## Continuous Variable Slope Delta Modulation

Continuous Variable Slope Delta Modulation was first proposed by Greefkes and Riemes in 1970. CVSD requires a 1-bit sample length compared to the 8 bits used in PCM, so more samples can be sent in the same bandwidth. As a result, CVSD is more tolerant of communications errors. Because of its error tolerance,

CVSD performs well in noisy channels, and for this reason, it has been widely used in military communications systems.The ability to tolerate errors is also what makes CVSD attractive for use in Bluetooth systems.

CVSD quantizes the difference in amplitude between two audio samples (that is, between the current input sample and the previous sample).The challenge is always to choose the appropriate step size 8(k). Small step sizes are better for tracking slowly changing low amplitude signals, but a larger step size is needed to accurately track a fast-changing high amplitude signal. This effect is shown in Figure 9.3.

Figure 9.3 The CVSD Operational Concept

Figure 9.3 The CVSD Operational Concept

Let's consider a random input voice signal that we would like to convert from analog samples to digital format using CVSD. Figure 9.4 shows how this happens. As the input signal increases, bits set to 1 are transmitted. If the input signal decreases, bits set to zero are transmitted. In the first declining cosine slope of the signal, we can see how poorly the signal was quantized, but since it is an adaptive differential quantizer, it starts to adapt by changing the step size. Given this, if the signal characteristics remain the same, it will excel in following almost exactly the trace of the input signal.

Figure 9.4 The CVSD Waveform

Figure 9.4 The CVSD Waveform

i i M (Hull i I 1 1 1 d i [■ u '1 il i n i li C u i [i - (id

In the CVSD algorithm, the adaptive changes in step size, 8(t), are based on the past three or four sample outputs (for example, b(k), b(k-1), b(k-2), b(k-3)) where it increases or decreases to catch up with the input signal as was shown in the example of Figure 9.4 earlier.The step size, 8(t), is controlled by the syllabic companding parameter, a, which determines when to increase 8(t) or allow it to decay.The step size decay time, P, is related to speech syllable length (sometimes called delay).The Bluetooth system specifies P to be 16 ms and the accumulator decay factor, h, to be 0.5 ms.

The accumulator decay factor decides the threshold of how quickly the output of the CVSD decoders decay to zero after an input; this determines how quickly the Codec will recover from errors in the received signal. Figure 9.5 shows flow diagrams of the algorithms for the encoder and decoder. The internal state of the accumulator depends upon the equations that follow.

hcku^jiij.iMlhil'l-ll.t;

Figure 9.5 The CVSD Encoder and Decoder Block Diagram

Figure 9.5 The CVSD Encoder and Decoder Block Diagram

A standard called Mean Opinion Scale (MOS) testing is used to assess the subjective quality of voice links. A rating of 4 to 4.5 is considered toll quality (equivalent to commercial telephony). As MOS decreases, so quality decreases; a value of just less than 4 indicates communication quality with some barely perceptible distortion. Figure 9.6 compares MOS ratings for ^,-law PCM and CVSD with various bit error rates on the channel.

Note_

The term toll quality was first used about 22 years ago when T1 multiplexers first started transporting voice over private T1 lines. The original idea was that a private wide area network (WAN) could provide voice quality equal to that of the long-distance public switched telephone network (PSTN), which charged a toll for each minute of use using what is nowadays known as Voice Over IP (VoIP).

Notice CVSD performs as well as ^,-law PCM in a clean communication medium. However, CVSD operates much better than ^,-law PCM in the presence of bit errors.To be more specific, CVSD retains quite good MOS ratings at low bit error rates; however, it drops to a MOS rating of 3 (fair quality but tends to be annoying) at higher bit error rates. This robustness to bit errors (channel noise) makes CVSD an ideal solution for many wireless speech communication applications, including Bluetooth technology. But because PCM is cheap and already available in a lot of devices, we really need both.

Figure 9.6 MOS versus Bit Error Rate for CVSD and |-Law PCM Codec

+2 -2

### Responses

• sophia
Why adaptive delta modulation is used?
8 years ago
• emilia
How adaptive delta modulation in bluetooth technology?
7 years ago
• margaret
How to calculate step size in cvsdm?
7 years ago
• VANNA
Why ADM system sometimes known as Continuous Variable Slope Delta modulation?
3 years ago
• Pandora Gawkroger
Why adm system sometimes known as cvsd modulatior?
2 years ago