1. SPACE-MAC: Enable spatial-reuse using MIMO channel aware MAC

2. SPACE-MAC Targets Beamforming MIMO
Enables multiple communicating node pairs by nullifying out known interferers
Solves two problems:
How to identify interferers?
How to null them?
RTS/CTS exchange
Zero-forcing beamforming B F A D

3. SPACE-MAC PHY Model r(t) = wRTHwTs(t) + wRTn where
Transmitter Receiver H
r(t) = wRTHwTs(t) + wRTn
where
wT = [wT1 wT2 wT3]T: tx weights,
wR = [wR1 wR2 wR3]T: rx weights,
H: 3x3 channel matrix,
n: Noise
wT1
wR1
s(t)
r(t)
wT2
wR2 
wT3
wR3

4. SPACE-MAC PHY Model (cont.)
r(t) = wRTHwTs(t) + n'
where
wT = [wT1 wT2 wT3]T: tx weights,
wR = [wR1 wR2 wR3]T: rx weights,
H: 3x3 channel matrix,
n’: Weighted noise
r(t) = wRThTs(t) + n'
H' = HwT: 3x1 channel vector
Estimate hT on
reception of RTS/CTS
Transmitter Receiver
wT1
wR1
s(t)
r(t)
wT2
wR2 
wT3
wR3
Real channel
Virtual channel hT
hT = HwT
wR1
s(t)
r(t)
wR2 
wR3
Transmitter Receiver

5. Operation of SPACE-MAC
When A wishes to transmit to B B F D A

6. Operation of SPACE-MAC
A sends RTS to B; F and D learns about A B F D A

7. Operation of SPACE-MAC
B responds with CTS; F and D learns about B B F D A

8. Operation of SPACE-MAC
D and F beamform s.t. signals from/to B and A are nulled; A and B start communicating B F D A

9. Operation of SPACE-MAC
While A and B are communicating, D and F also can start talking B F D A

10. Nullifying r(t) = wDThAs(t) + n'
On reception of RTS from A, Node D learns hA = HwA
Find wD
s.t. wDThA = 0
HwA = hA
wD1
s(t)
wD2 
wD3
Transmitter A Receiver D

11. Beamforming at Node S
Beamforming problem in presence of k interferers:
w* = wSHRwS (total interference energy)
where
HiS = Channel between Node i and S,
wi = weightvector of Node i,
ZiS = HiSwiwiHHiSH,
Rk =
w* is an eigenvector corresponding to min of Rk
Fact: wSHRkwS is bounded by min/max eigenvalues of Rk 1 2 3 4 S
We formulated the problem as an optimization problem. All the capability of the space-mac is capture in this formula.
Gain: wSHH1Sw1w1HH1SHwS

12. Simulation Qualnet Assuming quasi-static Rayleigh fading: Settings
Each channel coefficient hij is a complex Gaussian random variable with zero mean unit variance and static during a RTS/CTS/DATA/ACK handshaking
Settings
20 nodes
Saturated network: nodes always having packets
Fixed packet size of 1024B
2Mpbs link

13. Simulation results
As number of antenna increase the number of simultaneous transmissions that the network can accept also increases so we get throughput enhancement.
Throughput: (Total “Data” received) / (simulation duration)

14. Mux-SMAC Problems are: Difference from SPACE-MAC?
Spatial multiplexing MIMO
Allows multiple communicating node pairs as well as multiple streams
Problems are:
How to identify competitors?
=> RTS/CTS exchanges
How to suppress/not to disturb competitors?
How many steams?
Difference from SPACE-MAC?

15. Mux-SMAC PHY model Transmitter Receiver s1(t) r1(t)   r2(t)   … …
w11  
v11
r2(t)  
w12
v12 … … . .
rm(t)  
w1n
v1n
H = [hij]
s2(t)
w21
v21
w21
v21 . .
Steering Matrix
W = [wij]
w2n
v2n . .
sm(t)
wm1
vm1
wm2
vm1 . .
wmn
vmn