Data Transmission

EECE 542 Fall 2003

Time/Frequency Relationships
• The relationship between time and
frequency domain representation of signals
is defined by Fourier analysis.
• Unmodulated (non-sinusoidal) signals have
their frequency domain spectra centered
about 0 Hz. (i.e. baseband transmission)
• General rule:
– A faster (shorter period) signal in the time
domain results in a wider (larger bandwidth)
signal in the frequency domain
Ex: A random sequence of 0’s
and 1’s
Baseband Data Transmission
• Most physical layer transmission systems rely on
baseband transmission.
– Almost exclusively use a type of cable or fiber
• Supports only one current transmission
– No parallel transmissions on the same wire unless
multiple wires are used (both tx and rx)
– Exception: some fiber optic systems
• Transmission involves a mapping of binary data to
analog waveforms.
Baseband Data Reception
• Line components typically block the transmission
in the vicinity of 0 Hz (DC).
• The received signal is first filtered and amplified
to reduce the effects of noise and line attenuation.
• Correct decisions on the data being a 0 or 1
requires knowledge of the bit transition edges or
boundaries.
– Requires a bit clock which is not typically sent with the
data
Bit Synchronization
• A bit (data) clock must be generated at the
receiver for the data being received.
• The generation of this clock and the
alignment (phase adjustment) of it’s edges
with the edges of the received data is
performed by a bit synchronizer.
• A bit synch is basically a Phase Lock Loop
(PLL)
• PLL’s work best if bit transitions occur at
most if not all data bit boundaries
Line Coding
• Data embedded in a layer 2 frame may
easily contain long strings of 0’s or 1’s
– Few bit transitions for the PLL to work well
• Line coding is the translation of the binary
data into a new digital stream
– Good line coding schemes guarantee bit
transitions
– The spectral shape of the transmission is often
affected
Types of Line Coding
• Unipolar
• Polar
– NRZ (Nonreturn to Zero)
– RZ (Return to Zero)
– Biphase
• Bipolar
Unipolar Line Coding
• Simple
– Binary 1 = high voltage
– Binary 0 = low (zero) voltage
• Properties
– No edge transitions when the original data
doesn’t change
– No change in the spectral shape (still has DC
component)
Unipolar – cont.
NRZ (Nonreturn to Zero) Coding
• A type of polar (two non-zero voltage levels) coding
• Removes the DC component
• NRZ-I
• NRZ-L
• 0 -> positive (or neg.) voltage
• 1 -> negative (or pos.) voltage
• NRZ-I
• 0 -> voltage remains the same
• 1 -> causes an inversion in the voltage
– creates bit transitions in long strings of 1’s (but not 0’s)
NRZ – Cont.
RZ (Return to Zero) Coding
• Another type of polar encoding
• The first half of each bit is mapped as in
NRZ-L
• The second half of each bit is set to 0 volts
• Guarantees bit transitions
• Removes the DC component
• The width of the transmitted pulse is cut in
half so the spectral bandwidth increases
RZ – Cont.
Biphase Coding
• Another type of polar
• Like RZ, transitions are created in the middle of the bit
periods
• Most common methods used in LANs
• Manchester
– Middle transition =  if bit =1,  if bit = 0
– Ethernet
• Differential Manchester
– Middle transition always present, but a transition at the
beginning of a bit only occurs if the bit = 0
– Token Ring
Manchester & Diff. Manchester
Line Coding Spectra
Block Coding
• Enhances the performance of line coding
while also introducing some error-detecting
capability
• Based on substituting a block of n bits for a
block of m bits, where n > m
• A dictionary contains the mapping. Some
of the n-bit blocks are not used in the one-
to-one mapping
Block Coding – cont.
Block coding subsitution
(m=4, n=5)
Block coding –cont.
• Errors can be detected if the received n-bit
word is invalid
• Also called mBnB coding
• Used in some of the newer Ethernet
standards
– 100Base-TX (2-wire twisted pair)
– 100Base-FX (Fiber)
– 1000Base-T (2-wire Gigabit Ethernet)
RF Transmission
• Not baseband
• Requires modulation
– The placement of data onto a cosinusoidal signal
• Multiple bits may be mapped into one
modulation symbol
• Baud rate = modulation symbol rate
• Traditional schemes:ASK, FSK, PSK, QAM
ASK – Amplitude Shift Keying
• Susceptible to channel degradations
FSK – Frequency Shift Keying
PSK – Phase Shift Keying

– BPSK: bit rate = baud rate, 0 or 180 deg. phase

– QPSK: bit rate = 2 * baud rate, [45, 135, 225, 315] deg.
QAM – Quadrature Amplitude
Modulation
• Combined ASK and PSK
• Higher-order modulation scheme that
lowers the symbol rate
• More susceptible to noise and nonlinearities
• Used in most modern phone modems
8-QAM
Multiplexing
• Transmission resources are usually limited
in either time, frequency, or both
• Normally two separate signals cannot share
the same time and frequency space
• As multiple users or segments become
necessary, a method of sharing the these
resources is critical
• Multiplexing allows this sharing
FDM – Frequency Division
Multiplexing
• The frequency channel is divided and each
user receives one portion of the spectrum
• Requires at least one non-baseband signal
• Guard bands are used to limit the effect of
adjacent channel interference (ACI)
FDM – cont.
FDM – cont.
Time-division Multiplexing
(TDM)
• Dividing by time
• Supports any combination of baseband and
modulated signals
• Two types of TDM:
– Synchronous TDM
– Asynchronous TDM
Synchronous TDM
• Each user (1, 2, … n) is allocated a time slot
• A frame consists of one full cycle of a time
slot from every user
• Requires framing bits for time slot
synchronization
• Inefficient if data is not always being sent
by ALL users
Synch. TDM – Cont.
Asynchronous TDM
• m time slots for n users, m < n
– Time slots are not reserved for each user
• Scans user input lines for available data
• Tries to fill all time slots during each frame
• Requires addressing overhead for correct
de-multiplexing
• Typically more efficient that synch. TDM
Asynch. TDM – cont.
Limits on Data Throughput
• Nyquist Bit Rate
– Noiseless, bandlimited channels
– Bit Rate (bps) = 2 x B x log2(L)
• L = # of signal levels used to represent the data
• B = frequency bandwidth available (Hz)
• Shannon’s Capacity Theorem
– Bandlimited channels with noise
– C (bps) = B x log2(1 + SNR)
• SNR = signal-to-noise ratio of the channel
Nyquist Example
• A noiseless channel with a 5 kHz
bandwidth and binary transmission (2
levels) can deliver:
– Bit Rate = 2 x 5000 x log2(2) = 10,000 bit/sec.
• If transmission using 4 bits/symbol is used
(16 levels) then
– Bit Rate = 2 x 5000 x log2(16) = 40,000 bit/sec.
Shannon Capacity Example
• A modem operating over a telephone line
has a maximum useful bandwidth of about
3400 Hz (300 Hz to 3700 Hz). The
maximum SNR of the channel is 39 dB.
What is the maximum capacity?
– First, un-dB the SNR:
• SNR = 10^(39/10) = 7943
– C = 3400 x log2(1+7943) = 44 kbps