PHY LayerL3

WINNER II & Channel Simulation

RF Analysis Math

WINNER II CHANNEL MODELS + LINK ADAPTATION REFERENCE | IST WINNER II D1.1.2, 3GPP TR 38.901 | For TGn/TGax Channel Simulator, WINNER II Mapper, NR-U Coexistence, Link Adaptation tools6 rows#

Scenario	Environment	Path Loss Formula (dB)	Shadow Fading sigma (dB)	3GPP TR 38.901 Equiv.
B1	Urban micro-cell NLOS (outdoor/outdoor)	LOS: PL = 22.7log10(d) + 41.0 + 20log10(fc/5.0) NLOS: PL = 44.9log10(d) + 31.0 + 20log10(fc/5.0) Valid: 10m < d < 5000m	LOS: sigma=3 dB NLOS: sigma=4 dB	UMi (Urban Micro) Street Canyon
B3	Indoor hotspot NLOS (large hall, airport)	PL = 18.7log10(d) + 46.8 + 20log10(fc/5.0) Valid: 3m < d < 100m LOS component can be added with K=3dB for clear LOS	LOS: sigma=3 dB NLOS: sigma=4 dB	InH (Indoor Hotspot)
C1	Suburban macro-cell (residential outdoor)	LOS: PL = 23.8log10(d) + 41.2 + 20log10(fc/5.0) NLOS: PL = 26.7log10(d) + 39.1 + 20log10(fc/5.0) Valid: 30m < d < 1000m	LOS: sigma=4 dB NLOS: sigma=6 dB	UMa (Urban Macro)
C2	Urban macro-cell (typical dense city)	LOS: PL = 26.0log10(d) + 39.0 + 20log10(fc/5.0) NLOS: PL = (44.9-6.55log10(hBS))log10(d) + 34.46 + 5.83log10(hBS) + 23log10(fc/5.0) Valid: 30m < d < 2000m, hBS=10-25m	LOS: sigma=6 dB NLOS: sigma=8 dB	UMa (Urban Macro) alternate
D1	Rural macro-cell (sparse outdoor, agriculture)	LOS: PL = 21.5log10(d) + 44.2 + 20log10(fc/5.0) NLOS: PL = 40.0log10(d) + 10.5 + 20log10(fc/5.0) Valid: 30m < d < 10000m Applicable to 802.11ah HaLow long-range deployments	LOS: sigma=4 dB NLOS: sigma=8 dB	RMa (Rural Macro)
D2a	Large outdoor festival / event	PL = 23.5log10(d) + 42.5 + 20log10(fc/5.0) Dense crowd outdoor. Similar to B1 but open space.	sigma=6 dB	Outdoor scenario - no direct 3GPP equiv.

O2I (OUTDOOR-TO-INDOOR) PENETRATION LOSS -- Add to outdoor path loss when client is inside building5 rows#

Building Type	Loss @ 2.4 GHz	Loss @ 5 GHz	Loss @ 6 GHz	Notes
Residential (wood frame, drywall)	8-12 dB	12-15 dB	14-18 dB	Most common. Frequency-dependent. Add to outdoor PL for indoor client.
Office building (concrete, glass facade)	15-20 dB	18-25 dB	20-28 dB	Varies heavily with window ratio and glass type. Low-E glass: add 8-15 dB more.
Industrial/warehouse (concrete + metal)	20-30 dB	25-35 dB	28-38 dB	Heavy penetration loss. Outdoor AP rarely serves indoor industrial clients.
Underground (parking, basement)	25-35 dB	30-40 dB	35-45 dB	Typically requires dedicated indoor APs. Outdoor coverage unreliable.
Car (standard glass)	6-10 dB	8-12 dB	10-14 dB	802.11p/bd for V2V uses dedicated ITS bands. Wi-Fi in car uses standard values.

LINK ADAPTATION ALGORITHMS -- AARF & MINSTREL reference | For Link Adaptation Simulator tool6 rows#

Algorithm	Type	Rate Increase Rule	Rate Decrease Rule	Probe Mechanism	Notes
AARF (Adaptive Auto Rate Fallback)	Threshold-based	Increase rate after N_success consecutive successes at current rate (typically N=10 for increase)	Decrease rate after 2 consecutive failures at current rate (immediate response to failure)	None - relies purely on success/failure counts	Simple, widely deployed in legacy drivers. Fast fallback but slow climb. Oscillates in marginal conditions.
AARF-D (with Dual AARF)	Threshold-based	Increase after N_success successes. N_success increases after each failed probe (1, 2, 4, 8...)	2 consecutive failures -> decrease one step. Reset N_success counter.	Probes next rate after stable period. Failed probe doubles next probe threshold.	More cautious than AARF. Less oscillation in fading channels.
MINSTREL	Statistical sampling	Maintains per-rate statistics: success_count, retry_count, throughput_estimate. Always uses highest estimated throughput rate.	Continuous: if throughput estimate drops, rate switches automatically. No hard threshold.	10% of transmissions are probes at random rates. Random sample keeps statistics fresh.	Best for real-world performance. Used in Linux mac80211, hostapd. Adapts well to rapid channel changes. Default in modern Linux Wi-Fi.
MINSTREL-HT (for HT/VHT/HE)	Statistical sampling	Same as MINSTREL but works with A-MPDU aggregation. Optimises for aggregate throughput not per-frame rate.	Tracks per-rate success at multiple aggregation levels (1,2,4,8,16... subframes)	Multi-rate retry: tries rate A, falls back to rate B, then rate C within single A-MPDU	Handles MCS 0-31 (HT) or 0-9 (VHT) or 0-11 (HE). NSS selection included. Linux kernel 3.18+.
RRAA (Robust Rate Adaptation)	EWMA-based	Uses Exponentially Weighted Moving Average of success rate. Increase when EWMA > high_threshold (0.85)	Decrease when EWMA < low_threshold (0.35)	Probes one rate higher when stable. Uses conservative thresholds.	Better at distinguishing temporary fading from sustained bad channel. Less reactive than AARF.
Vendor proprietary (Cisco/Aruba)	Hybrid	Each vendor uses proprietary variation. Cisco CCX: client reports link quality metric. AP uses combined RSSI + link quality to predict optimal rate.	Vendor-specific. May include RF signature matching, spatial diversity info.	Vendor-specific probe scheduling.	Not standardized. Black box. Generally performs similarly to MINSTREL in practice.

NR-U vs 802.11ax COEXISTENCE PARAMETERS -- 3GPP TS 37.213 vs IEEE 802.11-20207 rows#

Parameter	NR-U (Cat 4)	802.11ax EDCA BE	NR-U (Cat 2)	Impact	Notes
Slot time	9 µs (same as Wi-Fi 5GHz)	9 µs	9 µs	Equal	Same slot = theoretically equal contention probability.
CWmin	15 (same as Wi-Fi BE)	15	N/A (no backoff in Cat 2)	Cat 4: Equal to BE. Cat 2: unfair advantage.	NR-U Cat 4 designed for fairness with 802.11ax BE. Cat 2 (no backoff) can dominate channel.
CWmax	1023 (same as Wi-Fi BE)	1023	N/A	Equal for Cat 4
AIFS equivalent	16 µs + 2×slot = 34 µs (same as DIFS)	AIFS[BE]=43µs (SIFS+3×slot)	16+1×slot = 25 µs (shorter than DIFS)	NR-U has slightly shorter effective AIFS than Wi-Fi BE in Cat 4. Cat 2: much shorter.	NR-U AIFS[VO]-equivalent access. Some studies show 5-15% airtime advantage for NR-U even Cat 4.
Spatial reuse (OBSS/PD)	No equivalent in NR-U Rel-16	BSS Color + OBSS/PD adaptive threshold	No equivalent	802.11ax can reuse channel when NR-U signal below OBSS/PD threshold. NR-U cannot.	802.11ax has structural advantage in dense mixed NR-U/Wi-Fi deployments via BSS Color.
UL multi-user	Each UE contends independently (no TF)	AP sends Trigger Frame -> simultaneous UL from multiple STAs	Same as Cat 4	802.11ax UL OFDMA more efficient. NR-U UL: N×EDCA overhead. ax UL: 1×EDCA overhead.	Key ax advantage in UL-heavy IoT/sensor applications.
Coexistence metric (ETSI TR 103 594)	Cat 4 achieves ~fair airtime share with 802.11ax BE	Reference	Cat 2 achieves 1.2-1.5x airtime share vs 802.11ax BE in studies	Acceptable with Cat 4. Cat 2 unfair.	ETSI TR 103 594 coexistence evaluation: recommends Cat 4 minimum for fairness.