Menoufia University – Faculty of Electronic Engineering

ECE 281: Digital Communications Module – Spring 2024

PCM

1) Uniform PCM
• For a message signal with amplitude (−𝑚𝑝 , 𝑚𝑝 ), the amplitude range is divided into 𝐿 uniformly
2𝑚𝑝
spaced intervals, each of width → ∆𝑣 = 𝐿
∆ ∆
• The quantization error lies in the range (− 2 , 2)
∆
• The Maximum quantization error → 𝑒𝑚𝑎𝑥 = 2

• The Signal power:

𝑚𝑝 2
𝑆𝑜 =
2
• The noise power:
1 ∆/2
𝑁𝑜 = ∆ ∫−∆/2 𝑞 2 𝑑𝑞 , where 𝑞 is the error signal

1 𝑞 3 ∆/2 1 ∆3 ∆2
𝑁𝑜 = |−∆/2 = =
∆ 3 3∆ 4 12

2𝑚𝑝 ∆2 𝑚𝑝 2
∵∆= ∴ 𝑁𝑜 = =
𝐿 12 3𝐿2

The Signal to Quantization Noise Ratio

𝑆𝑜 𝑚𝑝 2 3𝐿2 3
𝑆𝑄𝑁𝑅 = = . 2
= 𝐿2
𝑁𝑜 2 𝑚𝑝 2

Bandwidth requirements:

• A signal with bandwidth 𝑩 Hz and sampled at 𝒇𝒔 samples/sec. therefore, sample 𝒏 bits are used,
therefore a total of 𝒏𝒇𝒔 bits/sec are required for transmission of the PCM signal.

• The minimum bandwidth requirement for transmission of 𝑪 bits/sec (or, more precisely, 𝐶
pulses/sec) is 𝑩𝑾 = 𝑪/𝟐

1|Pa g e Sheet3
Menoufia University – Faculty of Electronic Engineering
ECE 281: Digital Communications Module – Spring 2024

2) Non-Uniform PCM

Figure 1 Non-uniform Quantization

• For speech waveform, small signal amplitudes occur more frequently than large ones.
• The problem can be solved by using smaller steps for smaller amplitudes (nonuniform quantizing), as shown
in Fig. 1.
• The same result is obtained by first compressing signal samples and then using a uniform quantization.
• In the reconstruction path, the inverse logarithmic relation is used to expand the signal amplitude.
• The combined compressor-expander pair is termed a compandor.

The Signal to Quantization Noise Ratio

When a µ-law compandor is used, the output SNR in linear scale is.

𝑆𝑜 3𝐿2 𝑚𝑝 2
𝑆𝑄𝑁𝑅 = ≈ 𝜇2 ≫
𝑁𝑜 [ln (1 + 𝜇)]2 𝑆𝑜
In dB scale
𝑆𝑜 𝑆𝑜
( ) = 10 𝑙𝑜𝑔10 ( )
𝑁𝑜 𝑑𝐵 𝑁𝑜

= 10 𝑙𝑜𝑔10 (𝑐 𝐿2 ) = 10 𝑙𝑜𝑔10 (𝑐 22𝑛 )

= 10 𝑙𝑜𝑔10 (𝑐) + 2𝑛 . 10 𝑙𝑜𝑔10 (2) = (𝛼 + 6𝑛)

𝑆𝑜
( ) = (𝛼 + 6𝑛)
𝑁𝑜 𝑑𝐵
3
Where 𝛼 = 10 𝑙𝑜𝑔 [ln
Sheet 3 (1+𝜇)]2

2|Pa g e Sheet3
Menoufia University – Faculty of Electronic Engineering
ECE 281: Digital Communications Module – Spring 2024

Sheet 3
1) A message signal 𝑚(𝑡) is transmitted by binary PCM without compression. If the SNR (signal-to-
quantization-noise ratio) is required to be at least 47 𝑑𝐵, determine the minimum value of 𝐿
required, assuming that 𝑚(𝑡) is sinusoidal. Determine the SNR obtained with this minimum 𝐿.

𝑆𝑁𝑅|𝑑𝐵 = 10 𝑙𝑜𝑔10 𝑆𝑁𝑅 = 47 𝑑𝐵

𝑆𝑁𝑅 = 1047/10 = 50119

3 2
𝑆𝑁𝑅 = 𝐿 ≥ 50119
2
𝐿 ≥ 182.8
Since 𝐿 should be an integer number power of 2
𝐿 = 28 = 256
Now, the actual SNR of this minimum 𝐿
3 3
𝑆𝑁𝑅 = 𝐿2 = 2562 = 98304 = 49.93 𝑑𝐵
2 2

2) Nine telemetry signals, each of bandwidth 0.36 𝑘𝐻𝑧, are to be transmitted simultaneously by
binary PCM. The maximum tolerable error in sample amplitudes is 0.32% of the peak signal
amplitude. The signals must be sampled at least 40% above the Nyquist rate. Framing and
synchronizing requires an additional 0.45% extra bits. Determine the minimum possible data rate
(bits per second) that must be transmitted, and the minimum bandwidth required to transmit this
signal.

The Nyquist sampling rate per each telemetry signal 𝑓𝑛 = 2𝑓𝑜 = 2 × 0.36 = 0.72 𝑘𝐻𝑧

The actual sampling rate per each telemetry signal 𝑓𝑠 = (1 + 40%) × 0.72 = 1.008 𝑘𝐻𝑧
∆𝑣 𝑚𝑝 0.32
𝑒𝑚𝑎𝑥 = = ≤ 𝑚
2 𝐿 100 𝑝
𝐿 ≥ 312.5
Since 𝐿 should be an integer number power of 2

𝐿 = 29 = 512
The minimum possible data rate per each telemetry signal is

𝐶𝑠𝑖𝑔𝑛𝑎𝑙 = 𝑛 𝑓𝑠 = 9 × 1.008 = 9.072 𝑘𝑏𝑝𝑠

The minimum possible data rate for all signals is

𝐶 = 9 × 𝐶𝑠𝑖𝑔𝑛𝑎𝑙 × (1 + 0.45%) = 82.015416 𝑘𝑏𝑝𝑠

The minimum BW required to transmit this signal is 𝐵𝑊 = 𝐶/2 = 41.007708 𝑘𝐻𝑧.

3|Pa g e Sheet3
Menoufia University – Faculty of Electronic Engineering
ECE 281: Digital Communications Module – Spring 2024

3) A signal 𝑚(𝑡) band-limited to 3.4 𝑘𝐻𝑧 is sampled at a rate 25% higher than the Nyquist rate. The
maximum acceptable error in the sample amplitude (the maximum quantization error) is 0.25%
of the peak amplitude 𝑚𝑝 . Find the minimum bandwidth of a channel required to transmit the
encoded binary signal. If 30 such signals are time-division-multiplexed, determine the minimum
transmission bandwidth required to transmit the multiplexed signal.

The Nyquist sampling rate per signal 𝑓𝑛 = 2𝑓𝑜 = 2 × 3.4 = 6.8 𝑘𝐻𝑧
The actual sampling rate per signal 𝑓𝑠 = (1 + 25%) × 6.8 = 8.5 𝑘𝐻𝑧
∆𝑣 𝑚𝑝 0.25
𝑒𝑚𝑎𝑥 = = ≤ 𝑚
2 𝐿 100 𝑝
𝐿 ≥ 400
Since 𝐿 should be an integer number power of 2
𝐿 = 29 = 512
The minimum possible data rate per signal is
𝐶𝑠𝑖𝑔𝑛𝑎𝑙 = 𝑛 𝑓𝑠 = 9 × 8.5 = 76.5 𝑘𝑏𝑝𝑠 The minimum Bandwidth → 𝐵𝑊𝑠𝑖𝑔𝑛𝑎𝑙 = 38.25 𝐾𝐻𝑧

The minimum possible data rate for all signals is

𝐶 = 30 × 𝐶𝑠𝑖𝑔𝑛𝑎𝑙 = 30 × 76.5 = 2.295 𝑀𝑏𝑝𝑠
The minimum BW required to transmit the multiplexed signal is 𝐵𝑊 = 𝐶/2 = 1.1475 𝑀𝐻𝑧

4) A signal 𝑚(𝑡) of bandwidth 𝐵 = 3.4 𝑘𝐻𝑧 is transmitted using a binary companded PCM with µ =
200. Compare the case of 𝐿 = 128 with that of 𝐿 = 256 from the point of view of transmission
BW and the output SNR.

For 𝐿 = 128 → 𝑛 = 𝑙𝑜𝑔2 128 = 7

𝐶 𝑛𝑓𝑠
𝐵𝑊 = = = 𝑛𝐵 = 7 × 3.4 = 23.8 KHz
2 2
𝑆𝑜
𝑆𝑄𝑁𝑅 = ( ) = (𝛼 + 6𝑛) = 𝛼 + 42 = −9.72 + 42 = 32.28 𝑑𝐵
𝑁𝑜 𝑑𝐵
3 3
Where 𝛼 = 10 𝑙𝑜𝑔 [ln (1+𝜇)]2 = 10 𝑙𝑜𝑔 [ln (201)]2 = −9.72 𝑑𝐵

For 𝐿 = 256 → 𝑛 = 𝑙𝑜𝑔2 256 = 8

𝐶 𝑛𝑓𝑠
𝐵𝑊 = = = 𝑛𝐵 = 8 × 3.4 = 27.2 KHz
2 2
𝑆𝑜
𝑆𝑄𝑁𝑅 = ( ) = (𝛼 + 6𝑛) = 𝛼 + 48 = −9.72 + 48 = 38.28 𝑑𝐵
𝑁𝑜 𝑑𝐵

The difference between the two SNRs is 6 dB, which is a ratio of 4. Thus, the SNR for L = 256 is 4
times the SNR for L = 128.
27.2−23.8
The 2nd case requires just about 23.8
= 14.3% more bandwidth compared to the 1st one.

4|Pa g e Sheet3
Menoufia University – Faculty of Electronic Engineering
ECE 281: Digital Communications Module – Spring 2024

5) A signal 𝑚(𝑡) of bandwidth 𝐵 = 4 𝑘𝐻𝑧 is sampled at a Nyquist rate and transmitted using a
binary companded PCM with µ = 255 and 𝐿 = 512, calculate the transmission rate and the
output SNR. If we would like to increase the sampling rate 25% higher than the Nyquist rate and
keep the transmission rate unchanged, calculate the new SNR required to achieve this.

For 𝐿 = 512 → 𝑛 = 𝑙𝑜𝑔2 512 = 9

𝐶 = 𝑛𝑓𝑠 = 2𝑛𝐵 = 2 × 9 × 4 = 72 𝐾𝑏𝑝𝑠

𝑆𝑜
𝑆𝑄𝑁𝑅 = ( ) = (𝛼 + 6𝑛) = −10.11 + 6 × 9 = 43.89 𝑑𝐵
𝑁𝑜 𝑑𝐵
3 3
Where 𝛼 = 10 𝑙𝑜𝑔 [ln (1+𝜇)]2 = 10 𝑙𝑜𝑔 [ln (256)]2 = −10.11 𝑑𝐵

If the sampling rate increased 25% higher than the Nyquist rate → 𝑓𝑠 = (1.25)𝑓𝑛 = 2.5 𝐵

𝐶 = 𝑛𝑓𝑠 = 𝑛 2.5 𝐵 = 𝑛 × 2.5 × 4 = 72 𝐾𝑏𝑝𝑠

𝑛 = 7.2 → 8
𝑆𝑜
𝑆𝑄𝑁𝑅 = ( ) = (𝛼 + 6𝑛) = −10.11 + 6 × 8 = 37.89 𝑑𝐵
𝑁𝑜 𝑑𝐵

6) A signal band-limited to 640 𝑘𝐻𝑧 is sampled at a rate 40% higher than the Nyquist rate and
quantized into 128 levels using µ − 𝑙𝑎𝑤 quantizer with µ = 100.
a) Determine the signal-to-quantization-noise ratio.
b) The SNR (the received signal quality) found in part (a) was unsatisfactory. It must be
increased at least by 5 𝑑𝐵. Would you be able to obtain the desired SNR without increasing
the transmission BW if it was found that a sampling rate 22.5% above the Nyquist rate is
adequate? If so, explain how.

a) 𝑛 = 𝑙𝑜𝑔2 128 = 7
𝑆𝑜 3
( ) = (𝛼 + 6𝑛) = 10 𝑙𝑜𝑔 + 6𝑛
𝑁𝑜 𝑑𝐵 [ln(1 + 𝜇)]2
3
= 10 𝑙𝑜𝑔 + 6 × 7 = 33.48 𝑑𝐵
[ln (101)]2
b) The Nyquist sampling rate 𝑓𝑛 = 2𝑓𝑜 = 2 × 640 = 1.28 𝑀𝐻𝑧
The actual sampling rate 𝑓𝑎 = (1 + 40%) × 1.28 = 1.792 𝑀𝐻𝑧
In part (a), the information rate 𝐶𝑎 = 𝑛. 𝑓𝑎 = 7 × 1.792 = 12.544 𝑀𝑏𝑝𝑠
The new sampling rate 𝑓𝑏 = (1 + 22.5%) × 1.28 = 1.568 𝑀𝐻𝑧
𝐶𝑏 = 𝑛𝑓𝑏 = 𝐶𝑎
𝐶𝑎 12.544
𝑛= = = 8 𝑏𝑖𝑡𝑠
𝑓𝑏 1.568
𝑆𝑜 3
( ) = 10 𝑙𝑜𝑔 + 6 × 8 = 39.48 𝑑𝐵
𝑁𝑜 𝑑𝐵 [ln (101)]2

5|Pa g e Sheet3
Menoufia University – Faculty of Electronic Engineering
ECE 281: Digital Communications Module – Spring 2024

Clearly, the SNR in this case is increased by 6 𝑑𝐵 (6 𝑑𝐵 for each 1 bit increase), i.e more
than the required 5 𝑑𝐵.
7) A signal band-limited to 1 𝑀𝐻𝑧 is sampled at a rate 50% higher than the Nyquist rate and
quantized into 256 levels using a µ − 𝑙𝑎𝑤 quantizer with µ = 255.
a) Determine the signal-to-quantization-noise ratio.
b) The SNR (the received signal quality) found in part (a) was unsatisfactory. It must be
increased at least by 10 𝑑𝐵. Would you be able to obtain the desired SNR without increasing
the transmission bandwidth if it was found that a sampling rate 20% above the Nyquist rate
is adequate? If so, explain how. What is the maximum SNR that can be realized in this way?

a) 𝑛 = 𝑙𝑜𝑔2 256 = 8
𝑆𝑜 3
( ) = (𝛼 + 6𝑛) = 10 𝑙𝑜𝑔 + 6𝑛
𝑁𝑜 𝑑𝐵 [ln(1 + 𝜇)]2
3
= 10 𝑙𝑜𝑔 + 6 × 8 = 37.89 𝑑𝐵
[ln (256)]2
b) The Nyquist sampling rate 𝑓𝑛 = 2𝑓𝑜 = 2 × 1 = 2 𝑀𝐻𝑧
The actual sampling rate 𝑓𝑎 = (1 + 50%) × 2 = 3 𝑀𝐻𝑧
In part (a), the information rate 𝐶𝑎 = 𝑛. 𝑓𝑎 = 8 × 3 = 24 𝑀𝑏𝑝𝑠
The new sampling rate 𝑓𝑏 = (1 + 20%) × 2 = 2.4 𝑀𝐻𝑧
𝐶𝑏 = 𝑛𝑓𝑏 = 𝐶𝑎
𝐶𝑎 24
𝑛= = = 10 𝑏𝑖𝑡𝑠
𝑓𝑏 2.4
𝑆𝑜 3
( ) = 10 𝑙𝑜𝑔 + 6 × 10 = 49.89 𝑑𝐵
𝑁𝑜 𝑑𝐵 [ln (256)]2

Clearly, the SNR is increased by 12 𝑑𝐵 (6 𝑑𝐵 for each 1 bit increase), i.e more that 10 𝑑𝐵.

6|Pa g e Sheet3