Library

14.3.5 Physical Media For Ethernet

Early Ethernet implementations used coaxial cable, which was quickly superseded by structured twisted-pair topologies. Although coaxial cable remains widely used in broadband cable distribution systems, it is no longer used in modern enterprise Ethernet LAN deployments. Contemporary Ethernet installations employ structured copper and fiber-optic cabling arranged in a physical star topology centered on switches. Unlike early shared coaxial systems, modern Ethernet uses point-to-point links between each device and a switch port.

Each transmission medium supports specific data rates and maximum link distances as defined by the corresponding IEEE 802.3 PHY specification. In contemporary LAN deployments, twisted-pair copper predominates at the access layer, while fiber-optic links are commonly used for backbone and inter-switch connections. The choice of medium affects achievable reach, bandwidth, electromagnetic immunity, installation complexity, and scalability.

We therefore begin with twisted-pair copper cabling before examining fiber-optic implementations and legacy media.

14.3.5.1 Twisted-Pair Copper Cabling

The most common medium for connecting end devices to Ethernet switches is balanced twisted-pair copper cable. Twisting the conductors reduces electromagnetic interference and crosstalk between adjacent pairs, enabling reliable high-speed transmission over cost-effective cabling. Modern installations follow structured cabling standards and use 8P8C modular connectors (commonly referred to as RJ-45).

Contemporary twisted-pair Ethernet operates in full-duplex mode. At lower data rates, separate pairs are used for transmit and receive; at higher rates (for example, 1000BASE-T and above), simultaneous bidirectional signaling occurs on all four pairs using advanced digital signal processing.

Common cable categories include:

The standard maximum horizontal channel length in structured installations is 100 m, typically comprising 90 m of fixed horizontal cabling plus up to 10 m of patch cords. This limit is determined by attenuation, crosstalk, return loss, and the signal-to-noise ratio required by the relevant PHY specification.

As data rates increase, copper implementations rely on progressively more sophisticated techniques, including multi-level pulse amplitude modulation (for example, PAM-5 and PAM-16), echo cancellation, crosstalk mitigation, and forward error correction (FEC).

Twisted-pair copper remains cost-effective and is therefore widely deployed at the network edge in offices, campuses, and industrial environments.

14.3.5.2 Fiber-Optic Ethernet

Fiber-optic cable is used where higher bandwidth, longer distance, or immunity to electromagnetic interference is required. Fiber links eliminate grounding problems and are well suited to data centers, backbone interconnections, and electrically noisy environments.

Two principal fiber types are used in Ethernet deployments:

Modern Ethernet standards over fiber support speeds including 1, 10, 25, 40, 100, and 400 Gbps and beyond. At very high data rates, multiple parallel optical lanes or wavelength-division multiplexing techniques may be employed within standardized Ethernet PHY definitions.

14.3.5.3 Power Over Ethernet (PoE)

Modern twisted-pair Ethernet supports Power over Ethernet (PoE), allowing electrical power to be delivered along with data over the same cable. In PoE systems, DC power (typically 44–57 V) is applied as a common-mode voltage to the twisted pairs used for data transmission. Because Ethernet signaling is differential and transformer-coupled, the DC component does not interfere with the AC data signal. Modern PoE standards distribute power across all four pairs and provide power levels ranging from approximately 15 W to 90 W, depending on the IEEE 802.3 variant implemented.

PoE enables switches to power devices such as wireless access points, IP telephones, surveillance cameras, IoT gateways, and building automation equipment. It simplifies installation by eliminating separate power supplies and supports centralized power management and backup.

14.3.5.4 Structured Cabling And Hierarchical Deployment

Modern Ethernet installations follow structured cabling principles in which horizontal copper (and sometimes fiber) links connect work areas to wiring closets, and fiber links interconnect distribution and core switches. This hierarchical approach improves scalability, manageability, and fault isolation.

Unlike early bus-based Ethernet, modern Ethernet is entirely point-to-point at the physical layer, with switching providing logical interconnection among devices. In this architecture, each link between an end device and a switch port forms its own independent point-to-point connection. Because modern Ethernet operates in full-duplex mode, collisions do not occur and CSMA/CD is disabled.

The historical concept of a “segment” as a shared collision domain therefore no longer applies. Earlier timing rules governing repeater counts, maximum network diameter, and user limits—such as the 5-4-3 rule—have disappeared from practical network design. Link length is determined by the physical-layer specification for the selected medium and data rate. For copper cabling, the standard maximum channel length is 100 m (typically 90 m of horizontal cable plus up to 10 m of patch cords). For fiber, permissible distance is determined by optical power budget, attenuation, and the PHY standard employed.

Scalability is governed not by collision-domain constraints but by switch port density, switching capacity, uplink bandwidth, and hierarchical network architecture. The physical medium determines signaling characteristics and maximum link distance, while Ethernet switching determines how frames are forwarded within the LAN.

Having examined both the logical operation of switching and the physical media that support Ethernet, we now consider how these elements are organized into modern LAN topologies.