Table 6.1.1.1-9 List of calibration study cases Definitions of cases 9, 10, 14 and 15
TR 38.821, Table 6.1.1.2-1 Calibration results on DL transmissions Reference 5th, 50th and 95th percentiles
R1-1913404 System Level Calibration Results for NTN on DL transmissions The CDF curve of each company

Four LEO cases: S-band, handheld UE, downlink

The cases cover two altitudes and two frequency re-use factors. All four use satellite parameter Set 1.

Case Satellite orbit Central beam elevation Terminal Frequency band Frequency re-use Beams in NetSim
SC9LEO-60090°HandheldS-bandOption 1: re-use factor 161
SC10LEO-60090°HandheldS-bandOption 2: re-use factor 3127
SC14LEO-120090°HandheldS-bandOption 1: re-use factor 161
SC15LEO-120090°HandheldS-bandOption 2: re-use factor 3127

Case definitions: TR 38.821 Table 6.1.1.1-9. Re-use options: Table 6.1.1.1-5. Each case has 19 metric beams with 10 UEs per beam, so each CDF contains 190 UEs.

19 metric beams inside an outer interference field

TR 38.821 takes its statistics from the inner 19 beams and requires the surrounding beams to be simulated independently.

SC10 geometry: 19 blue metric beams at the centre, 108 orange outer interference beams around them, and 190 black UE positions inside the metric beams

Figure 1: SC10 downlink geometry. Axes show distance from the central beam in km.

  • Metric beams. The 19 blue cells. Only UEs in these cells enter the CDFs.
  • Outer interference beams. The orange cells. They transmit on the same channel and supply the surrounding interference.
  • UE positions. The black dots. NetSim places 10 UEs uniformly inside the Voronoi cell of each metric beam.
Re-use factorAdditional tiersMetricOuterTotal
1 (SC9, SC14)2194261
3 (SC10, SC15)419108127

The tier counts follow TR 38.821 Section 6.1.1.1: two additional tiers for FRF = 1, and four additional tiers for FRF > 1.

The SC10 scenario in the NetSim GUI on a map: stars mark the 127 beam centres, a red outline marks the 19 metric beams, and red dots mark the 190 UEs

Figure 2: The same SC10 scenario in NetSim. Stars: beam centres. Red outline: metric beams. Red dots: UEs.

NetSim configuration

One configuration serves all four cases. Only the orbit and the re-use factor change.

Parameter NetSim setting 3GPP reference
SatelliteOne transparent LEO satellite per case. 600 km for SC9 and SC10; 1200 km for SC14 and SC15.Table 6.1.1.1-9
Satellite EIRP density34 dBW/MHz for LEO-600; 40 dBW/MHz for LEO-1200. Applied per beam.Table 6.1.1.1-1
Satellite Tx max gain30 dBiTable 6.1.1.1-1
Satellite antenna patternBessel functionTR 38.811 Section 6.4.1
CarrierS-band, NetSim band n256, 1985–2015 MHzTable 6.1.1.1-5: S-band (i.e. 2 GHz)
Bandwidth30 MHz, 160 PRBs, 15 kHz subcarrier spacing (μ = 0)Table 6.1.1.1-5: DL 30 MHz
Frequency re-use factor1 for SC9 and SC14; 3 for SC10 and SC15Table 6.1.1.1-5, Options 1 and 2
Central beam elevation90°Table 6.1.1.1-6
ChannelRural, 100% outdoor UEs, clear sky, line of sightTable 6.1.1.1-5
LossesAtmospheric absorption enabled. Scintillation, clutter and additional losses set to zero.Table 6.1.1.1-5, NOTE 2
InterferenceDownlink, exact geometric model. All co-channel beams interfere.Section 6.1.1.1
UE attachmentRSRPTable 6.1.1.1-5
UE radioHandheld, height 1.5 m, 23 dBm, antenna gain 0 dBi, noise figure 7 dB, antenna temperature 290 KTable 6.1.1.1-3
UE antenna branchesOne transmit and one receive branch in the saved NetSim cases. The 3GPP handheld reference has two receive branches.Table 6.1.1.1-3
UE distribution10 UEs per metric beam, uniform in the Voronoi cell of each beam centre. 190 UEs in total.Table 6.1.1.1-5
Mobility and durationStationary UEs, 5 seconds
TrafficOne UDP full-buffer application per UE, 1460-byte packets
PHY and schedulerQAM256 MCS table, zero BLER, round robin
LogsNTN Radio Measurement Log and NTN UE Beam Association Log

Table 1: NetSim parameters for the Set-1 downlink calibration. Table numbers refer to TR 38.821.

The two metrics, and how NetSim computes them

Table 6.1.1.1-5 names the calibration metrics: coupling loss and geometry.

Coupling loss

The signal loss from the satellite antenna port to the UE antenna port. It depends on the distance, the channel model and the two antenna patterns, so it tests the geometry and the propagation model.

Geometry SINR

\[ \text{Geometry SINR} = -10\log_{10}\!\left(\frac{I}{C} + \frac{N}{C}\right) \]

C, I and N are the carrier, interferer and noise power levels measured over the configured signal bandwidth (TR 38.821 Table 6.1.1.2-1, NOTE). It adds the interference model and the UE noise figure to the test.

  1. Run the case in NetSim
  2. Read the downlink records of the radio measurement log
  3. Take the time mean for each UE
  4. Form the CDF across the 190 UEs

The time mean gives every UE the same weight in the CDF, whatever its number of log records. The beam association log confirms the serving beam of each UE.

Results: NetSim on the 11 company curves

Solid lines: the company curves of R1-1913404. Dashed blue line: NetSim. The 3GPP columns are the values of TR 38.821 Table 6.1.1.2-1. All values are in dB.

SC9

LEO-600, re-use factor 1

SC9: NetSim coupling loss and Geometry SINR CDFs overlaid on 11 company curves from R1-1913404
Metric5th percentile50th percentile95th percentile
3GPPNetSim3GPPNetSim3GPPNetSim
Coupling loss123.7123.759125.3125.540127.0127.200
Geometry SINR−3.1−2.985−1.1−1.1071.00.917
SC10

LEO-600, re-use factor 3

SC10: NetSim coupling loss and Geometry SINR CDFs overlaid on 11 company curves from R1-1913404
Metric5th percentile50th percentile95th percentile
3GPPNetSim3GPPNetSim3GPPNetSim
Coupling loss123.7123.759125.3125.540127.0127.200
Geometry SINR7.37.2978.27.9378.58.318
SC14

LEO-1200, re-use factor 1

SC14: NetSim coupling loss and Geometry SINR CDFs overlaid on 11 company curves from R1-1913404
Metric5th percentile50th percentile95th percentile
3GPPNetSim3GPPNetSim3GPPNetSim
Coupling loss129.8129.715131.3131.670133.0133.126
Geometry SINR−3.1−3.048−1.1−1.5161.00.958
SC15

LEO-1200, re-use factor 3

SC15: NetSim coupling loss and Geometry SINR CDFs overlaid on 11 company curves from R1-1913404
Metric5th percentile50th percentile95th percentile
3GPPNetSim3GPPNetSim3GPPNetSim
Coupling loss129.8129.715131.4131.670133.0133.126
Geometry SINR7.37.2278.27.9378.58.366

The percentiles agree

Across the four cases and the two metrics, the largest difference between a NetSim percentile and the TR 38.821 value is 0.42 dB. It occurs at the SC14 median Geometry SINR.

Altitude sets the coupling loss

The median coupling loss rises by 6.1 dB from LEO-600 to LEO-1200. Doubling the distance adds 6.0 dB of free-space loss. Set 1 raises the EIRP density by 6 dB at LEO-1200, so Geometry SINR stays nearly the same.

Re-use factor sets the SINR

With re-use factor 1, every beam interferes and the median Geometry SINR is near −1 dB. With re-use factor 3, fewer beams share the channel and the median rises to 7.9 dB.

Throughput and capacity for the same four cases

The same runs also report application throughput. These are NetSim results with full-buffer traffic. The throughput results in TR 38.821 Section 6.1.1.3 use FTP traffic at set resource utilisation, so the two sets do not compare.

Case Satellite capacity (Mbps) Beam radius (km) Area traffic capacity (kbps/km²) Average spectral efficiency (bits/s/Hz/TRxP)
SC9251.45623.6119.1380.441
SC10330.48923.61112.0100.580
SC14242.79847.2212.2060.426
SC15330.41847.2213.0020.580

Table 2: Capacity metrics over the 19 metric beams.

Satellite capacity

The sum of the application throughput of the 190 downlink applications, one per UE.

Area traffic capacity

Satellite capacity divided by the area of the 19 metric beams, where R is the beam radius:

\[ A_{\text{total}} = 19 \cdot \frac{3\sqrt{3}}{2} R^2 \]

Average spectral efficiency

Satellite capacity divided by the bandwidth of the 19 metric beams:

\[ \frac{\text{Satellite capacity}}{19 \times 30\ \text{MHz}} \]
CaseThroughput per UE (Mbps)Throughput per beam (Mbps)
5th50th95th5th50th95th
SC90.6951.0921.87211.28313.29215.182
SC101.6001.7551.76116.97117.40917.748
SC140.6951.0891.68811.18312.76314.222
SC151.6571.7551.76117.23517.39217.555

Table 3: Percentiles of application throughput. The per-beam value is the sum over the 10 UEs of a metric beam.

CDF of application throughput per UE for SC9, SC10, SC14 and SC15

Figure 3: CDF of application throughput per UE.

CDF of per-beam application throughput for SC9, SC10, SC14 and SC15

Figure 4: CDF of application throughput per beam.

Throughput follows Geometry SINR. The re-use factor 3 cases deliver about 330 Mbps at both altitudes, because their SINR distributions match. The re-use factor 1 cases deliver less, and their UEs spread over a wider throughput range.

The example ships with NetSim, with its configuration files and results. The NetSim 5G NTN manual explains it step by step.

References: 3GPP TR 38.821 V16.2.0, “Solutions for NR to support non-terrestrial networks (NTN)”. R1-1913404, “System Level Calibration Results for NTN on DL transmissions”, Thales, RAN1#99.