Cable and IPTV Distribution
Cable and IPTV distribution covers the wired systems that carry television and broadband service from an operator facility to the subscriber premises. Unlike terrestrial or satellite broadcasting, where every receiver in a coverage area shares one radiated signal, wired distribution gives the operator a private, bounded transmission medium. That medium can be engineered channel by channel, segmented into small service groups, and reclaimed for new services as demand shifts. The result is a class of network that carries broadcast video, on-demand video, telephony, and high-speed data over the same physical plant.
Two technical traditions meet in this field. The first is radio-frequency carriage over coaxial cable, which began as community antenna television and matured into the hybrid fiber-coaxial plant that carries QAM channels and DOCSIS broadband today. The second is IP video, in which the program is an ordinary stream of packets on a managed network, delivered by multicast to live audiences and by unicast on demand. Modern operators run both at once, and the long-term direction of the industry is the steady conversion of the first tradition into the second: the coaxial drop survives as a physical medium, while everything above it becomes IP.
From Community Antenna Television to Broadband Plant
Cable television began in the late 1940s as community antenna television, or CATV. A well-sited master antenna picked up distant broadcast stations, an amplifier chain boosted the signal, and coaxial cable carried it down to homes in a valley or behind a ridge where over-the-air reception failed. The system was purely an analog relay: each broadcast channel appeared on the cable at a fixed frequency, and any television set could tune it.
Three changes turned that relay into a distribution industry. Operators began importing distant signals by microwave and, from the mid-1970s, by satellite, which made cable-only programming viable. Channel capacity grew as amplifier bandwidth improved, from a dozen channels to more than a hundred. Finally, digital compression in the 1990s allowed several programs to share one six-megahertz slot, and the DOCSIS specifications turned spare spectrum into a two-way data service. The plant that carries gigabit broadband today is a direct descendant of the antenna on the hill, and many of its engineering constraints—cascade noise, return-path ingress, and the frequency-dependent loss of coaxial cable—are unchanged.
Hybrid Fiber-Coaxial Plant Architecture
The hybrid fiber-coaxial (HFC) plant is the physical backbone of a cable system. Optical fiber runs from the headend or hub to an optical node in the neighborhood; coaxial cable runs from that node to the subscriber. The architecture combines the low loss and immunity of fiber over long distances with the low cost and simple powering of coaxial cable over the last few hundred meters.
Headend and Hub
The headend is the origination point. It receives programming from satellite downlinks, terrestrial off-air antennas, fiber contribution circuits, and local studios; it decodes, transcodes, and re-multiplexes that programming; and it modulates the result onto RF carriers or hands it to IP equipment. The headend also houses the cable modem termination system, video-on-demand servers, conditional access and digital rights management systems, and the network operations interfaces. Larger operators use a two-tier structure in which a regional headend feeds several hubs, each hub serving tens of thousands of homes and holding the equipment that must stay close to subscribers.
Optical Transport and the Node
Traditional HFC carried the full RF spectrum over fiber as an analog optical signal. A distributed-feedback laser at 1310 nanometers, or an externally modulated laser at 1550 nanometers for longer reaches, was driven directly by the composite RF waveform. The node contained a photodiode, a post-amplifier, and a return-path transmitter. This approach is simple but demands high optical link quality, because laser clipping and relative intensity noise translate directly into carrier-to-noise degradation on every channel.
Contemporary builds replace the analog optical link with digital Ethernet transport and move the modulation into the node, an approach described below under distributed access architectures. The fiber then carries 10-gigabit or 100-gigabit Ethernet rather than a linear replica of the RF spectrum, and optical performance ceases to limit signal quality.
Coaxial Distribution
From the node outward, the plant is a tree-and-branch coaxial network. Hard-line trunk and feeder cable, typically 0.500-inch to 0.875-inch aluminum-sheathed coaxial cable, carries the signal along the street. Directional couplers called taps bleed off a small, controlled fraction of the power at each pole or pedestal, and a flexible drop cable such as RG-6 runs from the tap to the home. Because coaxial attenuation rises roughly with the square root of frequency, the loss at 1 gigahertz is several times the loss at 50 megahertz, and every amplifier must apply tilt and equalization to restore a flat spectrum. Amplifier cascades are described by their depth—an N+5 plant places five amplifiers after the node, while an N+0 plant is entirely passive beyond the node.
The coaxial network also carries its own power. Ferro-resonant or switch-mode supplies inject a quasi-square-wave alternating voltage, commonly 60 or 90 volts, onto the center conductor through power-inserting passives; amplifiers and nodes draw from it, and standby batteries hold service through short outages. This shared power path is one reason coaxial plant remains attractive: it needs no separate utility feed at each active device.
Service Groups and Node Splitting
All subscribers fed by one node share the same downstream spectrum and contend for the same upstream spectrum. That population is the service group, and its size is the primary capacity lever available to an operator. Splitting a node in two—by adding fiber, adding an optical receiver, or segmenting the node internally—halves the number of homes sharing a channel and immediately doubles the capacity available per home. Pushing fiber deeper and shortening amplifier cascades, often called a fiber-deep or node-plus-zero strategy, improves noise performance at the same time and prepares the plant for higher spectrum limits.
Spectrum Planning and Frequency Splits
Cable spectrum is a shared, finite resource, and its division between upstream and downstream defines what services the plant can offer. The dividing point is set by diplex filters inside every amplifier and node, so changing it is a physical upgrade rather than a configuration change.
North American plants historically used a sub-split return of 5 to 42 megahertz, with downstream service starting near 54 megahertz. European plants used 5 to 65 megahertz. As upstream demand grew, operators moved to a mid-split of 5 to 85 megahertz and then a high-split of 5 to 204 megahertz, in which downstream carriage begins at 258 megahertz. DOCSIS 4.0 defines further options, adding ultra-high splits with upstream limits of 396, 492, or 684 megahertz. At the top end, plants have progressed from 550 to 750, 860, and 1002 megahertz. DOCSIS 3.1 raised the downstream band edge to 1218 megahertz, the usual target for recent rebuilds, and DOCSIS 4.0 extends it to 1794 megahertz, the figure ordinarily quoted as 1.8 gigahertz.
Within that range, spectrum is divided into channels: 6 megahertz in systems following North American practice and 8 megahertz in systems following European practice. A modern plant mixes legacy single-carrier QAM channels, DOCSIS OFDM blocks up to 192 megahertz wide, and, on many systems, a shrinking set of analog carriers retained for legacy equipment. Reclaiming spectrum—retiring analog channels, converting broadcast video to IP, or moving on-demand traffic into DOCSIS—is a continuous operational activity rather than a one-time migration.
QAM Carriage of Digital Video
Digital cable television carries programs as MPEG transport streams modulated onto RF carriers with quadrature amplitude modulation. QAM encodes data in both the amplitude and the phase of a carrier, so a 256-QAM constellation carries eight bits per symbol. Cable is a comparatively benign channel—shielded, with a high signal-to-noise ratio and no fading—so it supports constellation densities that would be impossible over the air.
Two profiles dominate. ITU-T J.83 Annex B, used in North America, applies concatenated Reed-Solomon and trellis coding with interleaving and delivers roughly 26.97 megabits per second at 64-QAM and 38.8 megabits per second at 256-QAM in a 6-megahertz channel. ITU-T J.83 Annex A, the basis of the DVB-C standard, uses Reed-Solomon coding without trellis coding and delivers approximately 51 megabits per second at 256-QAM in an 8-megahertz channel. DVB-C2, the second-generation European standard, replaces single-carrier QAM with OFDM and low-density parity-check coding for substantially better spectral efficiency, though its deployment has been limited because DOCSIS has absorbed most new capacity requirements.
One QAM channel typically carries several television services multiplexed together. A statistical multiplexer allocates bit rate dynamically across those services, giving more to a fast-moving sports feed and less to a static studio shot, which raises the average quality achievable within a fixed channel. Program-specific information tables, along with system information defined by ANSI/SCTE and DVB, let the set-top box find each service; out-of-band signaling channels carry conditional access entitlements, channel maps, and emergency alert messages to receivers.
Signal quality on a QAM channel is judged by modulation error ratio and by pre- and post-correction bit error rate, with the constellation display used to identify the underlying impairment. Phase noise rotates the constellation, compression spreads the outer points, and a strong single interferer produces characteristic rings or doughnuts around each symbol point.
DOCSIS and Cable Broadband
The Data Over Cable Service Interface Specification, developed by CableLabs, defines how cable modems and headend equipment share plant spectrum for two-way IP service. Its successive generations show a clear pattern: each adds spectrum, adds modulation efficiency, and improves the coding that makes higher-order modulation survivable.
Generations
DOCSIS 1.0 established the basic scheme of a downstream broadcast channel and a time-division upstream managed by grants from the headend. DOCSIS 1.1 added service flows, classifiers, and quality-of-service scheduling, which made packet telephony over cable practical. DOCSIS 2.0 improved the upstream with 64-QAM and an optional synchronous code-division multiple access mode for noisy plants.
DOCSIS 3.0 introduced channel bonding, in which a modem transmits and receives across multiple channels simultaneously—up to thirty-two downstream and eight upstream channels—raising throughput above a gigabit per second downstream and to roughly 200 megabits per second upstream. It also added IPv6 support and IP multicast improvements.
DOCSIS 3.1, issued in 2013, replaced single-carrier modulation with orthogonal frequency-division multiplexing downstream and OFDMA upstream, using subcarriers spaced 25 or 50 kilohertz apart within blocks up to 192 megahertz downstream and 96 megahertz upstream. Low-density parity-check coding replaced the older concatenated scheme, which made far denser constellations survivable: downstream modulation reaches 4096-QAM as the mandatory ceiling for modems, with 8192-QAM and 16384-QAM defined as options, and the upstream reaches 1024-QAM with higher orders optional. Crucially, DOCSIS 3.1 assigns modulation profiles per group of modems, so subscribers on clean plant receive dense constellations while those on impaired drops fall back without penalizing everyone else.
Capacity claims for DOCSIS 3.1 need care, because the specification ceiling and the deployed figure differ. The headline number is about 10 gigabits per second downstream and 1 to 2 gigabits per second upstream, but the downstream figure assumes the entire 1218-megahertz plant given over to OFDM at the densest profiles. CableLabs describes current DOCSIS 3.1 modems as supporting up to 5 gigabits per second downstream and 1.5 gigabits per second upstream, and retail tiers are lower again, because spectrum is still shared with QAM video and with every other subscriber in the service group.
DOCSIS 4.0, published in 2020, targets multi-gigabit symmetrical service and admits two alternative approaches within one specification. Extended-spectrum DOCSIS, which the specification treats as its frequency-division duplex mode, raises the downstream band edge to 1794 megahertz and the upstream limit to 396, 492, or 684 megahertz, keeping the two directions in separate bands. Full-duplex DOCSIS instead lets upstream and downstream share the 108-to-684-megahertz band at the same time, relying on echo cancellation in the node and on interference-group scheduling, which assigns modems that can hear one another to different frequencies at any given moment.
Full duplex was first conceived for passive plant beyond the node, because early designs could not cancel echo through an amplifier. That constraint has since been lifted: echo-cancelling full-duplex amplifiers are covered by SCTE 296 and are in commercial service, so full duplex is no longer confined to node-plus-zero builds. Operators have divided between the two options, with Comcast building on full duplex and Charter pursuing extended spectrum, and vendors have begun offering node electronics that support either mode.
CableLabs states a maximum capacity for DOCSIS 4.0 of up to 10 gigabits per second downstream and up to 6 gigabits per second upstream. Those are specification ceilings for a fully equipped plant rather than service rates; the first commercial DOCSIS 4.0 offerings launched at symmetrical speeds of up to 2 gigabits per second. The 10G label that appears in cable industry marketing names a program of network upgrades, not a modulation scheme, a subscriber data rate, or a DOCSIS version number.
Modem Initialization and Security
A cable modem coming online scans for a downstream carrier, acquires timing and MAC-layer synchronization, and then performs initial ranging: it transmits a burst in a contention window, and the headend replies with corrections to its power level, timing offset, and frequency. Pre-equalization coefficients returned during ranging compensate for micro-reflections on that specific drop. The modem then obtains an IP address by DHCP, downloads a configuration file by TFTP that defines its service flows and rate limits, and registers. Baseline Privacy Interface Plus authenticates the modem with an installed X.509 certificate and encrypts subscriber traffic over the shared medium, which prevents both eavesdropping and theft of service.
Latency and Scheduling
Upstream transmission is scheduled: the headend issues bandwidth allocation maps that assign transmit opportunities, and modems either request grants or use unsolicited grants for constant-rate services such as voice. The request-grant cycle is the main contributor to upstream latency, and buffering in the modem adds more under load. Low Latency DOCSIS addresses this with a dual-queue design that separates queue-building traffic from latency-sensitive traffic, adds active queue management and queue protection, and supports the L4S congestion-signaling framework. The result is a marked reduction in working latency for interactive applications without reserving dedicated bandwidth.
CMTS, CCAP, and Virtualized Cores
The cable modem termination system is the headend counterpart to the cable modem. It terminates DOCSIS service flows, routes subscriber traffic to the operator network, schedules upstream transmissions, enforces quality-of-service policy, manages modem registration and security, and reports the plant telemetry that operations teams depend on.
Early plants ran separate CMTS and edge-QAM equipment, the latter modulating video-on-demand and switched digital video streams. The Converged Cable Access Platform consolidated both functions into one chassis, simplifying spectrum management and reducing the number of RF combining stages in the headend.
The current direction is virtualization. A virtual CMTS, or vCMTS, runs the DOCSIS MAC and control software as containerized workloads on commercial off-the-shelf servers, pairing naturally with distributed access architectures that have already moved the physical layer out to the node. Virtualization decouples capacity growth from chassis replacement, allows software features to be deployed independently of hardware, and lets operators scale control-plane and data-plane resources separately. It also shifts the operational challenge toward data-center practice: server lifecycle management, container orchestration, and the deterministic packet processing needed to meet DOCSIS timing requirements.
Distributed Access Architectures
A distributed access architecture (DAA) relocates part of the access equipment from the headend into the optical node, replacing the analog optical link with digital Ethernet transport over a converged interconnect network.
Remote PHY
In the Remote PHY approach, the physical layer—modulation, demodulation, and RF conversion—moves into a Remote PHY device inside the node, while the DOCSIS MAC and all higher layers remain in the headend core. The core and the node exchange packets over two tunneling interfaces: the Downstream External PHY Interface carries downstream data toward the node, and the Upstream External PHY Interface carries received upstream data back. The Generic Control Plane protocol configures and monitors the remote device. Because the MAC scheduler is now separated from the PHY by a packet network, timing must be distributed explicitly, using IEEE 1588 Precision Time Protocol and the DOCSIS Timing Protocol.
Remote MACPHY and Flexible MAC Architecture
Remote MACPHY places both the MAC and the PHY in the node, leaving only routing and management functions upstream. This shortens timing paths and reduces the volume of traffic on the interconnect network, at the cost of more complex, more power-hungry, and less centrally manageable node electronics. CableLabs later defined the Flexible MAC Architecture to describe how MAC functions may be split and placed—in the node, in a headend appliance, or in software—under a common management model, so that an operator can mix placements within one network.
Why Operators Adopt DAA
The practical benefits are consistent. Digital optics remove analog link noise and laser clipping from the signal budget, which improves modulation error ratio and permits denser constellations. Headend space and power drop sharply because RF combining networks and analog transmitters disappear. The fiber between hub and node becomes ordinary Ethernet, so it can be shared with business services and mobile transport. Finally, DAA is a prerequisite for practical vCMTS deployment, because a virtualized core running on commodity servers has no radio-frequency output of its own, and it is the practical basis for the high-split and extended-spectrum node electronics that DOCSIS 4.0 upgrades depend on.
Plant Impairments and Proactive Maintenance
Cable plant is exposed to weather, physical damage, corrosion, and amateur wiring inside homes, and its impairments have well-known signatures.
Micro-reflections arise from impedance mismatches at connectors, damaged cable, or unterminated splitter ports; they produce standing-wave ripple in the frequency response and inter-symbol interference that adaptive equalizers must correct. Ingress is external signal entering through shielding faults, and because the tree-and-branch topology funnels every home's upstream noise back to one receiver, a single defective drop can raise the noise floor for an entire service group. Common-path distortion, caused by corroded connections acting as diodes, generates beat products across the return band. Group delay concentrates near the diplexer band edges, where it degrades the outermost carriers. Laser clipping in analog optical links produces bursts of errors across all channels at once.
Proactive network maintenance uses the measurements that DOCSIS equipment already produces to find these faults before subscribers report them. Pre-equalization coefficients reported by every modem reveal the delay and depth of reflections on that drop, and clustering modems with similar signatures localizes the fault to a shared passive. Full-band capture turns the modem tuner into a spectrum analyzer, showing ingress and suck-outs from the subscriber's own location. On DOCSIS 3.1 plant, per-subcarrier receive modulation error ratio gives a detailed spectral picture of impairment, and profile management applications adjust modulation profiles automatically as conditions change. These techniques convert maintenance from a reactive, truck-roll-driven activity into a data-driven one.
IPTV over Managed Networks
IPTV is television delivered as IP packets over a network that the service provider controls end to end. The defining characteristic is not the protocol but the management: the operator engineers capacity, applies quality-of-service policy, and admits sessions, so the video path is protected from the congestion that affects ordinary internet traffic. Telephone companies adopted IPTV to compete with cable over DSL and, later, fiber-to-the-home; cable operators use the same techniques to deliver IP video over DOCSIS.
System Components
A managed IPTV system begins with acquisition and encoding: satellite and fiber feeds are decoded, then encoded to H.264, HEVC, or AV1 at the bit rates the access network can support. Live channels are packaged as MPEG transport streams carried in RTP or UDP, or as segmented CMAF media for IP-native clients. Video-on-demand servers hold assets and stream them by unicast. A middleware platform provides the electronic program guide, subscriber entitlement, session management, and the application environment on the set-top box. Conditional access or digital rights management protects the content, and network probes monitor delivery quality at aggregation points.
The set-top box terminates the stream: it joins multicast groups, buffers to absorb jitter, decrypts, decodes, and renders to the television. Increasingly, that box is a general-purpose streaming device running the operator's application rather than a dedicated video receiver, which blurs the technical boundary between managed IPTV and internet streaming.
Access Network Options
IPTV rides on whatever access technology the operator owns. VDSL2 with vectoring supports a few high-definition streams over short copper loops. Passive optical networks—GPON, XGS-PON, and successors—provide symmetric multi-gigabit capacity and are now the dominant new build for telephone companies. Over cable plant, IP video shares DOCSIS capacity with broadband, which is why spectrum reclamation from legacy QAM video matters so much to operators pursuing all-IP delivery. Fixed wireless access adds a further option where trenching is uneconomic.
Multicast Delivery and Channel Change
Live television is the case that multicast was designed for: many subscribers want the same stream at the same instant. With IP multicast, the source sends one copy, and routers replicate packets only where paths diverge, so the load on the core is independent of audience size.
Receivers signal interest with the Internet Group Management Protocol—IGMPv2, or IGMPv3 for source-specific joins—and with Multicast Listener Discovery on IPv6. Within the operator network, Protocol Independent Multicast, usually in source-specific multicast mode, builds the distribution tree. Access nodes such as DSLAMs, optical line terminals, and Ethernet switches perform IGMP snooping so that a channel is forwarded onto a subscriber port only while someone is watching, and admission control limits how many streams a given link will accept.
Two problems dominate the engineering. The first is channel-change latency: joining a group, waiting for a random access point in the video, and filling the decoder buffer can take seconds. The standard remedy is a unicast burst that delivers the recent past of the channel at faster than real time while the multicast join completes, an approach standardized as unicast-based rapid acquisition of multicast RTP sessions. The second is packet loss, which multicast cannot repair by retransmission from the source. Operators use application-layer forward error correction, as defined in SMPTE ST 2022-1 and DVB-IPTV, and unicast RTP retransmission from a local server, so a receiver can request just the packets it missed.
Multicast adaptive bitrate delivery combines the two worlds. The operator distributes the most popular channels as multicast to a gateway in the home or at the network edge, which then republishes the segments over HTTP to ordinary adaptive-bitrate clients. The client software is unchanged, but the network carries one copy of a popular live stream rather than thousands.
Content Protection and Advertising Systems
Cable and IPTV systems protect content at the transport layer and at the application layer. Traditional conditional access encrypts the transport stream and delivers entitlement messages to a smart card or embedded secure element in the set-top box. Simulcrypt, standardized by DVB, allows one scrambled stream to be accompanied by entitlement messages from several conditional access vendors, so an operator may serve a mixed population of receivers. In North America, the CableCARD regime separated conditional access from the retail receiver for a period before regulatory support ended and IP delivery superseded it.
IP-delivered video instead uses digital rights management systems—Widevine, PlayReady, and FairPlay are the widely deployed examples—with keys delivered over HTTPS to a client that enforces output and security policy. Common Encryption allows one encrypted media file to be consumed under multiple DRM systems, which is what makes a single packaged asset serviceable to many device types.
Advertising and program signaling ride alongside. SCTE 35 messages inserted in the transport stream or the manifest mark splice points, and downstream systems replace national advertisements with local or addressable ones. SCTE 224 carries the event and policy metadata—blackouts, regional rights, alternate content—that determines what a given subscriber is permitted to see at a given moment.
Quality of Service and Quality of Experience
Managed video imposes strict transport requirements. Constant-bitrate transport streams tolerate very little loss: a single lost packet destroys a slice of a picture, and losses within a reference frame propagate visibly until the next random access point. Jitter must stay within the receiver buffer, and a plant-wide timing reference must keep encoder and decoder clocks aligned.
Operators measure delivery with the Media Delivery Index, which reports a delay factor describing buffering requirements and a media loss rate describing lost or out-of-order packets, and they enforce priority with DiffServ marking and per-service-flow scheduling in DOCSIS or PON equipment. Adaptive-bitrate video is more forgiving of loss because TCP or QUIC retransmits, but it converts network trouble into visible quality reduction and rebuffering instead. Quality-of-experience monitoring therefore tracks client-side metrics—startup time, rebuffer ratio, average delivered bitrate, and playback failures—and correlates them with network telemetry to locate the cause.
Over-the-Top Streaming and Its Relationship to Managed IPTV
Over-the-top (OTT) video travels across the public internet without any arrangement between the video provider and the subscriber's access provider. It differs from managed IPTV in three respects rather than in fundamental technology. Delivery is unicast over HTTP rather than multicast over RTP. Adaptation is performed by the client, which selects among encoded renditions listed in a manifest, rather than by network admission control. And responsibility for quality is shared and unenforceable: the video provider controls encoding and content delivery network placement, while the access provider controls the last mile.
The relationship is more cooperative than competitive at the infrastructure level. Access operators host caching nodes from major streaming services inside their own networks, which shortens delivery paths and reduces transit costs for both parties. Open caching specifications aim to standardize that arrangement so one operator cache can serve many content providers. Meanwhile, cable and telephone operators have converted their own services to the same adaptive-bitrate technology, delivering their channel lineups to applications on retail streaming devices and smart televisions.
The economic consequence has been substantial. Subscribers have moved away from bundled channel packages, and operators have responded by treating broadband as the primary product and video as an application over it. That reframing explains most current investment priorities: spectrum reclaimed from QAM video becomes DOCSIS capacity, distributed access architectures and virtualized cores reduce the cost of that capacity, and the set-top box gives way to software on devices the operator does not own.
Practical Engineering and Troubleshooting Guidance
A few habits distinguish reliable installations and quick diagnoses in this field.
Begin at the physical layer. Most subscriber-reported faults on coaxial plant come from connectors, splitters, and drop cable rather than from headend equipment. Check for corroded or improperly compressed connectors, unterminated splitter ports, and damaged shielding; verify that the drop is properly grounded and bonded. Confirm downstream receive level and modulation error ratio at the modem before investigating anything upstream of it.
Read the constellation and the error counters together. A high pre-correction error rate with a clean post-correction rate signals an impairment that the forward error correction is currently absorbing, which is a warning rather than an all-clear; the margin will disappear as conditions worsen. Distinguish uncorrectable bursts, which suggest ingress or laser clipping, from steady degradation, which suggests level or noise problems.
Treat upstream problems as service-group problems. Because return-path noise funnels, a fault in one home affects every subscriber on the node. Correlating full-band capture and pre-equalization data across many modems isolates the shared passive far faster than sequential site visits.
For IP video, separate transport faults from encoding faults. Macroblocking that is identical for every subscriber on a channel points to the encoder or the source feed; macroblocking confined to one access node points to loss or jitter in that segment. Media Delivery Index probes placed at successive aggregation points narrow the location quickly.
Finally, plan spectrum and service groups together. Adding channel capacity to an oversubscribed service group yields less improvement than splitting the group, and a split also lowers the noise contribution seen upstream. Capacity planning should model peak concurrent demand per service group rather than average traffic per subscriber.
Summary
Cable and IPTV distribution is the engineering of wired, operator-controlled paths for video and broadband. On the cable side, the hybrid fiber-coaxial plant carries QAM video and DOCSIS data over a shared spectrum whose split between upstream and downstream is set by physical filters; the cable modem termination system schedules that spectrum, and its functions are steadily migrating outward into the node and downward into software. On the IPTV side, video becomes ordinary packets on a managed network, delivered by multicast for live channels and unicast on demand, with quality assured by capacity engineering rather than by radio-frequency headroom.
The two converge. Distributed access architectures and virtualized cores make the cable plant a packet network with a radio-frequency last hop; multicast adaptive bitrate and operator applications on retail devices make managed video look like streaming. What persists is the underlying discipline: a fixed medium with measurable impairments, a shared resource that must be allocated fairly, and a subscriber whose judgment of the service is formed by startup time, picture quality, and reliability rather than by any specification.