RFoG (Radio Frequency over Glass) is a fiber-to-the-premises architecture that carries traditional cable RF signals (DOCSIS/QAM video) over passive fiber instead of coax in the outside plant. It lets cable operators keep their existing CMTS/video headend and in-home coax/CPE while replacing amplifiers and coax distribution with passive optical splitters and fiber.

How it works:

• At the headend/hub, RF services are converted to optical at 1550 nm and broadcast over fiber through passive splitters (PON-like).
• At the customer premises, an RFoG ONU (R-ONU or micro-node) converts the 1550 nm downstream optical back to RF on coax for set-tops/cable modems.
• For upstream, the R-ONU uses a burst-mode laser (commonly 1610 nm; some use 1310 or 1590/1610 CWDM) to transmit only during scheduled DOCSIS upstream bursts back to an optical receiver, which hands RF to the CMTS.
• WDM filters allow coexistence with FTTH PON: e.g., GPON 1490/1310, RFoG 1550 downstream, 1610 upstream, sharing the same fiber.

Benefits:

• Eliminates powered amplifiers and most coax, improving noise/SNR, reach, reliability, and plant maintenance.
• Preserves DOCSIS/QAM services and OSS, easing migration to all-fiber.

Limitations:

• Optical Beat Interference (OBI) can occur when multiple R-ONUs transmit simultaneously at the same wavelength; mitigations include OBI-free R-ONUs, upstream wavelength separation (CWDM), and tight MAC scheduling.
• Still bound by DOCSIS RF spectral constraints compared with native PON.

• Physical medium

  • HFC: Fiber to optical node, then powered coax with amplifiers.
  • RFoG: All-fiber to premises with passive splitters; at the home an R-ONU converts optical to legacy RF.
  • FTTH/PON: All-fiber with ONT; purely digital framed traffic (GPON/XGS-PON).

• Signaling/Services

  • HFC: RF/DOCSIS over coax; linear QAM video and DOCSIS data.
  • RFoG: Same RF/DOCSIS and QAM video, but carried over glass; reuses CMTS, STBs, cable modems.
  • PON: Ethernet/IP frames; IPTV/OTT for video; requires ONT and new CPE.

• Plant power/maintenance

  • HFC: Active outside plant (nodes, amps), ingress/impulse noise issues.
  • RFoG: Passive outside plant like PON; minimal ingress, cleaner upstream.
  • PON: Passive plant; no RF ingress.

• Capacity/latency

  • HFC: Bounded by RF spectrum and split architecture; DOCSIS 3.1/4.0.
  • RFoG: Same DOCSIS limits and latency characteristics as HFC, but lower noise floor; still shared RF spectrum.
  • PON: Higher line rates (e.g., 2.5G/1.25G, 10G/10G); TDMA scheduling; different latency profile.

• Upstream specifics

  • HFC: Coax return path; susceptible to ingress.
  • RFoG: 1610 nm upstream lasers in R-ONUs; risk of optical beat interference (OBI) when multiple ONUs transmit; mitigations via OBI-free lasers/wavelength plans.
  • PON: Standardized upstream wavelengths and MAC; no OBI in same sense.

• Reach/splits

  • RFoG: ~20 km, split ratios similar to PON (e.g., 1:32/64) subject to optical budget.
  • HFC: Shorter coax runs; amplifier cascades limit reach.
  • PON: Comparable or greater reach with defined optical classes.

• Migration/cost

  • RFoG: Eases migration for cable operators; reuse headend/OSS and CPE.
  • PON: Higher transformation effort but greater long-term scalability.

Yes. RFoG (RF over Glass) is designed to be natively compatible with DOCSIS and legacy cable headends.

• DOCSIS compatibility: RFoG transports the same RF carriers used in HFC. DOCSIS 2.0/3.0/3.1 (including OFDM downstream and OFDMA upstream) works transparently when the R-ONU converts optical to coax at the customer premises. Existing cable modems, STBs, and outside-plant passives on the in-home coax remain unchanged.

• Headend compatibility: You continue using your existing CMTS, QAM modulators, and video platforms. What changes is the optical layer: add RFoG optical transmit/receive gear (e.g., 1550 nm downstream transmit/EDFA, WDMs, RFoG optical receivers) and deploy R-ONUs at the edge. No DOCSIS MAC or CMTS software change is required.

• Coexistence: RFoG can overlay on the same fiber with GPON/EPON via wavelength division (typically 1550 nm downstream video/RF, 1610 nm upstream; PON at 1490/1310), enabling gradual migration.

• Caveats: Upstream optical beat interference (OBI) can occur if multiple R-ONU lasers transmit simultaneously, degrading DOCSIS—especially OFDMA. Mitigate via OBI-free/OBI-reduced R-ONUs (wavelength isolation), tighter service-group segmentation, MAC-layer scheduling, or node splits. Ensure R-ONU bandwidth, burst timing, and level plans align with DOCSIS 3.1 specs.

• Performance/limits: RFoG delivers HFC-like RF levels with lower noise and extended reach (often up to ~20 km) and typical splits of 32–64 homes, subject to optical budget and OBI management.

Bottom line: RFoG is a drop-in RF transport replacement for the HFC coaxial span, preserving DOCSIS and headend investments, with added optical-layer components and OBI considerations.

Advantages

• Preserves legacy HFC/DOCSIS and video QAM investments (CMTS, STBs, OSS), enabling low-disruption fiber deep/FTTH.
• Eliminates outside-plant actives/amplifiers: lower power/OPEX, fewer truck rolls, higher reliability.
• Better downstream SNR and ingress immunity vs coax; longer reach and more split options.
• Coexists with GPON/XG(S)-PON via WDM overlays (1550/1610 nm), easing gradual migration.
• Immunity to lightning/EMI; supports analog/digital RF services without re-architecting CPE.

Limitations

• Optical Beat Interference (OBI): When two or more R-ONUs transmit upstream simultaneously at close optical wavelengths, their optical fields mix at the headend receiver, producing beat products that fall in the RF band, causing burst packet loss, MER degradation, and intermittent service issues that are hard to diagnose under load.
• OBI mitigations add cost/complexity and are imperfect: tighter wavelength control/locker lasers, per-customer CWDM channelization, centralized seed-light with wavelength-assigned ONU optics, narrowband optical filtering, smaller serving groups/scheduling discipline, or moving upstream to PON for high-contention areas.
• Still a linear RF link: upstream “noise funneling” persists; plant ingress from legacy RF segments can propagate; optical clipping/CSO/CTB constraints remain.
• Upstream bandwidth limited by DOCSIS split (e.g., 5–42/85/204 MHz) and shared contention; cannot match symmetric multi-gig PON without major DOCSIS upgrades.
• CPE and headend require burst-mode optics/receivers; R-ONUs cost more than simple ONTs, with stricter optical budgets and turn-on transients to manage.
• Operational complexity when coexisting with PON (inventory, WDM plan, troubleshooting mixed layers).
• Scaling challenges: to avoid OBI, nodes must be kept small or wavelengths diversified, reducing statistical multiplexing gains and increasing wavelength management overhead.

• Core headend/hub

  • CMTS/CCAP and EdgeQAMs generating RF (54–1002+ MHz).
  • 1550 nm downstream optical transmitter (DFB/EML), often with external modulation for linearity.
  • EDFA(s) to amplify 1550 nm broadcast video/data.
  • RFoG burst-mode upstream receiver(s) at 1610 nm (and/or 1590 nm if used).
  • WDM mux/demux modules to combine/split RFoG and any PON wavelengths.
  • Management/monitoring (optical power meters/OTDR as needed).

• Outside plant (passive)

  • Optical splitters (1xN), patch panels, connectors.
  • WDM filters/couplers for coexistence with GPON/XG(S)-PON.
  • No RF amplifiers (all-passive fiber); optional optical attenuators/equalizers.

• Customer premises

  • RFoG R-ONU/R-ONT (micro-node):
    • Downstream: photodiode converts 1550 nm optical to RF coax for CPE/STBs/cable modems.
    • Upstream: burst-mode 1610 nm DFB laser converts RF return to optical.
    • Variants: “OBI-mitigating/OBI-free” R-ONUs (e.g., wavelength-seeded, RSOA-based, or offset lasers).
  • In-home splitters, coax, power supply.

• Wavelength plan (typical/coexistence)

  • RFoG downstream: 1550 nm (1540–1565 nm).
  • RFoG upstream: 1610 nm (≈1603–1617 nm), alternative deployments sometimes use 1590 nm.
  • GPON coexistence (via WDM filters):
    • 1490 nm downstream, 1310 nm upstream.
  • XG-PON/XGS-PON coexistence:
    • 1577 nm downstream, 1270 nm upstream.
  • Optional analog/narrowcast overlays may sit within the 1550 nm downstream band.

• Notes on OBI (optical beat interference)

  • Occurs when multiple R-ONU lasers transmit simultaneously on same wavelength; mitigated by:
    • Tight MAC scheduling (DOCSIS upstream contention control).
    • OBI-mitigating R-ONUs (wavelength offsets/seeding).
    • Segmentation (smaller split ratios), per-node optical isolation.
    • Dedicated 1610 nm burst receivers with fast AGC.

• Bandwidth

  • RFoG carries standard DOCSIS over fiber, so capacity matches the DOCSIS version and spectrum used.
  • DOCSIS 3.0: downstream ~300–1,200 Mbps (8–24 bonded QAMs); upstream ~50–150 Mbps (3–8 bonded channels), depending on return spectrum and profiles.
  • DOCSIS 3.1: downstream ~1–5 Gbps (1.2 GHz plant, OFDM); upstream ~100–1,000+ Mbps (OFDMA, return spectrum permitting). Common retail tiers are 500 Mbps–1 Gbps down and 35–200 Mbps up.
  • Practical throughput per service group depends on channelization, modulation (SNR), and load; RFoG does not inherently increase DOCSIS spectral capacity but reduces impairment vs coax.

• Latency

  • Similar to well-engineered HFC DOCSIS, often slightly better jitter due to the fiber last mile.
  • Typical one-way latency ~5–15 ms; RTT ~10–30 ms under normal load. With DOCSIS 3.1 Low Latency features, sub-10 ms RTT for marked traffic is possible; congestion and scheduling can increase latency.

• Subscribers per node

  • RFoG “node” is usually a passive optical split; common split ratios are 1:32 or 1:64 to limit Optical Beat Interference (OBI) and upstream contention.
  • Typical homes passed per node: 32–64. Some deployments use up to 1:128 with OBI mitigation (e.g., wavelength plans, OBI-free ONUs), but many operators keep ≤64 for performance.
  • Active subscribers per DOCSIS service group are engineered similarly to HFC (often 50–200), but RFoG plants frequently target smaller groups to maintain upstream performance.

• Design to avoid OBI

  • Use OBI-mitigating RFoG headend receivers (phase/polarization diversity, beat-cancellation).
  • Deploy multiple return wavelengths (CWDM grid per R-ONU) and WDM filters; assign unique upstream λ where feasible.
  • Reduce simultaneous talkers: smaller service groups/splits; more nodes; tighter DOCSIS upstream scheduling and fewer contention opportunities (tune initial maintenance, backoff).
  • Specify low-linewidth, low-RIN burst lasers with fast, well-controlled turn-on/off and high extinction ratio; verify burst timing.
  • Control optical power: keep R-ONU Tx within receiver window; add pads/attenuators; prevent receiver compression.
  • Minimize reflections/ORL: APC connectors end-to-end, clean/inspect fiber, correct bend radius, high-isolation splitters, short drop tails; test at 1610 nm.
  • Avoid in-path EDFAs on return; use isolators if needed.
  • Keep RF ingress low: good CPE shielding/grounding, tight F-connectors, LTE filters where needed.

• Troubleshooting playbook

  • Correlate upstream SNR/MER dips and T3/T4, FEC/CRC with OBI events; monitor headend spectrum (5–85/204 MHz) during busy hour for bursty wideband noise.
  • Measure optics: downstream at R-ONU, upstream at headend; confirm within spec and no clipping. Insert temporary attenuators to test compression.
  • Check burst parameters with a burst analyzer: rise/fall, extinction, preamble length, timing vs receiver.
  • Sweep for reflections: OTDR at 1610 nm; remediate high-reflection links/connectors; replace UPC with APC.
  • Isolate offenders: bring ONUs online one at a time; lock out suspected drops; move noisy subs to a different splitter/receiver.
  • Validate WDM orientation and filter health; replace suspect WDMs/splitters.
  • RF hygiene at premises: replace corroded fittings, tighten homes, add HPF/LTE traps.
  • Optimize DOCSIS: reduce contention windows, spread channels, adjust minislot sizes/profiles; push upgrades to CPE and CMTS.
  • Long term: node splits, migrate heavy talkers to separate λ or to PON.