EE 5103
Advanced Digital Communication
Lecture 9
Recap of Last Lectures…
• Methods to calculate the channel capacity
• Significance of Channel Capacity
• Yardstick to measure the performance of comm. Systems
• Behavior of Shannon’s Capacity formula i.e.
• How the Shannon capacity can be achieved in different ways with different
limitations on different systems
• We found the different tradeoffs
• Shannon’s capacity can be achieved in different ways in different
environments.
Recap of Last Lectures…
• We identified 2 distinct regions based on Shannon’s capacity formula.
• Bandlimited Region: we can increase the spectral efficiency but at the expense of power
efficiency.
• The value of r increases in the bandlimited region, but the required EBNO also increases.
• Power Limited Region: we can reduce the spectral efficiency (r) and we can also reduce the
value of Ebno, which means that we can increase the power efficiency.
In the extreme case if we allow the spectral efficiency to approach zero by
consuming a lot of BW (infinite amount of BW) then we can transmit with the value
of EBNO which we call the ultimate channel limit and the value is around -1.95dB.
• The tradeoffs between power and bandwidth can be achieved with the help of
appropriate modulation and coding schemes.
• Higher order modulation schemes that can pack more messages (spheres) in a given volume
are helpful in bandlimited region.
• Orthogonal modulation schemes (e.g. orthogonal FSK) and simple binary modulation
schemes with appropriate channel codes can be used in the power-limited region.
Spectral and Power Efficiency of M-PAM
Room for Improvement
Is Error Correction Coding Worth it in
Bandlimited Region
Example
• To transmit 2 bits simultaneously, we can use 4-PAM.
• To achieve probability of error of 10!" it requires 13.6 dB.
• To transmit 3 bits simultaneously, we can use 8-PAM.
• To achieve the same probability of error of 10!" it requires 18.2 dB.
• Suppose we are transmitting 2 bits simultaneously using 4-PAM and now
we want to improve the gap of 4-PAM from the Shannon limit.
• Suppose, we use an error correction code that introduces 1 additional bit
for every two data bits.
• Now, to accommodate the additional 3rd bit without increasing
bandwidth requirements we decide to expand the constellation and use
8-PAM
Example
• However, 8-PAM requires 18.2 − 13.6 = 4.6 𝑑𝐵 improvement just to
achieve the performance of 4-PAM.
• Thus, the code has to provide a gain of 4.6 𝑑𝐵 to reach 4-PAM
performance and then additional gain to beat it.
• This becomes too difficult and complex.
Therefore, the motivation of using classical error correction techniques is
lost in the bandlimited region.
Two Dimensional Schemes – QAM and PSK
PSK and QAM Constellation Diagrams
QAM Constellation Example
PSK Constellation Example
QAM Constellation Shapes
Comparison of PAM, PSK, QAM
Probability of Error – PSK
PSK
Probability of Error – QAM
QAM – Square Constellations
Performance Comparison of Modulation Scheme –
Bandlimited Region
Higher Dimensional Lattices
Dense Packing of Points
Coding Gain
&
𝑑#$%
𝐶𝑜𝑑𝑖𝑛𝑔 𝐺𝑎𝑖𝑛 = &
!
𝑉 '!
Where 𝑉 "# is normalized volume of the lattice
[Link]
Shaping Gain
[Link]
Combined Savings – Coding & Shaping Gains
Properties of Modulation Schemes for
Power-Limited Region
Modulation Schemes for
Power Limited Regions
Orthogonal FSK
Orthogonal Modulation – Vector Representation
Minimum Euclidean Distance
Probability of Error – Behavior
Performance Comparison of Modulation Schemes
Achieving Capacity in Power Limited Region
[Link]
Channel Coding
