3.1 INTRODUCTION

Communication systems are designed to transmit information generated by a source to a destination in the most efficient and reliable manner possible. Information sources may take many forms, including audio (speech or music), video, or textual data. Source coding concerns the process of forming an efficient representation of that information so that it can be conveyed using the fewest possible bits while maintaining an acceptable level of fidelity.

In a digital communication system, the input to a source coder may be an analog signal, while the output is a binary bitstream suitable for transmission and storage. This conversion—commonly referred to as analog-to-digital (A/D) conversion—can be achieved using waveform coding, model-based coding, or a combination of both approaches.

In waveform coding, the signal waveform itself is sampled and quantized, with the objective of reproducing the received waveform as accurately as possible. These techniques make little or no use of prior knowledge about the nature of the source signal and instead aim to preserve the waveform shape directly while minimizing distortion. Common examples include pulse-code modulation (PCM), differential PCM (DPCM), and delta modulation (DM).

In model-based coding, the encoder does not transmit individual waveform samples. Instead, it exploits knowledge of the signal generation process to extract a compact set of parameters that describe the essential characteristics of the source. The receiver then uses these parameters to synthesize or approximate the original signal. This approach is particularly effective for structured signals such as speech, images, and video, where the underlying production mechanisms are well understood—for example, linear predictive coding (LPC) for speech or motion-compensated coding for video

A fundamental objective of source coding—whether waveform-based or model-based—is to achieve a compact and bandwidth-efficient representation of the source. By reducing the number of bits required to describe the information, a communication system can either decrease the transmission bandwidth required or increase the number of simultaneous channels that can be supported within a fixed bandwidth.

In this chapter, we examine how these principles are realized through quantization, compression, and entropy coding, and how source coding interacts with channel coding to balance efficiency against robustness to noise and errors in practical communication systems.