5. SDMA
Spatial Division MultipleAccess
---- Uses an array of antennas to provide control of space
by providing virtual channels in an angle domain
6. Directional Antennas
Sectorised antenna
1) switched beam system
•Use a number of fixed beams
•Select one of several beams to
enhance receive signals
2) adaptive array system
•Be able to change its antenna pattern
dynamically;
Smart antenna
7. System Model
Uniform LinearArray of M elements
c
d
c
l
sin
tfj c
etmtx 2
1 )()(
d
)(2
2 )()(
tfj c
etmtx
))1((2
))1(()(
Mtfj
M
c
eMtmtx
8. System Model
)()( tmtm
tfj c
etmtx 2
1 )()(
c
dj
etxtx
sin2
12 )()(
Narrow Band array processing Assumption:
c
dMj
M etxtx
sin)1(2
1 )()(
sin)1(2
sin22
sin2
1
c
c
c
d
Mj
d
j
d
j
e
e
e
S
Array response vector
9. System Model
The Beam-former Structure
)(
)()(
1
*
tXw
txwty
H
i
M
i
i
Mw
w
w
w
2
1
)(
)(
)(
)( 2
1
tx
tx
tx
tX
M
)(1 tx
)(2 tx
)(txM
1
2
M
*
1w
*
1w
*
Mw
)(ty
12. Optimum Beamformer Design
Maximum SINR beamformer
SRS
SR
w
NI
H
NI
SINR
1
1
max
Under different criterions
Mean-Square-Error optimum beamformer
2
)(tmEP
SPRwMMSE
1
13. Optimum Beamformer Design
Minimum-Variance-Distortionless-Response beamformer
SRS
SR
w HMVDR
1
1
Under different criterion
Maximum Likelihood optimal beamformer
SRS
SR
w
NI
H
NI
ML
1
1
14. Practical Issues
In practice, neither R nor RI+N is available to calculate the optimal
weights of the array;
In practice, direction of arrival (DOA) is also unknown.
Issues
Solution
Adaptive beamforming algorithms – the weights
are adjusted by some means using the available
information derived from the array output, array signal
and so on to make an estimation of the optimal weights;
DOA estimation methods
17. Adaptive Beamforming Algorithms
1. SMIAlgorithm (Sample Matrix Inverse)
N
i
H
iiN rr
N
R
1
1ˆ
Estimate R using N samples:
n
rr
R
n
n
R
H
nn
nn
1
ˆ1ˆ
nn
H
n
n
H
nnn
nn
rRrn
RrrR
R
n
n
R
1
1
1
1
1
11
1
1
ˆ)1(
)ˆˆ
ˆ1ˆ
,....2,1
0
ˆ 1
0
k
c
cIR
Use matrix inversion lemma:
Then:
SRw nn
1ˆ
18. Adaptive Beamforming Algorithms
2. LMS Algorithm (Least Mean Square)
**
1 )( nnnnn
H
nnnn erwdwrrww
nn
H
nn drwe
• Need training bits and calculate the error between the received
signal after beamforming and desired signal;
• The step size u decides the convergence of LMS algorithm;
• Based on how to choose u, we have a set of LMS algorithm,
“unconstraint LMS”, “normalized LMS”, “constraint LMS”.
According to orthogonality principle (data| error) of MMSE beamformer:
0)()()( *
tdwtrtrE H
Solution:
19. 3. RLS Algorithm (Recursive Least Square)
Adaptive Beamforming Algorithms
)()(ˆ *
1
1
1 nn
H
nnnn dwrrnRww
Given n samples of received signal r(t), consider the optimization
problem—minimize the cumulative square error
n
k
k
kn
e
0
2
min 10
Solution:
• In some situation LMS algorithm will converge with very slow
speed, and this problem can be solved with RLS algorithm.
20. Adaptive Beamforming Algorithms
4. CMA (Constant Modulus Algorithm)
Assume the desired signal has a constant modulus, the existence of an
interference causes fluctuation in the amplitude of the array output.
Consider the optimization problem:
2
2
2
)(
2
1
min AtrwE
H
Solution:
)( 2
2
1 Arwwrrww n
H
nn
H
nnnn
• This is a blind online adaptation, i.e., don’t need training bits
• CMA is useful for eliminating correlated arrivals with different magnitude
and is effective for constant modulated envelope signals such as GMSK
and QPSK
21. DOA Estimation Method
1. MFAlgorithm (Matched Filter)
2. MVDRAlgorithm
3. MUSICAlgorithm (MUltiple SIgnal Classification)
22. DOA Estimation Method
1. MF Algorithm (Matched Filter)
The total output power of the conventional beamformer is:
wRwwtrtrEwtrwEtyEP
H
H
HH
)()()()(
2
2
• The output power is maximized when
• The beam is scanned over the angular region say,(-900,900), in discrete
steps and calculate the output power as a function of AOA
• The output power as a function of AOA is often termed as the spatial
spectrum
• The DOA can be estimated by locating peaks in the spatial spectrum
• This works well when there is only one signal present
• But when there is more than one signal present, the array output power
contains contribution from the desired signal as well as the undesired
ones from other directions, hence has poor resolution
0Sw
23. 2. MVDR Algorithm
DOA Estimation Method
This technique form a beam in the desired look direction while taking into
consideration of forming nulls in the direction of interfering signals.
wRwtyE
H
min)(min
2
1Swtosubject
H
Solution:
SRS
P HMVDR
1
1
)(
• By computing and plotting pMVDR over the whole angle range, the DOA’s
can be estimated by locating the peaks in the spectrum
• MVDR algorithm provides a better resolution when compared to MF
algorithm
• MVDR algorithm requires the computation of a matrix inverse, which can
be expensive for large arrays
24. DOA Estimation Method
Comparison of resolution performance of MF and MVDR algorithms
Scenario: Two signals of equal power at SNR of 20dB arrive at a 6-element uniformly
spaced array at angles 90 and 100 degrees, respectively
25. 3. MUSIC Algorithm (MUltiple SIgnal Classification)
DOA Estimation Method
MUSIC is a high resolution multiple signal classification technique based
on exploiting the eigenstructure of the input covariance matrix.
Step 1: Collect input samples and estimate the input covariance matrix
N
i
H
ii rr
N
R
1
1ˆ
Step 2: Perform eigen decomposition
VVRˆ
},,,{ 21 Mdiag M 21
MqqqV ,,, 21
26. 3. MUSIC Algorithm (MUltiple SIgnal Classification)
DOA Estimation Method
Step 3: Estimate the number of signals based on the fact :
DMK ˆ
• The first K eigen vectors represent the signal subspace, while the last
M-K eigen vectors represent the noise subspace
• The last M-K eigen values are equal and equal to the noise variance
find the D smallest eigen values that almost equal to each other
Step 4: Compute the MUSIC spectrum
SVVS
P
H
nn
HMUSIC
1
)( MKKn qqqV ,,, 21
find the largest peaks of Pmusic to obtain estimates of DOAKˆ
27. DOA Estimation Method
Comparison of resolution performance of MVDR and MUSIC
Scenario: Two signals of equal power at SNR of 20dB arrive at a 6-element uniformly
spaced array at angles 90 and 95 degrees, respectively
28. Summary of Part I
• System model
• Optimum beamformer design
• Adaptive beamforming algorithms
1) SMI
2) LMS
3) RLS
4) CMA
• DOA estimation method
1) MF
2) MVDR
3) MUSIC
29. Part II: Schemes using directional antennas
in MAC layer of ad hoc network
31. RTS/CTS mechanism in 802.11
Nodes are assumed to transmit using omni-directional antennas.
Both RTS and CTS packet contain the proposed duration of data
transmission
The area covered by the transmission range of both the sender(node B) and
the receiver (node C) is reserved during the data transfer
This mechanism reduce collisions due to the hidden terminal problem
However, it waste a large portion of network capacity.
32. Vaidya Scheme 1
Assumption:
Each node knows its exact location and the location of its neighbors
Each node is equipped with directional antennas
If node X received RTS or CTS related to other nodes, then node X will
not transmit anything in that direction until that other transfer is
completed
That direction or antenna element would be said to be “blocked”
While one directional at some node be blocked, other directional at the
same nodes may not be blocked, allowing transmission using the
unblocked antenna
33. Vaidya Scheme 1
A B C D E
DRTS
OCTS
OCTS
DATA
ACK
DRTS
OCTS
DATA
ACK
OCTS
34. Vaidya Scheme 1
Utilize a directional antenna for sending the RTS (DRTS), whereas
CTS are transmitted in all directions (OCTS).
Data andACK packets are sent directionally.
Any other node that hears the OCTS only blocks the antenna on
which the OCTS was received.
36. Vaidya Scheme 2
A node uses two types RTS packets: DRTS and ORTS according to the
following rules:
1) if none of the directional antennas at node X are blocked, then node X
will send ORTS;
2) otherwise, node X will send a DRTS provided that the desired
directional antenna is not blocked.
37. A B C D
ORTS
OCTS
OCTS
DATA
ACK
Vaidya Scheme 2
F
ORTS
DRTS
40. Nasipuri Scheme
NodeA that wishes to send a data packet to B first sends an omni-
directional RTS packet
Node B receives RTS correctly and responds by transmitting a CTS
packet, again on all directions.
In the meanwhile, B can do DOA estimation from receiving RTS packet
Similarly, node A estimates the direction of B while receiving the CTS
packet.
Then node A will proceed to transmit the data packets on the antenna
facing the direction of B.
43. Bagrodia Scheme
DirectionalVirtual Carrier Sensing(DVCS)
Three primary capabilities are added to original 802.11 MAC protocol for
directional communication with DVCS:
1) caching theAngle ofArrival (AOA)
2) beam locking and unlocking
3) the use of Directional NetworkAllocationVector (DNAV)
44. Bagrodia Scheme
1.AOA caching
Each node caches estimatedAOAs from neighboring nodes whenever it
hears any signal, regardless of whether the signal is sent to it or not
When node X has data to send, it searches its cache for theAOA
information, if theAOA is found, the node will send a directional RTS,
otherwise, the RTS is send omni-directionally.
The node updates itsAOA information each time it receives a newer
signal from the same neighbor.
It also invalidates the cache in case if it fails to get the CTS after 4
directional RTS transmission.
45. Bagrodia Scheme
2. Beam locking and unlocking
A B
B
(1)RTS
(2)CTS(3)Data
(4)ACK
When a node gets an RTS, it locks its beam pattern
towards the source to transmit CTS
The source locks the beam pattern after it receives CTS .
The beam patterns at both sides are used for both
transmission and reception, and are unlocked after ACK is
completed.
46. Bagrodia Scheme
3. DNAV setting
DNAV is a directional version of NAV(used in the original 802.11
MAC), which reserves the channel for others only in a range of
directions.
Available directions for transmission
In the fig:
Three DNAVs are set up
towards 300, 750 and 3000 with
600 width.
Until the expiration of these
DNAVs, this mode cannot
transmit any signals with
direction between 0-1050 or
270-3300 , but is allowed to
transmit signals towards 105-
2700 and 330-3600
47. Bagrodia Scheme
A network situation where DVCS can improve the network
capacity with DNAVs
F
BD
E
A C
49. Summary of Part II
Comparison of four schemes
RTS CTS Data ACK
802.11 omni omni omni omni
Vaidya 1 dir. omni dir. dir.
Vaidya 2 dir./omni omni dir. dir.
Nasipuri omni omni dir. dir.
Bagrodia dir./omni dir. dir. dir.
50. Conclusion
smart antenna is a technology for wireless systems that use a set of antenna
elements in an array.The signal from these antenna elements are combined
to form a movable beam pattern that can be steered to a desired direction
smart antennas enable spatial reuse and they increase the communication
range because of the directivity of the antennas
smart antennas can be beneficial for wireless ad hoc networks to enhance
the capacity of the network
To best utilize directional antennas, a suitable MAC protocol must be
designed
If the locations are unknown , DOA estimation may be needed before
sending directional signals
51. reference
LiliWei 2002 presentation on smart antennae
J.C.Liberti,T.S.Rappaport, “Smart antennas for wireless communications: IS-95 and
third generation CDMA applications”
L.C.Godara, “Application of antenna arrays to mobile communicaitions, part I:
performance improvement, feasiblility, and system considerations”
L.C.Godara, “Application of antenna arrays to mobile communications, part II: beam-
forming and direction-of-arrival considerations”
Y.b Ko,V.Shankarkumar and N.Vaidya, “Medium access control protocols using
directional antennas in ad hoc networks”
A.Nasipuri, S.Ye, J.You and R.Hiromoto, “A MAC protocol for mobile ad hoc networks
using directional antennas”
M.Takai, J.Martin,A.Ren and R.Bagrodia, “Directional virtual carrier sensing for
directional antennas in mobile ad hoc networks”
S.Bellofiore, J.Foutz, etc.. “Smart antenna system analysis, integration and performance
for mobile ad-hoc networks (MANETs)