Transcript A Distributed Relay-Assignment Algorithm for Cooperative

A Distributed Relay-Assignment Algorithm
for Cooperative Communications in
Wireless Networks
ICC 2006
Ahmed K. Sadek, Zhu Han, and K. J. Ray Liu
Department of Electrical and Computer Engineering,
and Institute for Systems Research
University of Maryland, College Park
Presented by Sookhyun Yang
Nearest Neighbor Protocol (NNP) Scheme
• Is the relay assignment scheme at the cooperative networking
– Selects the nearest neighbor from the source as the relay node
– Considers only the uplink (toward the BS/AP)
• Neighbor discovery
– Each relay sends out “Hello” message
– Each source node can know its distance to the BS/AP using TOA (Time of Arrival)
Relay
Nearest neighbor relay
BS/AP
Source
Relay
Direct transmission
Relay
2/14
Source’s Location Distribution Model
• Derive PDF for the source location distribution when the
distance between the user and the BS/AP is
– The user’s angle is uniformly distributed between, [0, 360o)
Source
: Cell radius
BS/AP
: the distance from
the source to
BS/AP
3/14
Received Signal Model
• Considered wireless link characteristics for received signal model
–
–
–
–
Random Rayleigh fading channel between two nodes Propagation path loss – path loss exponent
Additive white Gaussian noise
No mutual interference because the nodes use the orthogonal channel
• Each node has a single-element antenna and half-duplex mode
Voltage of
received
power
Depends on Antenna’s
design (constant)
Path loss
exponent
Transmitted data with
unit power (x=1)
noise
Transmitted power
(
Distance from
source to
destination
Channel fading
are assumed to be the same for all nodes)
4/14
Metric: Outage
• Is the event that the received SNR falls below a certain
threshold
– It the received SNR is higher than the threshold, the receiver is
assumed to be able to decode the received message with negligible
probability of error
– If the outage occurs, the packet is considered lost
• Outage probability
threshold
• For comparing the bandwidth efficiency, “outage
probability” is computed for both the direct transmission
and the nearest neighbor based cooperative transmission
5/14
Outage Probability for the Direct Transmission
Outage probability for the direct transmission when the source locates
with the distance
from BS/AP
Average outage probability over the cell
The probability that the
source is at the distance
6/14
Outage Probability for NNP-based
Cooperative Transmission
rsl
Source
Nearest Neighbor Relay
rld
rsd
BS/AP
7/14
Outage Probability for NNP-based
Cooperative Transmission
Outage probability for NNP-based cooperative transmission
Outage probability for the direct
transmission
Outage probability for the
transmission through relay
Average outage probability over the cell
depends on
Outage probability
PDF of source location
PDF of the nearest neighbor relay’s location?
8/14
Outage Probability for NNP-based
Cooperative Transmission
• The probability that the nearest neighbor is at distance
from the source is equivalent to “the probability that the
shaded area is empty”
Nearest Neighbor Relay
Source
BS/AP
Average outage probability for the nearest neighbor relay over the cell
=
9/14
Approximated Outage Probability Formula
for NNP-based Cooperative Transmission
• Because the formula in the previous slide can only be calculated
numerically, they derive an approximated expression
• Assumption for the approximation
– 1. The outage probability at the nearest neighbor relay is very low
– 2. Consider the worst case of the nearest neighbor selection, “a” or “b” when
=
Approximation
10/14
Simulation Setup
• Is modeled as a random Rayleigh fading channel
–
–
–
–
–
–
Similarly configures with the indoor WLAN
Cell radius is taken between 10m and 100m
AWGN (Additive white Gaussian noise): variance = -70dBm
Path loss exponent = 2.6
The number of users in the cell attached to the AP = 10
SNR threshold
= 20dB
• For comparing the bandwidth efficiency (low outage probability), the
average transmitted power is kept equal in the direct transmission and
the cooperative transmission
– The transmitted power does not affect the outage probability?
• Plotted the theoretical outage performance and the simulation results
for both direct transmission and cooperative transmission
– Which simulator?
11/14
Average Outage Probability (y-axis)
vs. Cell Radius (x-axis)
Direct transmission
Cooperative transmission
The cell coverage will increase in the cooperative
transmission!
12/14
Average Outage Probability (y-axis)
vs. Transmission Power (x-axis)
Direct transmission
Cooperative transmission
The cooperative
transmission is energy
efficient
13/14
Discussion
• They did not show that the nearest neighbor can have the
best performance comparing with the other neighbors
– What about the neighbor which locates in the middle between the
source and the BS/AP?
• They fixed the transmitted power as the same in both cases?
– They considered the orthogonal channels for the cooperative
transmission, so the direct transmission will use less transmitted power
than the cooperative transmission
14/14
Q&A
(1)
Received power
Transmitted power
Pt
2
Pr 
rsd  A
4
rsd
Distance
Pr  Pt  rsd

Pt
For transmitting the data, the antenna
focuses on the specific direction
(antenna directivity)
Pr  K  Pt  rsd
Pr
s
Path loss exponent

d
rsd
d
s
Pt
Depends on Antenna design
16/14
Pr
(1)
From the previous slide
Pr  K  Pt  rsd
The relationship between Voltage and Power

ysd  Pr
Voltage
ysd  K  Pt  rsd

Multi-path fading

ysd  K  Pt  rsd hsd x  nsd
Go to the next slide
Noise
Transmitted data per unit power
(Usually x is set as “1”)
17/14
?
(2)
From the previous slide

ysd  K  Pt  rsd hsd x  nsd
Compute the voltage of signal without noise, where x=1
Voltage

ysd  K  Pt  rsd hsd  Psignal
Power
Psignal  K  Pt  rsd
SNR(rsd ) 
Psignal
Pnoise


 hsd

sd
K  Pt  r
2
 hsd
nsd
Go to the next slide
2


sd
K  Pt  r
N0
 hsd
18/14
2
(3)
From the previous slide
SNR(rsd ) 

sd
K  Pt  r
 hsd
2
N0
P( SNR(rsd )   )  P(
Use

sd
K  Pt  r
 hsd
2
  )  P( hsd 
2
N0
  N0

sd
K  Pt  r
f h 2 ( x)  e  x , F h 2 ( x )  1  e  x
sd
sd
(h^2 is exponential distribution)
P( SNR(rsd )   )  P( hsd 
2
 1 e
  N0

 K  P r 
t sd





  N0

sd
K  Pt  r
 1 e

 N 0  rsd

 K  Pt

)  Fh
sd




2
(
  N0

sd
K  Pt  r
)
19/14
)