Wireless ad hoc networks: cross layer opportunities
NSF workshopWashington DC Aug 27-28
Mario Gerla
Computer Science Dept, UCLA
www.cs.ucla.edu
Ad hoc networking Current Status
Leading Applications
• Tactical battlefield: – no infrastructure
• Civilian emergency:– infrastructure, if present, was destroyed
• Critical Requirements: scalability, survivability, 100% reliable, QoS, jam protection, etc
• Non critical: Cost, Standards, Privacy
SURVEILLANCE MISSION
SURVEILLANCE MISSION
AIR-TO-AIR MISSION
STRIKE MISSION
FRIENDLY GROUND CONTROL
(MOBILE)
RESUPPLY MISSION
SATELLITE COMMS
Unmanned Control Platform
COMM/TASKING
COMM/TASKING
MannedControl Platform
COMM/TASKING
UAV-UAV NETWORK
Tactical Ad Hoc Network
UAV-UGV NETWORK
Emerging Landscape : “Opportunistic” Ad Hoc networks
Recreational, commercial, education applications
• Vehicle networks• Workgroups (eg, sharing 3G via Bluetooth)• Massive Network games • Patient monitoring
Access to Internet? • available, but - “bypass it” with “ad hoc” if too costly or
inadequate
Tolerant to delays: DTNs
Critical: Cost, Privacy, security, standards
Car to Car communications for Safe Driving
Vehicle type: Cadillac XLRCurb weight: 3,547 lbsSpeed: 65 mphAcceleration: - 5m/sec^2Coefficient of friction: .65Driver Attention: YesEtc.
Vehicle type: Cadillac XLRCurb weight: 3,547 lbsSpeed: 45 mphAcceleration: - 20m/sec^2Coefficient of friction: .65Driver Attention: NoEtc.
Vehicle type: Cadillac XLRCurb weight: 3,547 lbsSpeed: 75 mphAcceleration: + 20m/sec^2Coefficient of friction: .65Driver Attention: YesEtc.
Vehicle type: Cadillac XLRCurb weight: 3,547 lbsSpeed: 75 mphAcceleration: + 10m/sec^2Coefficient of friction: .65Driver Attention: YesEtc.
Alert Status: None
Alert Status: Passing Vehicle on left
Alert Status: Inattentive Driver on Right
Alert Status: None
Alert Status: Slowing vehicle aheadAlert Status: Passing vehicle on left
Co-operative Download: Car Torrent
Vehicle-Vehicle Communication
Internet
Exchanging Pieces of File Later
Vehicular Sensor Network (VSN)
VSN-enabled vehicle
Inter -vehiclecommunications
Vehicle -to-roadsidecommunications
Roadside base station
Vid e o Ch e m.
Sensors
S to ra g e
Systems
P ro c.
Personal Networking: BlueTorrent
A A
B
C
B
C
D D
A
D
B
C
Patient Monitoring
Nurses upload patient data; share data files
in P2P mode
1. Future expectations on wireless network research
• Network layer more tightly coupled with applications– Content sharing, environement sensing
• Besides data forwarding, additional services:– Location aware service discovery,
– content based routing;
– P2P networking
– Data collection, processing, filtering, storage, dissemination
• Network layer design must interact with:– applications
– PHY Layer
2. Major recent advances/breakthroughs in the physical layer
• Cognitive radios (spectrum scavenging)• MIMOs (for flexible topology designs;
interference mitigation etc)• Cooperative radios• Multi radio devices (BT, 802.11, 3G, etc)
3. Algorithms must adjust to PHY layer
“PHY layer aware” MAC, Network and Transport designs
Examples (based on MIMO):
• Topology control• A MIMO aware MAC protocol: SPACE-MAC• Multi-path Routing & MIMO• TCP & MIMO
MIMO Topology Control/Routing
• Topology control:– Exploit mode flexibility to dynamically shape topology– Meet different customer requirements
Topology with high capacity links: disconnected network
Topology with low capacity links: fully connected network
300Mbps
10Mbps
SPACE2 MAC
When A wishes to transmit to B
A
B
D
F1) A sends RTS to B; F and D learn about A
2) B responds with CTS; F and D learn about B
SPACE MAC (cont)
3) F and D beamform such that signals from/to B and A are nulled; then, A and B start talking
A
B
F
D
4) After A and B pair is established, F and D pair also can talk
Two-Path Routing using MIMO
• S sends two independent streams simultaneously to R
• Assume 2 antennas at each node (but extendible to systems with more antennas).
SS RR
sender receiver
relay nodes
MIMO yields 6-fold throughput gain
• In the traditional relay mode, the capacity is C/3.• Simulcast achieves 6-fold throughput increase.
SS RR
sender receiver
TCP and MIMO in Ad Hoc Networks
• Consider three flows in the same wireless domain
• As the flows get closer to each other:– Interference builds up
– Throughput decreases
– Fairness suffers
• Can MIMO Help?
FTP 1
FTP 2
FTP 3
(100, 100) (600, 100)
(350, 350)
(350, 700)(0, 700)
(350, 1050)
(100, 1300) (600, 1300)
(700, 700)
TCP over SPACE MAC (MIMO)Distance = 400m (interference range)
3 F T P / T C P F l o w s
0
50
100
150
200
250
300
350
802.11 S P A C E - M A C
M A C P r otoc ol
Throughput (Kbits/s)
F low 1 F low 2 F low 3
Fig 0. The throughput of 3 FTP/TCP flows with the distance between flows being 400m
TCP over SPACE MAC (MIMO)Distance = 350m (tx range)
3 F T P F l o w s
0
50
100
150
200
250
300
350
802.11 Space-MAC
M A C P rotoc ol
Throughput (Kbits/s)
F low 1 F low 2 F low 3
Fig 0. The throughput of 3 FTP/TCP flows with the distance between flows being 350m
Identify gaps
• Question: How to exploit the wealth of PHY emerging technology?
• Do not limit your scope to LINK capacity gains• Look for cross layer optimization opportunities at
all layers:– MAC
– Network (routing, topology control, multicast, bandwidth scavenging, etc)
– transport,
– applications and PHY layer
The End
Thank You
Simul-Cast
Brian Choi
Mario Gerla
MIMO System Model
s(t)WHVH = r(t)
weight vector w1 = [w11 w21 … wm1]
W V
Assumptions
• Fading is flat (i.e. freq. independent).• Channel is symmetric and quasi-static.• Two subchannels - control channel and data channel• Rich scattering - H is full-rank• Antenna’s capable of transmitting and receiving
signals simultaneously.• We ignore additive channel noise.• Perfect sychronization between nodes
Two Path Routing Problem
• S sends two independent streams under two paths simultaneously to R.
• Assume 2 antennas at each node (but extensible to systems with more antennas).
SS RR
sender receiver
relay nodes
The 6-fold Benefit of MIMO
• If C = (capacity of a point-to-point link) in the traditional relay mode, the capacity is C/3.
• Simulcast achieves 6X throughput increase.
SS RR
sender receiver
Sender
• A wants to send a stream (s1(t)) to B and another stream (s2(t)) to C simultaneously.
AA
BB
CC
s1(t)
s2(t)
s(t) = [s1(t) s2(t)]
Sender: Linear Coding
• B receives rB(t) = s(t)WAHABWBH.
• For B to recover s1(t), B must consume 2 degrees of freedom.
AA
BB
CC
s1(t)
s2(t)
s(t) = [s1(t) s2(t)] HAB
HAC
rB1(t)
rB2(t)
rC1(t)
rC2(t)
WA
WB
Sender: Pre-coding
• If A knows the channel and the steering matrices of B and C, then A can precode its data such that s1(t) is received at rB1(t), s2(t) is received at rC1(t), without interfering each other.
• B and C needs to comsume only one DOF each.
AA
BB
CC
s1(t)
s2(t)
rB1(t)
rB2(t)
rC1(t)
rC2(t)
Dirty Paper Coding
HABwB1H
HACwC1H
Let H = = QR
QR factorization, Q = unitary, R = upper triangular
AA
BB
CC
s1(t)
s2(t)
rB1(t)
rB2(t)
rC1(t)
rC2(t)
wB1
wC1
Dirty Paper Coding
• Let r(t) = [rB1(t) rC1(t)]. Then r(t) = s(t)H.• Multiply s(t) by QH, such that s’(t) = s(t)QH. • Then r(t) = s’(t)H = s(t)QHH = s(t)QHQR = s(t)R
• rB1(t) = s1(t)R1,1 (no interference)
• rC1(t) = s1(t)R1,2 + s2(t)R2,2
• Sender can estimate this interference and subtract it from s(t) before transmitting.
(interference!)
Relay Node
• There is one DOF left for us to use. We use it to simultaneously relay the received data to the next node.
• We set weight vectors such that they are orthogonal to each other.
used to receive data from the previous node
used to send a stream to the next node
Receiver
• This reduces to the problem of spatial multiplexing.• If R knows the channels and the weight vectors used
for both streams, then R can decode the received data.
AA
BB
RR
HAR
HBR
Network-wise Benefit
• If C = (capacity of a point-to-point link) in the traditional relay mode, the capacity is C/3.
• Simulcast achieves 6X throughput increase.
SS RR
sender receiver
Multiple Paths
• We can run OLSR-type of routing protocol for the nodes to pre-determine the paths.
• This suggests a cross-layer approach (between network layer and MAC layer).
Summary
• With MIMO and Pre-coding techniques, one can effectively reduce the DOF consumption at the receiving nodes.
• We can utilize the idle DOF to relay the data simultaneously.
• With two independent simultaneous paths, we can achieve up to 6X throughput increase.