Midterm Review

Midterm only covers material from lectures and HWs

 Overview of Wireless Systems:

 Nothing from Chapter 1 is on the MT

 Signal Propagation and Channel Models

 Chapter 2.1-2.4, 2.6-2.10

 Modulation and Performance Metrics

 Chapter 3.1,3.2.1-3.2.2, 3.3

 Fundamental Capacity Limits

 Chapter 4

 Impact of Channel on Performance

 Chapter 6

 Diversity Techniques
 Chapter 7.1,7.2.1-7.2.2,7.2.4,7.3.1,7.4.1
 Future Wireless Not on MT

Networks
Ubiquitous Communication Among People and Devices
Wireless Internet
access
Nth generation
Cellular
Wireless Ad Hoc
Networks
Sensor Networks
Wireless
• HardEntertainment
Delay/Energy Constraints
• Hard
SmartRate Requirements
Homes/Spaces
Automated
Highways
All this and more…
 Not on MT

Design Challenges
 Wirelesschannels are a difficult and
capacity-limited broadcast
communications medium
 Traffic
patterns, user locations, and
network conditions are constantly
changing
 Applicationsare heterogeneous with
hard constraints that must be met by
the network
 Energy,
delay, and rate constraints
change design principles across all
 Current/Futre Wireless
Not on MT

Systems
 Current Systems
 4G Cellular Systems
 802.11b/a/n/ac Wireless LANs
 Satellite Systems
 Paging Systems
 Bluetooth
 Zigbee radios

 Emerging Systems (Can cover in bonus

lecture)
 Ad hoc/mesh wireless networks
 Cognitive radio networks
 Wireless sensor networks
 Energy-harvesting radios
 Distributed control networks

 Signal Propagation
 Path Loss
 Free space, 2-path,…
 Simplified model

 d0  d
Pr Pt K   , 2  8
d
 Shadowing
 dB value is Gaussian
 Find path loss Pr/Pt
exponent and shadow
STD by curve fitting
 Multipath
 Ray tracing d=vt
 Statistical model
 Outage Probability
and Cell Coverage
Area
Pr
 Path loss: circular cells
 Path loss+shadowing: amoeba
cells
Tradeoff between coverage and
interference
 Outage probability
Probabilityreceived power below
given minimum
 Cell coverage area
% of cell locations at desired power
 Statistical Multipath
Model

 Random # of multipath components, each

with varying amplitude, phase, doppler, and
delay
 Leads to time-varyingN channel impulse
response c ( , t )   n (t )e
 j ( t )
n
 (   n (t ))
n 1

 Narrowband channel
 No signal distortion, just a complex amplitude
gain
 Wideband Channels
 Individual multipath components
resolvable
 True when time difference between
components
  1 / B
exceeds signal
Bu  Bc
bandwidth
u

 1  2 Ac(f)
t f
Wideband 0 Bc

r Doppler Power Spectrum

s(t,r)=FDt[Ac(t,Dt
 Scattering t
function Delay Power Spectrum

)]
Yields delay spread/coherence BW
Capacity of Flat Fading
Channels
 Channel Capacity
Maximum data rate that can be
transmitted over a channel with
arbitrarily small error
 Capacity
of AWGN Channel:
Blog2[1+g] bps
g=P /(N0B) is the receiver SNR
r
 Capacity of Flat-Fading Channels
Nothing known: capacity typically
zero

C  B log
Fading 2 1   p ( )Known
Statistics d B log(few
2 (1   )
results)
 0
 Capacity in Flat-Fading: g known at
TX/RX max 
 P ( ) 
C  B log 2  1   p ( )d
P () : E[ P ( )] P 0  P 

Optimal Rate and Power

Waterfilling
 
P ( )  1  1
 1
Adaptation
0
 0

P  0 else 0

 1
C 
 log 2   p()d. g
B 0  0  g0 g
 Channel Inversion
 Fadinginverted to maintain
constant SNR
 Simplifies design (fixed rate)
 Greatly reduces capacity
Capacity is zero in Rayleigh
fading
 Truncated inversion
Invert channel above cutoff fade
depth
Constant SNR (fixed rate) above
 Frequency Selective
Fading Channels
 For time-invariant channels,
capacity achieved by water-
filling in frequency
 Capacity of time-varying channel
unknown
 Approximate by dividing into
subbands
Each subband has width Bc
1/|H(f)|2
Independent fading in each
subband P Bc
Capacity is the sum of subband f
 AWGN: MPSK and
MQAM
 ML detection induces decision
regions
Example: 8PSK
dmin

P depends on
s
# of nearest neighbors
Minimum distance d (depends on

g s)   min
Ps  M Q  M s
Approximate expression
 Linear Modulation in
Fading
 In fading gs and therefore Ps
random
 Metrics: outage, average P ,
s
T
combined
s outage and P
average.
Outag s

e Ps(target)

Ps

Ts Ps Ps (s ) p (s )ds

 Moment Generating
Function Approach

 Simplifies average Ps calculation

 Usesalternate Q function
representation
P reduces to MGF of gs
s
distribution
 Closedform or simple numerical
calculation for general fading
 Doppler Effects
 Highdoppler causes channel
phase to decorrelate between
symbols
 Leads to an irreducible error
floor for differential modulation
Increasing power does not reduce
error
 Error floor depends on fDTb as
 Delay Spread (ISI)
Effects
 Delay
spread exceeding a symbol time
causes ISI (self interference).
1 2 4
Delay Tm

0 Ts Tm

3 5
 ISI leads to irreducible error floor:
 Increasing signal power increases ISI
power
 ISI
imposes data rate constraint:
Ts>>Tm (Rs<<Bc)
 Diversity
 Send bits over independent fading
paths
 Combine paths to mitigate fading effects.

 Independent fading paths

 Space, time, frequency, polarization
diversity.

 Combining techniques
 Selection combining (SC)
 Maximal ratio combining (MRC)

 Can have diversity at TX or RX


 Selection Combining
 Selects the path with the highest
gain
 Combiner SNR is the maximum of
the branch SNRs.
 CDF easy to obtain (Pip(gi<gthr)), pdf
found by differentiating the CDF
P obtained from CDF. Average Ps
out

typically found numerically

 Diminishing returns with number of
 MRC and its
Performance
 With MRC, gS=gi for branch SNRs gi
 Optimal technique to maximize output
SNR
 Yields 20-40 dB performance gains
 Distribution of g hard to obtain
S

 Standard average BER calculation

Psb Pbs( ) p( )d ...Psb ( ) p(1 ) * p(2 ) * ... * p( M )d1d2 ...d M

 Hard to obtain in closed form

 Integral often diverges

 MGF Approach:

 TX diversity gain with CSI same as RX

 Main Points
 Wirelesschannels introduce path-
loss, shadowing and multipath fading
 Shadowing introduced outage
 Flat-fading causes large power
fluctuations
 ISI causes self-interference
 Performance
of digital
communications in wireless channels
random
 Characterized by outage probability and
average probability of error in flat-fading
 Characterized by irreducible error floors
in ISI/Doppler