mm Principles

40 60 80
of Communication 100
Systems 120

(Subject Code: EC-3501)

40

Lecture 27
Date: 12/10/2020

60
Dr. Pawan Kumar
Assistant Professor
Department of Electronics and Communication Engineering
National Institute of Technology Rourkela
80
Rourkela, Odisha

navigation symbols
Autumn (2020-21)
1 / 13
Part 4: Analog to digital conversion and digital communications
Reveiwmm
of the last lecture
40 60 80 100 120
• Non-uniform quantizer: compressor, expander, compandor
(compressor+expander)
∗ Variable step size depending on the strength of the input analog signal
∗40Results in a constant signal-to-quantization-noise ratio (SQNR/SNR) all
amplitude levels of the signal
• Pulse-code modulation (PCM): generator and demodulator
∗ Generator is basically performs analog to digital conversion using
60sampling, quantization, and encoding techniques
∗ Regenerative repeater(s), pulse detector, and low-pass filter
• Time division multiplexing (TDM)
∗ Transmission of multiple signals by allocating different time slots in a
80way that all signals can use the same channel bandwidth.
∗ In TDM for PCM, the main building blocks are commutator, encoder and
navigation symbols
pulse modulator, pulse demodulator, and decommutator.
2 / 13
Time division multiplexing
• Themmidea is to allow
40
many signals
60
to share 80
the same medium
100
(channel)
120
in time. (Recall that un FDM (frequency division multiplexing) the
channel is shared by signals in frequency.)
• The signals to be multiplexed may come from various sources such as
PCM, computer output, telemetry data, digital facsimile, etc.
40
• Following is an illustration of multiplexing (samples of) two signals
m1 (t) and m2 (t) in time.

60

80

navigation symbols

3 / 13
• This is basically multiplexing of PAM signals, s1 (t) and s2 (t).
• The sampling time Ts should be sufficiently larger than the pulse time T
mm 40 60 80 100 120
to accommodate the two signals.
• Both signals’ sample pulses are of duration T, hence their amplitude
spectra are same. Although, the phase spectra would differ due to
relative delay in the pulses.
• Same40 amplitude spectra (sinc shape for rectangular pulses) suggests the
two signals multiplexed in time share the same channel. (the following
diagram is only for illustration of the spectra of m1 (t) and m2 (t))
• Available bandwidth BT will be shared by both signals at different time.
(No cross-talk like in FDM (refer to the previous related lectures))
60

80

navigation symbols

4 / 13
• Consider another example of TDM of four signals m1 (t), m2 (t), m3 (t),
m4 (t).
andmm
40 60 80 100 120

40

60

80

navigation symbols

5 / 13
• Transmitter end
mm 40 60 80 100 120

40

60

80

navigation symbols

6 / 13
• Receiver end
mm 40 60 80 100 120

40

60

80

navigation symbols

7 / 13
• A schematic of TDM system
∗ The main components are – commutator, pulse modulator, pulse
mm demodulator,40and decommutator
60 80 100 120
∗ Commutator is basically a switching circuitry which selects samples of N
incoming signals alternatively. This results in multiplexing of the signals
in time domain
∗ Pulse modulator is a circuitry designed to convert the input (analog)
40samples in digital form and then do proper arrangement for transmission
∗ On reception, pulse demodulator performs operations to convert the input
signal into binary form. Then bits are regrouped to make samples
∗ The regrouped samples are separated by decommutator to extract
samples corresponding to individual signals
∗60The samples are then passed through sufficiently designed low-pass
filters to extract the N analog signals

80

navigation symbols

8 / 13
More on TDM
• Application: One of the applications of TDM is found in T1 system. T1
mm carries multiplexed
system 40 2460PAM voice signals.
80 100
Regenerative 120
repeaters are used in between the transmitter and the receiver to nullify
the effects of channel noise. (refer to the textbooks for more details)
• Digital Multiplexing: Instead of multiplexing the samples, we can also
multiplex the bits or group of bits (word). This results in bit
40
interleaving and word interleaving, respectively.
∗ In bit interleaving, bits are taken one at time from various signals
(samples)
∗ A group of bits are of the samples are multiplexed in word interleaving.
T1 system uses this interleaving.
60
• Synchronization: Synchronization between the transmitter and receiver
is required to correctly identify the bits at the receiver. It is done by
adding framing and synchronization bits to the data (information) bits.
These extra bits other than data bits are known as overload bits.
80 coming from various sources/channels (PCM, computer output,
• Signals
telemetry data, digital facsimile) may have different data rate (number
of bits per second). In such a case, high-bit-rate channels are allocatednavigation symbols

proportionately more slots. 9 / 13

Alternatives to PCM
mm 40 60 80 100 120
• PCM is not a very efficient system as it generates so many bits and
requires large bandwidth for transmission
• In PCM, value of a sample can be guessed from the previous samples –
40 are not independent. Also, it generally happens that there
samples
exists a large amount of redundancy in samples

• Differential PCM (DPCM), adaptive DPCM (ADPCM), delta

modulation
60 (DM), adaptive DM (ADM) are the alternatives to improve
the encoding efficiency of the analog-to-digital convertor (PCM)

• These schemes exploit the non-independency of the samples to avoid

the 80
redundancy and, hence, get an efficient encoding – use of fewer bits
• Prediction techniques are employed for this purpose
navigation symbols

10 / 13
The underlying concept of prediction
• Consider a simple case of transmitting the difference of samples instead
mm 40 60 80 100 120
of the samples
• If m[k] = m[kTs ] is the kth sample, instead of transmitting the samples
directly, we transmit the difference d[k] = m[k] − m[k − 1]
• At receiver, knowing d[k] and the previous sample value m[k − 1], we
40
can reconstruct m[k] iteratively using relation m[k] = d[k] + m[k − 1]
• An improvement: Use of relation d[k] = m[k] − m̂[k], instead of
d[k] = m[k] − m[k − 1]. m̂[k] is an estimate (prediction) of m[k]
obtained using the previous samples
∗60m̂[k] = Estimation[m[k − 1], m[k − 2], m[k − 3], . . . ]
∗ Transmitted sample is the difference sample d[k]
∗ At receiver, the previous received samples are used to estimate the
present sample m[k], given by m̂[k]. Hence, we can get the output sample
as m[k] = d[k] + m̂[k]
80
• Due to operation on the difference of the successive samples, the
method employing this idea for encoding of samples is commonly
navigation symbols
known as differential PCM.
11 / 13
• Future prediction: The idea of estimating m[k] form the past samples
m[kmm
− 1], m[k − 40
2], m[k − 3],.60
. . , is a kind 80 100
of future prediction 120
• This can be understood better with the help of Taylor series expansion
∗ Using Taylor series, a signal m(t) can be expanded at t = Ts as
Ts2 T 3 ...
40 m(t + Ts ) = m(t) + Ts ṁ(t) + m̈(t) + s m(t) + . . .
|2! 3!
{z }
→0 for small Ts
≈ m(t) + Ts ṁ(t)

∗60For sampling at t = kTs

m[k] − m[k − 1]
m[kTs + Ts ] = m[k + 1] ≈ m[k] + Ts = 2m[k] − m[k − 1]
Ts

∗80This is second order prediction of m[k + 1]. A prediction of m[k] can be

given by
m[k] ≈ 2m[k − 1] − m[k − 2] navigation symbols
| {z }
m̂[k] 12 / 13
 ∗ Nth order prediction of m[k] can be given by
mm 40 60 80 100 120
m̂[k] = a1 m[k − 1] + a2 m[k − 2] + · · · + aN m[k − N]

• For large N, m̂[k] would be closer to m[k]. This suggests, we will get
40
smaller difference d[k] for large N.
• Smaller d[k], which is output to the quantizer, results in smaller step
size if the number of levels (in PCM and DPCM) remains same.
60
• The smaller step size will induce reduction in the quantization error
• This means, for given quantization levels (same number bits or same
amount of consumed bandwidth), the SNR is increased in DPCM.
80
• We can also say that for same amount of SNR we need smaller
bandwidth in DPCM navigation symbols

13 / 13