3.8.1 What Is Source Coding?
- Why Do Communications Systems Need Source Coding?
- What Types of Information Can Be Source Coded?
- How Does Source Coding Improve Efficiency?
- What Is Redundancy?
- Is All Redundancy Unnecessary?
- What Is the Difference Between Lossless and Lossy Coding?
- How Does Source Coding Relate to Analog-to-Digital Conversion?
- What Are Waveform Coders?
- What Are Model-Based Coders?
- What Are Hybrid Coders?
- How Much Compression Is Possible?
- Where Is Source Coding Used Today?
- How Does Source Coding Fit into a Communications System?
- Why Is Source Coding Important?
Every communications system exists for a single purpose: to transfer information from a source to a destination. The source may be a person speaking into a telephone, a computer generating data, a camera capturing an image, or a microphone recording music. Regardless of the source, the information must be represented in a form that can be transmitted, stored, and reconstructed at the destination.
One of the fundamental challenges in communications engineering is doing this efficiently. If information can be represented using fewer bits, less transmission bandwidth is required, storage requirements are reduced, and more users can share the same communications infrastructure. The process of converting information into an efficient representation is known as source coding.
Source coding forms one of the most important building blocks of modern communications systems. It underpins digital telephony, music streaming, television broadcasting, video conferencing, Internet communications, and data storage systems. Whether transmitting speech across a telephone network, streaming a movie, or storing photographs on a smartphone, source coding helps reduce the amount of information that must be transmitted or stored while maintaining an acceptable level of quality.
Why Do Communications Systems Need Source Coding?
At first glance, it might seem that the simplest approach would be to transmit information exactly as it is generated. In practice, however, this is rarely efficient.
Consider a voice signal. The original analog waveform contains an effectively infinite number of amplitude values and changes continuously with time. Transmitting the waveform directly would require considerable bandwidth and would be highly susceptible to noise.
Similarly, a digital image may contain millions of pixels, while a video stream may contain thousands of images every minute. Transmitting all of this information without compression would require enormous data rates.
Source coding addresses this problem by creating a more compact representation of the information. The objective is to reduce the number of bits required while preserving the information needed by the receiver.
The benefits include:
- Reduced bandwidth requirements.
- Reduced storage requirements.
- Lower transmission costs.
- Increased network capacity.
- Faster transmission times.
These advantages make source coding an essential component of modern communications systems.
What Types of Information Can Be Source Coded?
Almost any form of information can be source coded. Common examples include:
- Speech. Telephone systems, mobile networks, and Voice-over-IP services all employ source coding to reduce the number of bits required to represent speech.
- Audio. Music streaming services and digital audio systems use source coding to compress sound recordings.
- Images. Digital cameras, websites, and image archives use source coding to reduce image file sizes.
- Video. Television broadcasting, video conferencing, and streaming services rely heavily on source coding to reduce transmission bandwidth.
- Computer data. Text, documents, software, and databases can all be compressed using source-coding techniques.
Although the details vary between applications, the underlying objective remains the same: represent information using as few bits as possible.
How Does Source Coding Improve Efficiency?
Source coding improves efficiency by exploiting redundancy.
Many information sources contain repeated or predictable patterns. If these patterns can be identified, they can often be represented more compactly.
For example, consider the following sequence of characters AAAAAA. Instead of transmitting six separate letters, the sequence could be represented as 6 × A. The information content remains the same, but fewer symbols are required.
Real-world compression techniques use far more sophisticated methods, but they are based on the same principle: eliminate unnecessary repetition and represent information more efficiently.
What Is Redundancy?
Redundancy refers to information that can be predicted from other information already available. Natural information sources often contain considerable redundancy. For example:
- Adjacent pixels in an image are often similar.
- Successive frames in a video are frequently nearly identical.
- Consecutive speech samples are highly correlated.
- Certain letters and words occur more frequently than others in written language.
Because these patterns are predictable, they do not need to be represented independently.
Source coding exploits this predictability to reduce the number of bits required.
Is All Redundancy Unnecessary?
Not necessarily. Some redundancy is useful and even essential.
For example, human language contains significant redundancy that helps listeners understand speech in noisy environments. Similarly, communications systems often deliberately add redundancy through channel coding to improve reliability.
The purpose of source coding is therefore not to eliminate all redundancy, but rather to remove redundancy that does not contribute significantly to the desired information.
This distinction becomes important when comparing source coding with channel coding, which is discussed later in this chapter.
What Is the Difference Between Lossless and Lossy Coding?
Source-coding techniques are commonly divided into two categories:
- Lossless coding. Lossless coding allows the original information to be reconstructed exactly. No information is lost during compression. Lossless coding is essential when exact reconstruction is required. Examples include: ZIP files, PNG images, many text-compression systems.
- Lossy coding. Lossy coding deliberately discards information that is considered unimportant or imperceptible. The reconstructed signal is not identical to the original, but the differences may be difficult for users to detect. Examples include: MP3 audio, JPEG images, most video compression systems.
Lossy coding generally achieves much greater compression than lossless coding. The choice depends on the application's quality requirements.
How Does Source Coding Relate to Analog-to-Digital Conversion?
In many communications systems, the original information exists in analog form. Examples include:
- Speech.
- Music.
- Video signals.
- Sensor outputs.
Before such signals can be transmitted digitally, they must first be converted into binary form.
This process typically involves:
- Sampling.
- Quantization.
- Encoding.
The resulting bitstream is then suitable for digital transmission and storage.
Techniques such as pulse-code modulation (PCM), differential PCM (DPCM), and delta modulation (DM) are examples of source-coding methods that perform this conversion.
What Are Waveform Coders?
One major class of source coders attempts to preserve the shape of the original signal waveform. These are known as waveform coders. Rather than relying on detailed knowledge of how the signal was generated, waveform coders simply attempt to reproduce the signal as accurately as possible. Examples include:
- Pulse-code modulation (PCM).
- Differential PCM (DPCM).
- Adaptive differential PCM (ADPCM).
- Delta modulation (DM).
Waveform coders generally provide good signal quality but often require relatively high bit rates.
What Are Model-Based Coders?
A second class of source coders exploits knowledge about how the source generates information. These are known as model-based coders. Instead of transmitting waveform samples directly, the encoder extracts a compact set of parameters describing the source.
The receiver then uses these parameters to synthesize an approximation of the original signal. Speech coding provides a good example. Rather than transmitting every speech sample, a model-based speech coder may transmit information describing:
- Vocal-tract resonances.
- Pitch.
- Excitation type.
- Loudness.
The receiver uses these parameters to reconstruct speech.
Model-based coders can often achieve dramatically lower bit rates than waveform coders, although the reconstructed signal may sound less natural.
What Are Hybrid Coders?
Modern source coders frequently combine waveform and model-based techniques. These are known as hybrid coders. Hybrid systems attempt to combine:
- The efficiency of model-based coding.
- The quality of waveform coding.
Many modern speech coders, including CELP-based systems, use hybrid approaches. Similar concepts are widely used in image and video compression systems.
Hybrid coding has become one of the most important developments in modern source coding because it provides an attractive balance between quality and efficiency.
How Much Compression Is Possible?
The achievable compression depends on the characteristics of the source. Highly predictable information can often be compressed dramatically. The exact values depend on the compression method and the desired quality. Examples include:
| Source | Typical Compression |
|---|---|
| Text documents | 2:1 to 5:1 |
| Speech | 4:1 to 20:1 |
| Images | 10:1 to 50:1 |
| Video | 20:1 to 200:1 |
In general, greater compression can be achieved when some loss of fidelity is acceptable.
Where Is Source Coding Used Today?
Source coding is present in almost every modern communications system. Applications include:
- Telephone networks.
- Cellular systems.
- Satellite communications.
- Internet communications.
- Television broadcasting.
- Video streaming.
- Digital radio.
- Computer networks.
- Cloud storage.
- Mobile devices.
Without source coding, the bandwidth and storage requirements of modern communications systems would be prohibitively large.
Indeed, many services that are now taken for granted—including streaming video and mobile Internet access—would be impractical.
How Does Source Coding Fit into a Communications System?
A simplified digital communications system typically consists of:
Source → Source Coder → Channel Coder → Modulator → Channel → Receiver
The source coder operates near the beginning of the transmission chain. Its purpose is to create an efficient representation of the information source. The resulting bitstream is then passed to the channel coder, which adds redundancy to improve reliability during transmission.
Although these two processes appear contradictory, they are complementary. Source coding removes unnecessary redundancy, while channel coding adds carefully structured redundancy that helps combat transmission errors.
Together, they allow communications systems to achieve both efficiency and reliability.
Why Is Source Coding Important?
Source coding is one of the foundations of modern digital communications.
By reducing the number of bits required to represent information, source coding enables communications systems to use bandwidth and storage resources more efficiently. The resulting savings make possible the high-capacity networks, streaming services, digital media systems, and mobile communications technologies that underpin modern society.
As communications demands continue to grow, source coding remains a critical technology for delivering more information to more users while making efficient use of limited resources.
Summary
Source coding is the process of representing information in a compact and efficient form so that it can be transmitted or stored using the fewest possible bits. By identifying and removing redundancy, source coding reduces bandwidth requirements, increases network capacity, and lowers storage demands. Techniques range from waveform coding methods such as PCM and delta modulation to model-based and hybrid coders used in modern speech, image, and video compression systems. Together, these techniques form one of the fundamental building blocks of modern digital communications.
Back to reading