Digital Communication Techniques
Digital Communication Techniques
The word "communication" is now a catch-all in modern society; in its most basic sense, it makes it possible to share information. A department that in any French university or technical school might historically have been labeled as "humanities" (at the end of the 1960s, particularly focused on human resources or sociology); was often later reduced to "communication and humanities", both terms having become interchangeable in the meantime. Perhaps, now devoid of a clear meaning, nothing will be left apart from the term communication?
This word must not be amalgamated into others: information (transport), (en)coding. Perhaps later semantics are involved in this book, in a strict, technical sense, certainly not in any modernistic sense.
I.1. Why digitize the world?
For broadband communications, transmissions are limited by physical constraints, such as noise or interference, resulting from system imperfections and physical components modifying the transmission of the signal sent. Distortion of the signal over the course of the broadcast is, similarly, a concern. Hence, there is a need for a clear separation of the signals sent, so that, in practice, they remain distinct when they are received.
The transmission of a set of signals undergoes data dispersion over time, leading to intersymbol interference. Signals reflected from buildings, the ground or vehicles cause this dispersion, depending on the length of the paths traveled. The significance of this phenomenon depends on the frequency (above all high frequency), which can vary stochastically, via, for example, the signal's phases over time (after reflection of obstacles: echoes). They often generate signals, added destructively, or at reception. The resulting signal will therefore be very weak, or sometimes almost nonexistent. These signals can also be added constructively; the final signal will therefore be more powerful than one that arrives via a direct path. We note that multiple paths do not present only drawbacks, since they enable communication even when the transmitter and receiver are not in direct contact (for example, Transcontinental Communications).
A signal is often corrupted when it crosses different paths between transmitter and receiver: data bits that reach the receiver are subject to delays. This distorted signal will be interpreted poorly by the receiver.
In broadband communications, signals are limited by constraints: transmission errors are attenuated when the signal is digitized. For example, for the voice, the amplitude of the signal is typically measured 8,000 times per second and its value is coded in an 8-bit sequence (of 0s and 1s) - we refer here to sampling . The receiver decodes the sequence of the original signal, thus reconstructing the signal sent. Using only 0s and 1s leads to a low (or indeed non-existent) probability of error. The propagation channel can be modeled via an impulse response (see: linear system, Dirac comb); the signal received r(t) is therefore none other than the filtering of the signal sent x (T) through the propagation channel c (t) and can therefore be written in baseband, via a convolution to which noise is often added (see Langevin term added), modeling the system imperfections. Reference is made to frequency-selective channels when the signal transmitted x (t) occupies a [-W / 2, W / 2] frequency band, which is wider than the propagation channel's coherence bandwidth, c (t), (propagation channel defined as the inverse of the propagation channel's maximum delay spread Tr).
In this case, the frequential components of x(t) separated from the coherence bandwidth undergo different attenuations. In broadband digital systems, symbols are often sent at a regular interval of time T, at a maximum path delay time Tr; the signal received at an instant t can be expressed as a weighted sum (affected by path atte