Computer Networks Advanced

Layered Network Models and the Physical Layer

A message from one computer to another must pass through many separate tasks: electrical signalling on a cable, addressing, finding a path across networks, recovering from lost data, and interpreting what the data means. Writing one program that handles all of this at once would be fragile, and every change to one task would disturb the others. Network engineers solve this by dividing the work into layers. Each layer handles one job, relies on the layer beneath it for help, and offers a service to the layer above. This article introduces that layered approach through the two reference models most often used to describe it, the OSI model and the TCP/IP model, and then examines the lowest layer, the physical layer, which carries the raw bits that everything else depends on.

OSI Reference Model Layers

The Open Systems Interconnection (OSI) reference model is a seven-layer description of network communication, published by the International Organization for Standardization. A reference model is a conceptual framework used for teaching and comparison; it describes what functions must exist and how they relate, without being the protocol stack that machines actually run. The seven layers, numbered from the bottom, are physical (1), data link (2), network (3), transport (4), session (5), presentation (6) and application (7).

Layering works through two relationships. Each layer provides a service to the layer directly above it, and it obtains its own service from the layer directly below. On each host, the layers form a vertical stack. When an application sends data, each layer adds its own control information, a process called encapsulation, and passes the result downward. On the receiving host the process reverses. The key point is that each layer communicates logically with its peer, meaning the same layer on the other machine, even though the data physically travels down one stack and up the other.

Each layer has one responsibility. The lower layers move bits and deliver data across individual links and across networks. The upper layers concern whether the data is delivered reliably between processes and how the application interprets it. The detailed responsibilities of every OSI layer are covered in the child article on OSI layers.

-free note check below.

spec: static diagram of two host computers side by side, each drawn as a vertical stack of seven labelled boxes from bottom to top: 1 Physical, 2 Data Link, 3 Network, 4 Transport, 5 Session, 6 Presentation, 7 Application. Horizontal dashed arrows connect matching layers across the two hosts, labelled "peer communication" at each level. A solid vertical arrow inside each host points downward on the sending side labelled "encapsulation: each layer adds its header", and upward on the receiving side labelled "decapsulation". A physical cable line joins the two Physical layer boxes at the bottom. Title: OSI reference model, seven layers.

TCP/IP Model Layers

The TCP/IP model is the layered structure of the Internet protocol suite, which was developed for the ARPANET and became the basis of the modern Internet. It has fewer layers than OSI: the Network Access (or link) layer, the Internet layer, the Transport layer and the Application layer. Where OSI was designed first as a reference standard, TCP/IP grew out of working protocols, and its layers describe what those protocols actually do.

The defining feature of the TCP/IP model is that it is built around the Internet Protocol (IP). The Internet layer, which carries IP, is the point where traffic from many different networks is made to look the same. Above it, applications and transport protocols can be written without knowing what physical network the data crosses. Below it, any link technology that can carry IP packets can be used. For this reason the Internet layer is often described as the narrow waist of the model: many protocols sit above it and many link technologies sit below it, and IP is the single common element.

The Network Access layer covers how a host sends packets over a particular local link, including its hardware and framing. The Transport layer provides end-to-end communication between processes, and the Application layer contains the protocols that programs use directly. The specific responsibilities of each TCP/IP layer are covered in the child article on the TCP/IP model.

spec: static diagram of the TCP/IP protocol suite drawn as four stacked horizontal bands from top to bottom: Application (example labels: HTTP, SMTP, DNS), Transport (TCP, UDP), Internet (IP), Network Access (Ethernet, Wi-Fi, other link technologies). The Internet band is visually emphasised as a narrow waist: draw the Internet band narrower than the bands above and below it, with an annotation reading "IP: single common protocol". Arrows fan out from the Internet band upward to the Transport and Application bands and downward to several link technology boxes, illustrating many protocols above and many below. Title: TCP/IP model, four layers.

Mapping OSI Layers to TCP/IP Layers

The two models describe the same kind of structure, so their layers can be matched. The TCP/IP Application layer corresponds roughly to the OSI application, presentation and session layers combined, because TCP/IP leaves those concerns to the application program rather than assigning them separate layers. The TCP/IP Transport layer corresponds directly to the OSI transport layer, and the TCP/IP Internet layer corresponds directly to the OSI network layer. The TCP/IP Network Access layer covers the OSI data link and physical layers together.

The mapping is useful but not exact, and the differences matter. OSI separates session and presentation functions into their own layers, while TCP/IP does not. OSI defines its layers strictly so that each has a well-defined interface, whereas TCP/IP layers are looser and were shaped by the protocols that became dominant. Most practical discussion of networks uses the TCP/IP names, while OSI remains the common vocabulary for pinpointing where a problem or function lies. Engineers therefore often say that a router works at "layer 3", meaning the OSI network layer, which corresponds to the TCP/IP Internet layer. The detailed layer-by-layer comparison is covered in the child article on mapping the two models.

spec: static diagram showing two vertical columns of layer boxes. Left column labelled OSI, top to bottom: 7 Application, 6 Presentation, 5 Session, 4 Transport, 3 Network, 2 Data Link, 1 Physical. Right column labelled TCP/IP, top to bottom: Application, Transport, Internet, Network Access. Draw connecting lines from the right column to the left: TCP/IP Application spans OSI 5, 6 and 7 (a bracket covering all three boxes); TCP/IP Transport connects to OSI 4; TCP/IP Internet connects to OSI 3; TCP/IP Network Access spans OSI 2 and 1 (a bracket covering both boxes). Annotate the bracketed regions with "combined" to show they are one TCP/IP layer. Title: mapping OSI layers to TCP/IP layers.

Physical Layer Role in Transmitting Raw Bits

The physical layer is the lowest layer in both models. Its job is to move raw bits, the ones and zeros that the layers above produce, from one device to another across a physical medium, meaning copper cable, optical fibre or radio spectrum. It does this by converting each bit into a signal, a physical quantity such as a voltage level, a pulse of light or a radio wave amplitude, sending that signal across the medium, and converting it back into bits at the receiver. The receiver must decide, from the signal it observes, which bit was sent. Noise and attenuation on the medium make this decision imperfect, which is why signal design matters.

The physical layer deliberately does not understand what the bits mean. It does not know where one message ends and another begins, which device a bit is meant for, or whether a bit arrived corrupted in a way that matters to the sender's data. Those responsibilities belong to the data link layer, which groups bits into frames, and to higher layers. This separation is what makes the lower layers reusable: the same physical technique can carry Ethernet frames, Wi-Fi frames or any other link protocol's output without change.

Several concepts govern how fast and how reliably the physical layer can work, including the choice of signal encoding, the bit rate achievable on a medium, and the bandwidth the medium supports. These are developed in their own child articles, which explain signals, media and bandwidth in detail. For now the essential idea is that the physical layer turns an abstract sequence of bits into a physical event, transports it, and reconstructs the bits at the far end, and nothing more.

spec: static diagram of a sender device on the left and receiver device on the right connected by a horizontal line labelled "physical medium (copper cable)". Above the sender, the bit sequence 1 0 1 1 is shown as four labelled boxes. Beneath the medium, a chart-style voltage-versus-time waveform is drawn with two levels labelled "high voltage = 1" and "low voltage = 0", showing a square-wave signal representing 1 0 1 1 with time along the horizontal axis. At the right, the receiver shows the recovered bits 1 0 1 1 in boxes, with an arrow labelled "decode signal to bits". A note near the medium reads "physical layer: converts bits to signal and back; no frame, address or meaning". Title: physical layer transmitting raw bits.

AI-generated study article from SophistAI. Sign up free to study it with an AI tutor.