Developinggap named

Wireless — RF, 802.11 & Point-to-Multipoint

Nikita Tarikin runs MikroTik RouterOS point-to-multipoint (PtMP) wireless in production — including NV2 links, the MikroTik time-division-multiple-access (TDMA) protocol built to replace the 802.11 contention model on outdoor radio — verified by the MikroTik Certified Wireless Engineer exam (MTCWE); long-distance acknowledgement-timeout (ACK-timeout) tuning at wireless-ISP (WISP) scale, and the legacy mesh and the dual-radio modes of Nstreme — MikroTik’s first-generation proprietary radio protocol — are understood but sit outside his deployed sub-kilometre link profile.

Deployed where the access is scheduled and the hop is short; the WISP-scale long-haul, the legacy protocols, and the academic RF formulas are each marked on the map below for exactly what they are — run, studied, or both.

4.3
Knowledge K̄: 4.2
Deployment Ē: 35
leaves mapped: 9
studied-ahead: 5
deployed-ahead

Verified by MTCNA 100%MTCWE 92%

The map

Every wireless leaf in the radio-and-link half of the official MikroTik curriculum, placed by what is known against what is run in production. A mark on the diagonal is a leaf where study and deployment reach the same depth; marks that fall below it are studied ahead of production, and a rarer set sits above it — leaves operated in the field a notch ahead of their own textbook physics. The dense mastery corner is the radio that ships every day; the daylight on both sides of the diagonal is the shape of this domain, as the data draws it.

Read left-to-right as knowledge, bottom-to-top as deployment. The dense cluster in the upper-right is the deployed RF stack — the 802.11 standards, the NV2 TDMA link, antennas, channels, and the survey tool-set — run in production. The comb of drop-lines below it is the studied-ahead set: the legacy Nstreme and mesh modes superseded by NV2, the long-haul WISP NV2 tuning edge, and the MPLS-over-wireless intersection. The short up-ticks above the diagonal are the rarer direction — the propagation, decibel, and 802.11n internals operated in the field ahead of where their academic formulas are held.

The topography

The full leaf-level ledger, grouped by topic — the wireless hardware and antennas, the RF characteristics, the 802.11 standards, the 802.11n specifics, WDS and mesh, wireless bridging, the legacy Nstreme protocol, the NV2 TDMA protocol, the non-CAPsMAN link modes, and the survey tool-set. Each row carries its own two verdicts — what is known and what is deployed — and the field note that earned them, quoted verbatim from the assessment.

Leaf	Knowledge	Deployment	Field note
Wireless Hardware & RouterBOARD		K̄ 5.0 · Ē 5.0 · 3 leaves
Wireless routers & MikroTik wireless cards	5Teach	5Teach	`wireless`=Atheros driver vs `wifi`/wifiwave2=newer ac-wave2/ax chipsets. Runs both driver worlds (cf. MTCEWE).
RouterBOARD with integrated wireless	5Teach	5Teach	Integrated directional radio (SXTsq etc.): set antenna-gain to the enclosure's built-in figure so EIRP/regulatory computes right.
Antenna types	5Teach	5Teach	MIMO needs decorrelated paths → a single dual-pol (V+H) antenna gives two spatial streams on one boresight.
RF Wireless Characteristics		K̄ 3.5 · Ē 4.5 · 6 leaves · 3 splits
RF spectrum, electromagnetic energy & decibelsdeployed ahead	2Aware	4Fluent	Honest don't-know on the dB→power conversion (6 dB = ×4): 'I didn't get the question'. Works with dBm operationally; the academic decibel math is the soft spot. Theory edge.
Antenna theory — isotropic, directional, omnidirectional & polarizationdeployed ahead	3Confident	5Teach	Honest don't-know on the polarization-mismatch formula (20·log cos θ): 'I didn't study that'. But knows antenna types/directional/omni/polarization PRACTICALLY (MTCWE antenna-types = Teach). Academic-formula edge, practical knowledge solid.
Attenuation, absorption & reflectiondeployed ahead	2Aware	4Fluent	Picked the direct-ray-blockage trap; with clear LOS the real Fresnel-zone mechanism is destructive phase interference from diffracted edge signals. Deploys 5GHz links knowing clearance is needed, not the physics why. Theory edge.
2.4/5 GHz cell sizes & transmitter power	5Teach	5Teach	The practical one he GOT: cranking AP power hurts a multi-client cell via the asymmetric link (clients can't reply) + added contention. Operational RF = Teach.
Station roaming	5Teach	5Teach	802.11r FT speeds roaming by pre-computing the PMK with the target AP so the 4-way handshake is skipped. Consistent with v1 802.11r/k/v Teach.
Co-channel & adjacent-channel interference; AP placement & airtime	4Fluent	4Fluent	wide-channel co-channel-interference / channel reuse in dense cells; auto channel = frequency-list + reselect-interval + skip-dfs, but per-AP LOCAL selection (knows RouterOS isn't a global RRM controller like Cisco/Aruba)
Wireless Standards (802.11)		K̄ 5.0 · Ē 5.0 · 9 leaves
802.11 a/b/g/n features & bands	5Teach	5Teach	Legacy 802.11b client triggers protection mode -> whole-cell slowdown.
802.11ac features	5Teach	5Teach	Wider channels/256-QAM/streams — AND volunteered 5GHz shorter-wavelength -> shorter range -> less neighbor noise -> higher SNR (exceeded framing, field insight).
802.11ax overview	5Teach	5Teach	OFDMA = resource units serving many clients per TXOP (great for dense/small-packet); distinct from MU-MIMO (spatial streams)
Channels, frequencies & channel widths	5Teach	5Teach	1/6/11 = only non-overlapping 20MHz channels in 2.4GHz.
Modulation, MCS & data rates	5Teach	5Teach	Single-stream/20MHz/long-GI caps MCS despite strong signal.
Channel bonding & frame aggregation	5Teach	5Teach	Ce = control at the configured freq, extension 20MHz above; eC mirrors below. Knows occupied-spectrum + neighbor coordination -- a knob most cert-passers can't explain
Chains — SISO, MIMO & MU-MIMO	5Teach	5Teach	PHY rate = channel width x MCS x spatial-streams x guard-interval; single-stream-despite-RSSI = one degraded chain (polarization/antenna), MIMO needs 2 decorrelated paths; his actual 5ac gear
CSMA/CA & HW protection (RTS/CTS)	5Teach	5Teach	Hidden-node collisions at the AP → RTS/CTS (`hw-protection-mode=rts-cts`): stations reserve the medium via the AP first. Real PtMP mitigation.
QoS / WMM	5Teach	5Teach	WMM = four EDCA access categories (voice/video/best-effort/background) that bias medium contention by priority.
802.11n Specifics		K̄ 3.3 · Ē 4.7 · 3 leaves · 2 splits
802.11n features & data ratesdeployed ahead	3Confident	5Teach	Knows n features — A-MPDU aggregation correct — but the short-GI tradeoff (~11% throughput / multipath-sensitivity) was missed. Inverted split: deploys n daily, one PHY internal soft.
TX-power & chain settingsdeployed ahead	2Aware	4Fluent	Sets tx-power correctly in production (uses default/card-rates mode — the right operational call), but soft on the RF math: missed why over-driving raises errors (EVM / PA-linearity) AND multi-chain power combining (+~5 dB for 3 chains: thinks per-chain=total). Knowledge-edge / experience-strong split.
Link debugging	5Teach	5Teach	Strong RSSI + low CCQ + high hw-retries with both chains similar = low SNR / co-channel interference, not a chain fault. Real troubleshooting command.
WDS & MESH		K̄ 3.5 · Ē 2.0 · 2 leaves · 2 splits
WDS protocol — dynamic & staticstudied ahead	3Confident	2Aware	Reasoned dynamic-WDS correctly — auto-creates the interface + bridges it — but self-flagged very shallow: MTCWE-lab-only ~10yr ago, never production. Distinct from station-bridge / 4-address (transparent-wds) which he DOES run.
MESH — RSTP bridge & HWMP+studied ahead	4Fluent	2Aware	Knew HWMP+ does L2 path-selection / reactive route discovery vs RSTP just blocking loops in a flat WDS bridge. Mesh is niche in his profile — experience inferred-undeployed
Wireless Bridging		K̄ 3.5 · Ē 3.0 · 2 leaves · 1 split
Transparent bridge — WDS bridging, AP/Station-WDS & pseudobridge	5Teach	5Teach	station mode sends 3-address frames and rewrites client src MACs to its own; transparent bridging needs station-bridge (MikroTik-proprietary) or WDS/station-wds for the 4-address frame
MPLS/VPLS over wirelessstudied ahead	2Aware	1Learning	Honest 'MPLS is not my thing'. The named carrier MPLS/VPLS blind spot (the global deep edge), confirmed at the wireless intersection. Knows the L2MTU principle generally (Teach elsewhere) but didn't connect it to label-stack overhead.
Nstreme Protocol		K̄ 3.3 · Ē 2.0 · 3 leaves · 3 splits
Nstreme features & config optionsstudied ahead	3Confident	2Aware	Knows Nstreme dual + proprietary-interop, but mapped its CORE access method to NV2's TDMA — it's actually AP-controlled polling. Runs NV2 (the successor), not Nstreme. Knowledge>experience legacy split.
Nstreme dualstudied ahead	4Fluent	2Aware	Knew Nstreme Dual's distinctive design — two radios (one TX-only, one RX-only) for full-duplex PtP. Legacy feature, never deployed (uses NV2).
Nstreme troubleshootingstudied ahead	3Confident	2Aware	Knew Nstreme is MikroTik-proprietary — non-RouterOS stations can't associate. Minimal Nstreme field exposure (NV2 in production).
NV2 (TDMA) Protocol		K̄ 5.0 · Ē 4.0 · 2 leaves · 1 split
NV2 features & config options	5Teach	5Teach	PRODUCTION-PROVEN 5-8yr 24/7 (home ~400m NV2, rice-field ~900m NV2, Moscow estate OmniTik 5ac->4x SXTsq-5ac PtMP). TDMA AP-scheduled (no hidden-node); cell-radius gates range NOT distance; NV2 has its OWN security (nv2-security + nv2-preshared-key, AES-CCM), not the 802.11 WPA profile. Runs his family's links -- not book-knowledge.
NV2 troubleshootingstudied ahead	5Teach	3Confident	Knew NV2 sync aligns co-located sectors' TDMA TX/RX windows so they don't transmit into each other's receive (advanced). Self-flagged NO co-located multi-sector / large-WISP experience — Teach at his PtMP scale, untested at WISP scale (mirrors the v1 long-haul edge).
Non-CAPsMAN Wireless Modes		K̄ 5.0 · Ē 4.0 · 2 leaves · 1 split
Repeaters & extenders	5Teach	5Teach	Single-radio repeater inherently ~halves throughput (one radio receives then retransmits each frame on the same channel).
MikroTik 60 GHz Wireless Wire bridgingstudied ahead	5Teach	3Confident	Knew the 60GHz tradeoff cold: multi-gig throughput, very short range, near-zero non-LOS (O₂ absorption peak). Knowledge Teach; product/deployment (Cube/wAP 60G) not directly probed — experience inferred Confident.
Wireless Surveys		K̄ 5.0 · Ē 4.3 · 3 leaves · 1 split
Pre- & post-install site surveys	5Teach	5Teach	Post-install survey = verify the as-built coverage/SNR/roaming matches the predicted design (vs pre-install planning).
Spectrum analysis	5Teach	5Teach	A strong constant-carrier with no 802.11 framing = a non-WiFi interferer (radar/analog video); avoid that channel, Wi-Fi can't share a constant carrier. Consistent with spectral-scan Teach.
Prediction softwarestudied ahead	5Teach	3Confident	Knew what prediction software (Ekahau/TamoGraph/iBwave) models: coverage/SNR from a floor-plan + material-attenuation model to place APs + pick channels. Knowledge Teach; formal-tool use vs hands-on spectral-scan surveys not confirmed — experience inferred Confident.

The edges

The daylight here runs in both directions. On the studied-ahead side sit the legacy radio protocols — the Nstreme dual-radio full-duplex design, and the WDS and mesh modes — held to teaching depth but superseded in production by NV2, which carries the live links instead; the co-located multi-sector NV2 timing that a wireless-ISP build would lean on, understood but never run at that scale; and the MPLS-over-wireless intersection, which reaches into the same carrier label-switching frontier owned elsewhere on this map. The rarer direction runs the other way: the RF propagation and decibel formulas, the polarisation maths, and the 802.11n physical-layer internals are all operated in the field a notch ahead of where the academic derivation is held — the links run daily while the textbook proof is the softer side. Each gap is quoted below exactly as the assessment recorded it.

Knowledge ahead of deployment

Known deeper than it currently runs — named, not rounded over.

MESH — RSTP bridge & HWMP+know 4 Fluent→run 2 Aware
Knew HWMP+ does L2 path-selection / reactive route discovery vs RSTP just blocking loops in a flat WDS bridge. Mesh is niche in his profile — experience inferred-undeployed
Nstreme dualknow 4 Fluent→run 2 Aware
Knew Nstreme Dual's distinctive design — two radios (one TX-only, one RX-only) for full-duplex PtP. Legacy feature, never deployed (uses NV2).
NV2 troubleshootingknow 5 Teach→run 3 Confident
Knew NV2 sync aligns co-located sectors' TDMA TX/RX windows so they don't transmit into each other's receive (advanced). Self-flagged NO co-located multi-sector / large-WISP experience — Teach at his PtMP scale, untested at WISP scale (mirrors the v1 long-haul edge).
MikroTik 60 GHz Wireless Wire bridgingknow 5 Teach→run 3 Confident
Knew the 60GHz tradeoff cold: multi-gig throughput, very short range, near-zero non-LOS (O₂ absorption peak). Knowledge Teach; product/deployment (Cube/wAP 60G) not directly probed — experience inferred Confident.
Prediction softwareknow 5 Teach→run 3 Confident
Knew what prediction software (Ekahau/TamoGraph/iBwave) models: coverage/SNR from a floor-plan + material-attenuation model to place APs + pick channels. Knowledge Teach; formal-tool use vs hands-on spectral-scan surveys not confirmed — experience inferred Confident.
WDS protocol — dynamic & staticknow 3 Confident→run 2 Aware
Reasoned dynamic-WDS correctly — auto-creates the interface + bridges it — but self-flagged very shallow: MTCWE-lab-only ~10yr ago, never production. Distinct from station-bridge / 4-address (transparent-wds) which he DOES run.
MPLS/VPLS over wirelessknow 2 Aware→run 1 Learning
Honest 'MPLS is not my thing'. The named carrier MPLS/VPLS blind spot (the global deep edge), confirmed at the wireless intersection. Knows the L2MTU principle generally (Teach elsewhere) but didn't connect it to label-stack overhead.
Nstreme features & config optionsknow 3 Confident→run 2 Aware
Knows Nstreme dual + proprietary-interop, but mapped its CORE access method to NV2's TDMA — it's actually AP-controlled polling. Runs NV2 (the successor), not Nstreme. Knowledge>experience legacy split.
Nstreme troubleshootingknow 3 Confident→run 2 Aware
Knew Nstreme is MikroTik-proprietary — non-RouterOS stations can't associate. Minimal Nstreme field exposure (NV2 in production).

Deployed ahead of the textbook

Run in production a notch beyond the studied internals — the rarer direction.

RF spectrum, electromagnetic energy & decibelsknow 2 Aware→run 4 Fluent
Honest don't-know on the dB→power conversion (6 dB = ×4): 'I didn't get the question'. Works with dBm operationally; the academic decibel math is the soft spot. Theory edge.
Antenna theory — isotropic, directional, omnidirectional & polarizationknow 3 Confident→run 5 Teach
Honest don't-know on the polarization-mismatch formula (20·log cos θ): 'I didn't study that'. But knows antenna types/directional/omni/polarization PRACTICALLY (MTCWE antenna-types = Teach). Academic-formula edge, practical knowledge solid.
Attenuation, absorption & reflectionknow 2 Aware→run 4 Fluent
Picked the direct-ray-blockage trap; with clear LOS the real Fresnel-zone mechanism is destructive phase interference from diffracted edge signals. Deploys 5GHz links knowing clearance is needed, not the physics why. Theory edge.
802.11n features & data ratesknow 3 Confident→run 5 Teach
Knows n features — A-MPDU aggregation correct — but the short-GI tradeoff (~11% throughput / multipath-sensitivity) was missed. Inverted split: deploys n daily, one PHY internal soft.
TX-power & chain settingsknow 2 Aware→run 4 Fluent
Sets tx-power correctly in production (uses default/card-rates mode — the right operational call), but soft on the RF math: missed why over-driving raises errors (EVM / PA-linearity) AND multi-chain power combining (+~5 dB for 3 chains: thinks per-chain=total). Knowledge-edge / experience-strong split.

The warrant

Third-party verification of the wireless depth above — the wireless engineer exam stands behind the deployed RF and point-to-multipoint work as an amplifier of it, never a substitute; the enterprise-wireless exam is recorded here as not yet taken, a marker recorded, not a gap rounded over.

MTCEWEEnterprise Wireless Engineer — in progress / not yet taken
MTCNAVerified — 100%, Network Associate
MTCWEVerified — 92%, Wireless Engineer

Wireless does not end at the outdoor link. Where it connects:

The questions

What point-to-multipoint and wireless work is genuinely production-proven?

The honest verdict: I run point-to-multipoint (PtMP) and point-to-point NV2 links in production at sub-kilometre scale, alongside the RF planning, antenna, channel and survey work around them — verified by MTCWE (MikroTik Certified Wireless Engineer); the wireless-ISP-scale long-distance and legacy-protocol leaves are the studied edge, named on the ledger above. The mechanism, publicly: NV2 is MikroTik’s proprietary protocol built on TDMA (Time Division Multiple Access) rather than the CSMA/CA contention that regular 802.11 uses — and per MikroTik’s own documentation that is exactly what makes it strong for PtMP, because TDMA schedules each station’s slot and so removes the hidden-node problem where two stations that cannot hear each other collide at the access point. The one setting that gates range is nv2-cell-radius (in kilometres): it sizes the contention slot the access point reserves for the farthest client, and if it is set too small a distant client cannot complete ranging and disconnects with a “ranging timeout.” That is the radio I run every day, not book-knowledge.

Do you tune long-distance, wireless-ISP-scale links?

Not at that scale — this is the named edge of the map. I deploy NV2 links at sub-kilometre distances; the wireless-ISP (WISP) long-haul and co-located multi-sector builds I hold at teaching depth but have not run. The mechanics I can place, publicly, sharpen exactly where that line is. On a regular 802.11 link the distance setting tunes the acknowledgement-timeout (ACK-timeout) so a far station’s ACK is not given up on too early — but in NV2 and Nstreme, per MikroTik’s documentation, acknowledgements are not used at all, because the TDMA scheduler already knows when each station will speak. So the long-distance knob on an NV2 link is not ACK-timeout but tdma-period-size: on a long hop the access point must leave a propagation gap each period for the last frame to reach the client and return, and widening the period reclaims throughput lost to that gap at the cost of added latency. Co-located sector sync — aligning several sectors’ TDMA transmit and receive windows so they do not transmit into each other’s receive — is the WISP-scale skill I understand but have not operated. I would lab it before I bill it.

Which RouterOS 7 driver carries your NV2 links — the legacy wireless package or the new wifi stack?

The legacy one, by necessity — and that is a current, deliberate distinction, not nostalgia. Since RouterOS 7.13 the drivers split into two worlds that cannot run at once: the legacy wireless package (the Atheros driver) and the newer wifi-qcom-ac and wifi-qcom packages (the 802.11ac-wave2 and 802.11ax drivers behind the menu once called wifiwave2, renamed wifi in 7.13). Per MikroTik’s own package matrix, Nstreme and NV2 are carried only by the legacy wireless driver — on a pure 802.11ax device they are simply “not possible,” and the wireless package conflicts with the wifi-qcom packages, so the two driver worlds cannot be active together on one radio. A production NV2 PtMP link therefore stays on the legacy driver, while the new wifi stack — WPA3, fast roaming, the version-two central controller — is the enterprise-Wi-Fi world. I run both, but the central-controller half of the story lives on its own page; this one is the outdoor link.

Do you run the legacy Nstreme protocol, or Layer-2 tunnels over the radio?

Nstreme: no longer — I run its successor. I hold the Nstreme design, including Nstreme-dual’s distinctive trick of using two radios, one transmit-only and one receive-only, for full-duplex point-to-point — but in production NV2 (Nstreme’s TDMA successor) replaced it, so Nstreme sits in the studied-ahead column. Both are MikroTik-proprietary, so a non-RouterOS station cannot associate with either. As for Layer-2 tunnels over the radio: carrying MPLS or VPLS across a wireless link is the one wireless leaf that reaches into my named frontier rather than my deployed ground. The radio underneath it I run; the label-switching layer on top — and the way its label stack eats into the link’s usable frame size — is the carrier MPLS/VPLS edge I own as a named frontier elsewhere on this site. If the job is carrier Layer-2 over wireless, see the carrier inter-networking page, where that frontier is named; the link itself, I will build.

Bring the wireless link.

If it is a point-to-multipoint or point-to-point NV2 link, RF planning, antenna alignment, or interference and survey work on RouterOS, that is ground I run in production. If it reaches into wireless-ISP-scale long-distance tuning, the legacy Nstreme or mesh modes, or MPLS over the radio, I will tell you where it sits against my deployed edge — and lab it before I bill it — before the work starts, not after.

Start the conversation