Introduction to Communication Systems

From the material in Chapters 4-6, we now have an understanding of commonly used modulation formats, noise models, and optimum demodulation for the AWGN channel model. Chapter 7 discusses channel coding strategies for these idealized models. In this final chapter, we discuss more sophisticated channel models, and the corresponding signal processing schemes required at the demodulator.

We first consider the following basic model for a dispersive channel: the transmitted signal passes through a linear time-invariant system, and is then corrupted by white Gaussian noise. The LTI model is broadly applicable to wireline channels, including copper wires, cable and fiber-optic communication (at least over shorter distances, over which fiber nonlinearities can be neglected), as well as to wireless channels with quasi-stationary transmitters and receivers. For wireless mobile channels, the LTI model is a good approximation over durations that are small compared with the time constants of mobility, but still fairly long on an electronic timescale (e.g., of the order of milliseconds). Methods for compensating for the effects of a dispersive channel are generically termed equalization. We introduce two common design approaches for this purpose.

The first approach is single-carrier modulation, which refers to the linear modulation schemes discussed in Chapter 4, where the symbol sequence modulates a transmit pulse occupying the entire available bandwidth. We discuss linear zero forcing (ZF) and minimum mean-squared error (MMSE) equalization techniques, which are suboptimal from the point of view of minimizing error probability, but are intuitively appealing and less computationally complex than optimum equalization.