Andrea Goldsmith
Stanford University
IEEE Communication
Theory Workshop
Maui, HI
May 14, 2012
Future Wireless Networks
Ubiquitous Communication Among People and Devices
Next-generation Cellular
Wireless Internet Access
Wireless Multimedia
Sensor Networks
Smart Homes/Spaces
Automated Highways
Smart Grid
Body-Area Networks
All this and more …
Future Cell Phones
Everything
in one device
Burden for wireless
this performance
is on the backbone network
San Francisco
BS
BS
LTE backbone is the Internet
Internet
Nth-Gen
Cellular
Phone
System
Nth-Gen
Paris
Cellular
BS
Much better performance and reliability than today
- Gbps rates, low latency, 99% coverage indoors and out
Careful what you wish for…
Source: FCC
Source: Unstrung Pyramid Research 2010
Growth in mobile data, massive spectrum deficit and stagnant revenues
require technical and political breakthroughs for ongoing success of cellular
Can we increase cellular system capacity to
compensate for a 300MHz spectrum deficit?
Without increasing cost?
or power consumption?
What would Shannon say?
Are we at the Shannon
limit of the Physical Layer?
We don’t know the Shannon
capacity of most wireless channels
 Time-varying channels with memory/feedback.
 Channels with interference or relays.
 Uplink and downlink channels with
frequency reuse, i.e. cellular systems.
 Channels with delay/energy/$$$ constraints.
Rethinking “Cells” in Cellular
Small
Cell
Coop
MIMO
How should cellular
systems be designed?
Relay
DAS
Will gains in practice be
big or incremental; in
capacity or coverage?
 Traditional cellular design “interference-limited”






MIMO/multiuser detection can remove interference
Cooperating BSs form a MIMO array: what is a cell?
Relays change cell shape and boundaries
Distributed antennas move BS towards cell boundary
Small cells create a cell within a cell
Mobile cooperation via relaying, virtual MIMO, analog network coding.
Are small cells the solution to
increase cellular system capacity?
Yes, with reuse one and adaptive
techniques (Alouini/Goldsmith 1999)
Area Spectral Efficiency
A=.25D2p
 S/I increases with reuse distance (increases link capacity).
 Tradeoff between reuse distance and link spectral efficiency (bps/Hz).
 Area Spectral Efficiency: Ae=SRi/(.25D2p) bps/Hz/Km2.
The Future Cellular Network: Hierarchical
Architecture
Today’s architecture
MACRO: solving
• 3M Macrocells serving 5 billion users
initial coverage • Anticipated
issue, existing
network
PICO: solving
street, enterprise
& home
coverage/capacity
issue
FEMTO: solving
enterprise &
home
Picocell
Macrocell
coverage/capacity
issue
10x Lower COST/Mbps
(more
with WiFi
Offload)
10x CAPACITY
Improvement
Near 100%
COVERAGE
Femtocell
Future systems require Self-Organization (SON) (and WiFi Offload)
SON Premise and Architecture
Mobile Gateway
Or Cloud
Node
Installation
SoN
Server
Self
Healing
Initial
Measurements
IP Network
SON
Self
Configuration
Measurement
Server
Self
Optimization
X2
X2
Small cell BS
Macrocell BS
X2
X2
Algorithmic Challenge: Complexity
 Optimal channel allocation was NP hard in
2nd-generation (voice) IS-54 systems
 Now we have MIMO, multiple frequency
bands, hierarchical networks, …
 But convex optimization has advanced a lot
in the last 20 years
Innovation needed to tame the complexity
Green Cellular Networks
Pico/Femto
Coop
MIMO
Relay
DAS
How should cellular
systems be redesigned
for minimum energy?
Research indicates that
significant savings is possible
 Minimize energy at both the mobile and base station via
 New Infrastuctures: cell size, BS placement, DAS, Picos, relays
 New Protocols: Cell Zooming, Coop MIMO, RRM, Scheduling,
Sleeping, Relaying
 Low-Power (Green) Radios: Radio Architectures, Modulation,
coding, MIMO
Gerhard Fettweis’ talk
Antenna Placement in DAS
 Optimize distributed BS antenna location
 Primal/dual optimization framework
 Convex; standard solutions apply
 For 4+ ports, one moves to the center
 Up to 23 dB power gain in downlink
 Gain higher when CSIT not available
6 Ports
3 Ports
Coding for minimum total power
Is Shannon-capacity still a good metric for system design?
X5
B1
B2
B3
Computational
Nodes
B4
X6
X7
X8
On-chip
interconnects
Extends early work of El Gamal et. al.’84 and Thompson’80
Fundamental area-time-performance tradeoffs
 For encoding/decoding “good” codes,
X5
B1
B2
B3
B4
Area occupied by wires
 Stay away from capacity!
 Close to capacity



Large chip-area
More time
More power
Encoding/decoding clock cycles
X6
X7
X8
Reduced-Dimension
Communication System Design
 Compressed sensing ideas have found widespread
application in signal processing and other areas.
 Basic premise of CS: exploit sparsity to approximate a
high-dimensional system/signal in a few dimensions.
 We ask: how can sparsity be exploited to reduce the
complexity of communication system design
Sparsity: where art thou?
To exploit sparsity, we need to find
communication systems where it exists

Sparse signals: e.g. white-space detection

Sparse samples: e.g. sub-Nyquist sampling

Sparse users: e.g. reduced-dimension multiuser detection

Sparse state space: e.g reduced-dimension network
control
Compressed Sensing
 Basic premise is that signals with some sparse
structure can be sampled below their Nyquist rate
Rob Calderbank’s
Talk
 Signal can be perfectly reconstructed from these
samples by exploiting signal sparsity
 This significantly reduces the burden on the front-end
A/D converter, as well as the DSP and storage
 Enabler for white space, SD and low-energy radios?
 Only for incoming signals “sparse” in time, freq., space, etc.
Software-Defined (SD) Radio:
BT
Cellular
FM/XM
A/D
GPS
DVB-H
Apps
Processor
WLAN
Media
Processor
Wimax
A/D
A/D
DSP
A/D
 Wideband antennas and A/Ds span BW of desired signals
 DSP programmed to process desired signal: no specialized HW
Today, this is not cost, size, or power efficient
Compressed sensing reduces A/D and DSP burden
Sparse Samples
New Channel
Sampling
Mechanism
(rate fs)
 For a given sampling mechanism (i.e. a “new” channel)



What is the optimal input signal?
What is the tradeoff between capacity and sampling rate?
What known sampling methods lead to highest capacity?
 What is the optimal sampling mechanism?

Among all possible (known and unknown) sampling schemes
Capacity under Sub-Nyquist Sampling
 Theorem 1:
 Theorem 2:
 Bank of Modulator+FilterSingle Branch  Filter Bank
t  n(mTs )
q(t)
zzzz
p(t)
zzzz
zz

y1[n]
s1 (t )
zzzzz
s(t )
zzzzz
y[n]
t  n(mTs )
equals
yi [n]
si (t )
t  n(mTs )
sm (t )
ym[n]
 Theorem 3:
 Optimal among all time-preserving nonuniform sampling
techniques of rate fs (ISIT’12; Arxiv)
Joint Optimization of Input and Filter Bank
 Selects the m branches with m highest SNR
 Example (Bank of 2 branches)
low
SNR
H ( f  2kfs )
X  f  2kfs 

H ( f  kfs )
highest
SNR
X  f  kfs 
highest
SNR
Xf 

H ( f  kfs )
low SNR
X  f  kfs 

N ( f  kfs ) S ( f  kfs )

H( f )
2nd
N ( f  2kfs ) S ( f  2kfs )


N( f )
S( f )

Y1  f 
Capacity monotonic in fs
Y2  f 
N ( f  kfs ) S ( f  kfs )

How does this translate to practical modulation and coding schemes
Ideal Multiuser Detection
-
Signal 1
=
A/D
A/D
A/D
Signal 1
Demod
A/D
Iterative
Multiuser
Detection
Signal 2
Signal 2
Demod
-
MUD proposed for LTE
= link at -7dB SNR)
(closes
Why Not Ubiquitous Today? Power and A/D Precision
 Exploits that number of active users G is random and
much smaller than total users (ala compressed sensing)
 Using compressed sensing ideas, can correlate with
M~log(G) waveforms
 Reduced complexity, size, and power consumption
10% Performance Degradation
1
Tb

0
1
Tb

0
Linear
Transformation

Reduced-Dimension MUD
Tb
c1
h1 (t)

r(t) 
g1t 
2g4 
t
n(t)
Tb

b˜2

b˜i 
c2
h2 (t)
Decision
Decision
M

1
Tb

b˜1
a c
ij
 
Tb
cM  
0
hM (t)




j 1
j
b˜N
Decision
Reduced-Dimension Network Design
Random Network State
Reduced-Dimension
State-Space Representation
Projection
Sparse
Sampling
Approximate
Stochastic Control
and Optimization
Utility estimation
Sampling
and
Learning
Feedback in Communications
• Memoryless point-to-point channels:
• Capacity unchanged with perfect feedback
• Feedback drastically increases error exponent (L-fold exponential)
• Feedback reduces energy consumption
• Capacity of channels with feedback largely unknown
•
•
•
•
For channels with memory and perfect feedback
Under finite rate and/or noisy feedback
For multiuser channels
For multihop networks
• ARQ is ubiquitious in practice: Why?
- Output feedback
- CSI
- Acknowledgements
- Something else?
Cognitive Radios
PTx
PRx
CRRx
CRTx
IP
CRTx
MIMO Cognitive Underlay
CRRx
PTx
PRx
Cognitive Overlay
 Cognitive radios support new wireless users in existing
crowded spectrum without degrading licensed users
 Utilize advanced communication and DSP techniques
 Coupled with novel spectrum allocation policies
 Technology could
 Revolutionize the way spectrum is allocated worldwide
 Provide more bandwidth for new applications/services
 Multiple paradigms
 Underlay (exploiting unused spatial dimensions) and Overlay
(exploiting relaying and interference cancellation) promising
The Smart Grid:
Fusion of Sensing, Control, Communications
carbonmetrics.eu
Wireless and Health, Biomedicine and Neuroscience
Body-Area
Networks
Todd
Coleman’s
Talk
Doctor-on-a-chip
-Cell phone info repository
-Monitoring, remote
intervention and services
Cloud
Ubli Mitra’s talk
The brain as a wireless network
- EKG signal reception/modeling
- Information flow (directed MI)
- Signal encoding and decoding
- Nerve network (re)configuration
Summary
 Communications research alive and well
 Communications technology will enable new
applications that will change people’s lives worldwide
 Design innovation will be needed to meet the
requirements of next-generation wireless networks
 A systems view and interdisciplinary design approach
holds the key to these innovations