Data and Signals

Data must be transformed to electronic signals

(why?)

Analog data: Continuous information, e.g., voice has a value at any

time. Analog clock have values at any time!

Digital data: Discrete state information. e.g., digital clock.

Analog signal: They have infinitely many levels over a period of time.

Digital signals: They have limited number of defined values.

Periodic Signals: “Repeat a pattern every measurable time frame,

“period” or “cycle”. (most used for analog signals)

Non-Periodic (Aperiodic) Signals: No pattern or cycle. (most used for

digital signals)

Figure 3.1 Comparison of analog and digital signals

3.6
 Periodic Analog Signals

Figure 3.2 A sine wave

3.9

Frequency (f) = 1/T and T = 1/f

Phase: It is the position of waveform relative to time = 0.

Ex: The “sine” phase is 0, where as the “cos” phase = π/2

Figure 3.3 Two signals with the same phase and frequency,
but different amplitudes

3.12
 Figure 3.4 Two signals with the same amplitude and phase,
but different frequencies

3.15

Figure 3.5 Three sine waves with the same amplitude and frequency,
but different phases

3.23

Wave length: It relates the frequency / period of a signal to its propagation

speed in the medium. In case of “light” over fiber:
λ=c/f
c = propagation speed of light signal.
f = frequency of light signal.
λ = Wave length of light signal.
Time vs. Frequency domain:
The amplitude values are plotted versus time in case of time domain
signals.

Figure 3.7 The time-domain and frequency-domain plots of a sine wave

3.26
The amplitude is shown for each frequency component of a signal in
the frequency domain. The above sin(6t) has one frequency of 6 Hz,
with 5 volt peak amplitude.

Figure 3.8 The time domain and frequency domain of three sine waves

3.29
Composite Signals: Every composite signal is made of many sine waves of
different amp’s, freq’s, phases. (Fourier analysis)

A periodic composite signal can be decomposed into a number of

signals with discrete frequencies in the frequency domain.

Whereas, a non-periodic composite signal is a group of sine waves with

continuous frequencies in the frequency domain.

Figure 3.9 A composite periodic signal

3.34

Figure 3.10 Decomposition of a composite periodic signal in the time and

frequency domains

3.35
 Whereas, a non-periodic composite signal is a group of sine waves
with continuous frequencies in the frequency domain.

Figure 3.11 The time and frequency domains of a nonperiodic signal

3.37
Band width: (non/composite signals). Refer fig 3.12

BWcs = fh – fl Hz

Figure 3.12 The bandwidth of periodic and nonperiodic composite signals

3.39
 Transmission Impairment
Causes:

1) Attenuation: Loss of energy as a function of the signal traveling

distance and its power.
Decibel: Unit measure of signal’s power loss or gain

Power units dB = 10 log10 (Pdest/Psrc)

Pdest: Signal Power at destination Psrc: Signal power at source

2) Distortion: Signal changes in form / shape.

Composite signal components have different propagation speed
through the medium; hence destination arrivals delays/phase shifts
causing incorrect receiving of the original signal at the destination.

3) Noise:
i) Thermal (white) caused by electrons random movement in the
medium (extra superimposed signals)
ii) Induced: Caused by appliances acting as a sending antenna with
medium as receiving one
iii) Crosstalk: Mutual affect between wires acting as sending/
receiving antenna
iv) Impulse: caused by a power lines, lightning shaped as a “spike”
(high frequency on very short time)

Signal to Noise Ratio (SNR):

SNR = avg. signal power / avg. noise power

SNR plays a profound role in deciding the bit rate limit (max).

It is always desirable to have it high (lower noise than signal power)

for good reception of signals passing through noisy environment (low
quality channel).
It is measured (and always given) in decibel units (powers ratio) dB
To convert to watt: (SNRdB/10)
SNRwatt =10
A SNRdB = 40 is to be converted to watt --> 10(40/10) = 10000 watt
 Digital Signals

Figure 3.16 Two digital signals: one with two signal levels and the other
with four signal levels

3.50

• Bit Rate: “non-periodic” digital signals are the most used in

digital data transfer, hence period or frequency are not used,
instead we use “Bit rate” which is the number of bits Tx per sec.

• Bit length = propagation speed * bit duration (distance one bit

occupies on the Tx medium)
• Bit interval: Time to Tx a bit = 1/bit rate = s/b
• Baud rate: Number of distinct changes (in signal) over the
medium per sec

Bit rate = Baud rate X number of bits per a signal change

In binary signals encoding, bit rate = baud rate (why?)

 Figure 3.17 The time and frequency domains of periodic and nonperiodic
digital signals

3.57

Transmission of Digital Signals:

1) Baseband (digital/encoding): We use low-pass channel with its BW
starting from 0 (i.e., flow = 0), since we are Tx digital signal.

Figure 3.18 Baseband transmission

2) 3.58

Since a digital signal is a composite of (infinite) analog signal with

infinite BW (frequency components), the idle case is to have low-
pass channel with infinite bandwidth (not in real life!) that must start
from frequency component zero and have a very wide BW channel, in
the frequency spectrum.(why?)

Remember: Due to the filtering effect of the channel, The loss of low
and high frequency components of the transmitted digital signal will
result in the loss of the flat part and the sharp edges (respectively) of
the received digital signal.
 Figure 3.19 Bandwidths of two low-pass channels

3.60

Yet, we can still send only the signal’s frequency components with
significant amplitudes within some acceptable threshold for minimum
distortion and receive a reasonably acceptable digital signal at the
receiver. As the bit rate increases, the signal significant BW (BWss)
increases, hence we need a wider medium low-pass wider BW (BWm)
For safe digital signal reception: BWm >= BWss

Figure 3.20 Baseband transmission using a dedicated medium

3.61

In fig 3.20, the input signal is Non-Periodic digital (user transmitted burst
data hence it has continuous BW (0 to ∞). The channel BW filters out the
low and high frequency component of the input signal, hence the receives
gets a distorted output signal. In case of the channel passing enough signal
BWss then we are OK, otherwise bad signal reception!
 Figure 3.21 Rough approximation of a digital signal using the first harmonic
for worst case

3.64

Figure 3.22 Simulating a digital signal with first three harmonics

3.65
Fig. 3.22: As you capture more (low and high frequencies)
harmonics you get better digital signal representation.
2) Broadband (analog/modulation): Digital signal  analog signal
If digital signals are so so so desirable for many reasons (easier
manipulation of refreshing for long channels, compressing for BW
saving, securing via encryption, ...) WHY do we ever need analog?
If the available channel BW is scarce, hence we are forced to to use a
bandpass channel (Fig. 3.23) with its flow = f1 is non-zero value.
Consequently, We do not have enough BW for digital transmission! Then
we need to convert our digital signal to analog signal (modulation
process), as shown in fig. 3.24
Figure 3.23 Bandwidth of a bandpass channel

3.70

Figure 3.24 Modulation of a digital signal for transmission on a bandpass

channel

3.72
Since we are modulating our digital data into analog signals (sin wave
carriers with different amplitude/frequencies/phases) we need not to
start from the zero in the spectrum, but anywhere we select our carrier
frequency (no need to have infinite number of frequency
components), hence band-pass channel instead of low-pass as in the
case of baseband digital Tx.
Signal to Noise Ratio (given in dB always):
SNRdB = 10log10 SNR where
to convert to watts:
SNRwatt = Signal power watt / Noise power watt
(SNR /10)
= 10 dB

Noise Free channel: Nyquist Capacity (max channel bit rate)

Max Bit Rate ––– CNyq = 2 BWch * log2 L -----------(1)
Where L is the number of signal levels

Since there is no upper limit on the value of L, we assume that the

channel is noise free for our analysis, otherwise large L will make
no sense over noisy channel! Yet, if we choose a value for L,
regardless of the channel noise, then in practice, we might be right or
wrong in such choice of the value of L. Hence, we need another
measure to judge our choice as depicted in the next theorem for
Shannon:

Noisy Channel: Shannon Capacity

Max Bit Rate ––– CShan = BWch log2 (1 + SNRwatt) ---(2)

Both limits are important CNyq to decide the encoding bit/level and CShan for the
actual noisy environment limitation. In case of noisy channel, obtaining CShan from
(2) and plugging it in lhs of (1) will guide us in the practical and correct choice of
encoding (bit/level). Otherwise, if we made a wrong choice of encoding (not
following the above) , then we might end up with an upper bound CNyq that is much
lower than a valid CShan.

Example: What is the max bit-rate capacity of a 4000 Hz binary channel (2

levels encoding) has a SNR of 30 dB?
CNyq = 2*4000 * log2 2 = 8000 b/s
CShan = 4000 * log2 (1+ 10(30/10)) = 4000 * 9.9672 ≈ 39869 b/s
Even though, the CShan is much higher than the CNyq , yet we are forced to the
lowest value of the two obtained capacities (i.e., 8000 b/s), as a result of bad
encoding decision of only two levels!
 Performance of Networks

1) Throughput: The actual speed of data traveling a link (notice,

the BW of a link is its potential speed)
Throughput (T) <= BW
Example: bad encoding might waste BW in case of
Ethernet ME. Therefore TME < BW

2) Delay (Latency): L = propagation + Tx + queuing + processing

Propagation delay = Distance (m) / propagation speed (m/s)

Tx delay = message size (bit) / data-rate (bit/sec)

3) Jitter: Packets of data will encounter different delays.

(not very good for real time multimedia applications!)