At the very bottom of the networking stack, all communication reduces to transmitting bits — zeros and ones — across a physical medium. The medium determines what kind of signal carries the information.

Signaling Methods

There are three primary physical signaling methods:

• Electrical (copper) — Voltage levels on a wire represent binary values. For example, +5V might represent a 1 and 0V represents a 0. Ethernet cables (Cat5e, Cat6) and coaxial cables use this approach. Copper is inexpensive but susceptible to electromagnetic interference (EMI) and signal degradation over distance.
• Optical (fiber) — Pulses of light travel through glass or plastic fibers. Light on = 1, light off = 0. Fiber optic cables offer enormous bandwidth, immunity to EMI, and can span tens of kilometers without repeaters. Single-mode fiber uses a single light path for long distances; multi-mode fiber uses multiple paths for shorter runs.
• Wireless (radio) — Radio frequency electromagnetic waves carry data through air. Modulation techniques (amplitude, frequency, phase) encode bits onto a carrier wave. Wi-Fi, Bluetooth, and cellular networks all use radio signaling. Wireless is convenient but faces interference, multipath fading, and shared-medium contention.

Analog vs Digital Signals

An analog signal varies continuously over time — think of a smooth sine wave. A digital signal is discrete, jumping between a finite set of levels (typically two: high and low). Real-world transmission always involves analog waveforms; the receiver must interpret the analog signal and recover the original digital bits. This process is called signal detection or demodulation.

Attenuation and Noise

As a signal travels through any medium, it loses strength — this is attenuation. The longer the cable or the higher the frequency, the greater the attenuation. Additionally, noise (random electrical disturbances, crosstalk from adjacent wires, thermal noise) corrupts the signal. The ratio of signal power to noise power is the Signal-to-Noise Ratio (SNR), usually measured in decibels (dB). A higher SNR means a cleaner signal and fewer bit errors.

Nyquist Theorem

The Nyquist theorem defines the maximum symbol rate for a noiseless channel: maximum symbol rate = 2 * bandwidth (in Hz). If each symbol encodes b bits (using 2^b signal levels), the maximum bit rate becomes 2 * bandwidth * b. For example, a 4 kHz channel using 16 signal levels (4 bits per symbol) can carry at most 2 * 4000 * 4 = 32,000 bps. The Nyquist limit assumes no noise — in practice, noise limits how many distinct signal levels the receiver can reliably distinguish.

Shannon Capacity

The Shannon-Hartley theorem gives the absolute maximum bit rate for a noisy channel: C = B * log2(1 + SNR), where C is channel capacity in bits per second, B is bandwidth in Hz, and SNR is the linear (not dB) signal-to-noise ratio. This is a theoretical upper bound — no encoding scheme can exceed it. For example, a channel with 3 kHz bandwidth and 30 dB SNR (linear ratio 1000) has capacity = 3000 * log2(1001) ~ 30,000 bps.

Bit Rate vs Baud Rate

Baud rate is the number of symbol changes per second. Bit rate is the number of bits transmitted per second. If each symbol carries exactly one bit, baud rate equals bit rate. But with multi-level signaling (e.g., QAM-16 uses 16 levels = 4 bits per symbol), the bit rate is baud rate * bits-per-symbol. A modem running at 2400 baud with 16-QAM achieves 2400 * 4 = 9600 bps.