OSP Design for Cable TV Headend Connectivity: Headend-to-Hub and Hub-to-Node Fiber

A cable operator can freeze a node map and still not have a transport design that will survive DAA. The map names serving areas. It does not assign a diverse headend-to-hub path, a fiber count that can host Remote PHY Devices, or a hub-face record a construction crew can splice to.

This guide stays on MSO transport: headend-to-hub rings and hub-to-node fiber, then fiber-count allocation for DAA and the hub-site handoff. We do not size generic trunk capacity here. We do not redraw node boundaries. We show how the OSP package turns an architecture decision into a named route and a fiber assignment the operator can accept.

What OSP Design for Cable TV Headend Connectivity Controls

OSP design for cable tv headend connectivity is the outside-plant package that assigns fiber from the headend to each hub and from each hub to each node. Draftech's package covers 4 transport decisions: ring diversity and DAA fiber-count allocation plus hub-to-node routing and hub-site handoff. The design names the path. It does not set architecture.

The buyer is the incumbent cable operator moving analog optics toward Distributed Access Architecture, not an overbuilder writing a greenfield HLD. The operator already has hubs and nodes in the street. The question is which fiber will carry CCAP Core traffic to those sites, and which record a hub technician will trust when the sheath arrives. We write that question on the kickoff sheet before a designer opens a route file, because a drawing that starts without a named DAA split will copy analog 2-fiber node math onto a port that now needs Ethernet and timing, plus a protect path.

Coverage is not the same fact. CostQuest Associates' Broadband in America Report, Cable Market Focus, May 2026 edition, reports coaxial service at 100/20 Mbps available to approximately 95.1 million broadband serviceable locations, or 81.7 percent of locations, as of June 30, 2025. That figure describes reach. It says nothing about whether the working and protect fibers to a given hub share a duct.

Fiber pressure is real and still does not write the transport sheet. The Fiber Broadband Association, in "U.S. Home Fiber Deployments Top 88M homes passed" dated April 17, 2025, citing the RVA LLC North American Fiber Deployment Report released in January 2025, reported 88.1 million U.S. homes passed with fiber. An operator facing that passing count still needs named headend-to-hub and hub-to-node paths before it spends DAA capital on a corridor whose only spare is a drawing note.

Safety bounds the field work that feeds this design. The OSHA Telecommunications standard, 29 CFR 1910.268, is the named federal reference for telecommunications work. The employer's approved procedure still governs the crew. We never treat a hub-face photograph as a splice instruction.

The trade-off in our 4-decision package is extra office time before route production. A single-line hub feeder looks faster. It leaves diversity and DAA counts as later surprises, which is when construction quantities start inventing spare fiber. We accept the setup cost because an unknown that stays named is cheaper than an unknown that ships as a bid item nobody can terminate.

Headend-to-Hub Fiber Rings and Diversity Paths

Start with the ring, not the sheath. If the working path and the protect path share a duct bank or a pole line, stop. A shared splice case is the same defect. You do not have a ring. You have two fibers with one failure domain.

The table is the preview map. Later sections expand each row without turning this post into generic trunk engineering or node-boundary work.

CableLabs Modular Headend Architecture v2 Technical Report, CM-TR-MHAv2-V01-150615, released June 15, 2015, describes Remote PHY as moving the PHY element out of a CCAP platform onto a Remote PHY Device interconnected on an IP network. That interconnection is the Converged Interconnect Network. It runs on the headend-to-hub plant this section assigns. MHAv2 does not draw your ring. It tells you why a shared duct is no longer a small optics inconvenience. A cut that used to drop analog light now drops Ethernet sessions to every RPD behind that hub.

Name the ring before the sheath

We record two independent routes between the headend and each hub: working and protect, with named splice points and a named failure domain each path is supposed to survive. A second buffer tube in the same sheath is a spare, not a ring. Call it a spare if that is what it is.

Common-structure audit is the first production gate. Walk the two routes against the same GIS and the same as-built IDs. If they meet at a manhole, the diversity claim is false until one path moves. We would rather hold a hub than ship a "diverse" pair that dies together.

Analog HFC often ran one downstream fiber and one upstream fiber to a node, and operators sometimes treated a second tube in the same cable as protection. DAA does not forgive that habit. Digital optics still fail when the sheath fails. Name the shared structure. Move one path, or drop the diversity claim from the construction package.

We defer generic trunk capacity and optical-budget math to fiber trunk line engineering. This ring is an MSO assignment problem: which physical path is working and which is protect. Hub IDs ride those paths by assignment. A beautiful loss budget on a single path does not create a second path.

Fiber-Count Allocation for DAA: R-PHY versus R-MACPHY

CableLabs Distributed CCAP Architectures Overview Technical Report, CM-TR-DCA-V01-150908, released September 8, 2015, names 3 distributed approaches: Remote PHY and Remote MAC-PHY, plus Split-MAC. Those splits change what the fiber must carry. They do not change the duty to count it.

R-PHY, in CableLabs Remote PHY Specification CM-SP-R-PHY-I21-251211, issued December 11, 2025, keeps MAC in the CCAP Core and places PHY in the Remote PHY Device. The hub-to-node link then carries DEPI and UEPI sessions, plus timing and control. R-MACPHY moves MAC and PHY together, so the same physical route carries a different payload. Copying analog node fiber math onto either split is how a 2-fiber assignment arrives at a 10G Ethernet face and does not light the Remote PHY Device at that hub.

We allocate fiber in named pools, not as a round leftover. Working traffic gets a count and a far-end. Protect traffic gets its own count on the independent path. Timing or sync gets a dedicated pair only when the operator's DAA platform requires it. Named growth gets a year and a trigger, not a hopeful extra tube. Video or out-of-band overlays get fibers only if the operator still runs those overlays on separate glass. A pool without a far-end is a note. It is not an assignment.

DOCSIS 4.0 Physical Layer Specification, CM-SP-PHYv4.0-I07-250627, issued June 27, 2025, assumes a distributed architecture for FDD systems and treats the downstream modulator as located in an FDD node. That is an RF placement fact. It is not a fiber-count recipe. We keep the PHY document out of the assignment spreadsheet so a designer cannot "pass" a count against a specification that never counted fibers.

One thing I still refuse is treating a hub drawing's unused tube as spare DAA capacity. Spare is a terminated, identified fiber with a port. A dark tube with no faceplate is inventory. Inventory does not light an RPD.

R-PHY fiber is not R-MACPHY fiber

Write the architecture split on the assignment sheet before the first count. If the operator has not chosen R-PHY or R-MACPHY for that hub, the count stays held. Guessing the split to close a BOM is how a later Core upgrade inherits the wrong Ethernet rate and the wrong timing plant for that hub.

Port math follows the split. An RPD needs Core sessions and a CIN path back to the headend. An RMD needs a Layer 2 Ethernet or PON payload and a different control plane. Mixing those on one unlabeled fiber is a records failure, not a clever reuse. We would rather show two held ports than one "universal" pair that no platform can terminate.

The CATV to fiber infrastructure survey is the occupancy evidence this count will spend. If the survey held overlash or duct space, the DAA count cannot convert that hold into spare fibers. Occupancy is a pathway fact. Count is an assignment fact. Keep them on the same IDs and keep their statuses independent.

Hub-to-Node Fiber and Fiber-Deep Routing

Hub-to-node fiber is the access transport, not a miniature trunk. This section assigns which node gets which fibers from which hub, including the laterals a fiber-deep program will add. It does not size a regional backbone.

Our HFC network design and coax plant engineering work already describes fiber-deep routing: new fiber from the existing hub or headend to new, deeper node sites, often with new pole attachments. Legacy nodes in that model served 500 to 1,000 homes through a long coax cascade. Fiber-deep architecture serves 50 to 250 homes per node. Those home-count targets are operator policy. The OSP job is to route fiber to the new sites without pretending the old analog node fiber can host every new RPD.

Fiber-deep routing is a hub-to-node problem

Each new node needs a lawful site and a path, plus a fiber assignment that matches the DAA split. Node-boundary geometry and service-group options belong to node segmentation engineering. Cutover evidence lives in that package. We take the approved node ID as an input. We do not invent a smaller serving area to make the fiber look shorter.

Pathway occupancy from the CATV survey is an input, not a route. If the survey held overlash or duct space, the hub-to-node design shows a hold, not a guessed spare sheath. A designer who converts a held duct into a bid quantity is writing fiction that a later crew will have to explain in the street.

New laterals pick up make-ready and attachment work the analog node never needed. That work is real. It is not this post. We flag the attachment need on the route sheet so the pole-loading package can start. We do not hide it inside a fiber-deep slogan.

We keep analog optics and digital optics on separate assignment rows when both still exist during migration. A node that still takes analog light while an RPD is staged on the same site is two services, not one "upgrade fiber." Mix those labels and the cutover plan will splice the wrong buffer.

Hub-Site Handoff the Operator Must Accept

The hub face is where this design either becomes buildable or becomes an argument. We hand the operator a hub-face sheet, not a folder of unlabeled photos. Each accepted assignment names the far-end. Each held assignment names the gap and the design consequence.

• Hub assignment: site ID, tray, port, fiber ID, plus the far-end hub or node on one row.
• Diversity record: working and protect routes with shared-structure status left visible.
• DAA pool sheet: working, protect, timing, and named growth, each with a far-end.
• Hold log: missing ports, missing slack, missing occupancy, with the quantity that cannot ship.

Export is part of the design, not a later IT task. We test at least 1 representative hub-to-node assignment through the receiving GIS or CAD workflow before we call the package closed. If export strips IDs or far-ends, the native dataset is not finished. Correct field work that dies in a shapefile is still a failed handoff.

Revision history stays with the record. When a port changes after a second hub visit, keep the first observation and the reason it changed. Construction needs to see that the value moved, because a silent overwrite is how two crews argue about which tray was "always" the RPD face.

Draftech performs this engineering in-house. When construction is later included on a DAA build, delivery is full turnkey through Draftech-managed subcontract crews under our QA/QC and safety program. Field deviations still return to engineering. A splice photograph does not rewrite an unaccepted assignment.

OSP Design for Cable TV Headend Connectivity: Next Steps by Role

MSO transport planner: freeze the DAA split per hub before route production. Do not send designers a node map and a slogan. If R-PHY and R-MACPHY will coexist in one market, isolate those hubs on separate assignment sheets so a later Core change cannot rewrite the wrong CIN.

Hub-site OSP lead: accept construction quantities only where the hub-face sheet has tray, port, fiber ID, plus far-end. A dark tube is not spare. If the protect path shares a structure with the working path, send the ring back before the sheath is ordered.

DAA program owner: fund the 4 transport decisions as a gate, not as a week of CAD. The Fiber Broadband Association's April 17, 2025 passing count explains why the program exists. It does not tell you which of your hubs has an independent protect path. Spend the assignment money before the RPD capital, not after the first dark port.

Construction coordinator: schedule against hub access and approved method, not against a mile target. A locked tray is a held day. Counting it as production hides the face the splicer will ask for on Monday.

The recurring MSO transport failures we still see start with analog 2-fiber counts copied onto DAA ports. Protect fibers still ride the same duct as working fibers. Fiber-deep laterals arrive with no second path. Hub faces arrive with no far-end. We remove those by keeping the 4 transport decisions separate until each one has evidence. If a hub ring needs that register built before DAA construction, email our HFC transport team.

A cable operator that wants the transport question tested on a qualifying corridor can start with a route. We engineer the first 20,000 linear feet of a qualifying build at no cost, from feasibility and field survey through permit approval, and the 4-decision MSO transport record described above is how headend-to-hub and hub-to-node fiber enter that design. The offer is open to cable operators and ISPs. Utilities and cooperatives qualify too.