Wireless Beyond the Third GenerationWireless Beyond the Third Generation——Facing The Energy ChallengeFacing The Energy Challenge

Jan M. RabaeyBWRCUniversity of California @ Berkeleyhttp://www.eecs.berkeley.edu/~jan

ISLPED 2001, Huntington Beach

It’s all about Laws …It’s all about Laws …

Mo o re ’ s

M o o re ’ s

L awLaw

Sha n n on ’ s

L aw

Sh a n n on ’ s

L aw

Me t c al f e

’ s L aw

Me t c al f e

’ s L aw

Ge n e ’ s L aw

Ge n e ’ s L aw

Maxwe l l ’ s

Ma xwe l l ’ s

L a w sL a w s

L aw s o f P h y s i c s

L aw s o f P h y s i c s

The The FibonacciFibonacci Law on Wireless GrowthLaw on Wireless Growth

Source: Goldman-Sachs

“The number of world-wide wireless subscribers (in tens of millions)grows as a Fibonacci series”

1999 2000 2001 2002 2003

46

57

67

?

The Shift to Wireless DataThe Shift to Wireless Data——The New InternetThe New Internet

Clusters

Massive Cluster

Gigabit Ethernet

Pre-1990:Client-Server Systems

The 2000s:Extending toward the SmallEnabled by integration(Moore’s Law at Work!)and wireless connectivity

The 1990s:Conquering the worldThe network revolution

Courtesy: R. Katz, UCB

(Projected) Growth in 802.11 WLAN(Projected) Growth in 802.11 WLAN

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

2001 2002 2003 2004 2005

1-2Mbps

10-11Mbps

20Mbps+

54 MbpsTotal

Units, k

Source: Cahners In-Stat 2001

The Evolving Wireless SceneThe Evolving Wireless Scene

Range

Dat

a R

ate

1m 10m 100m 1km 10km

1Kb

10Kb

100Kb

1Mb

10Mb

100Mb

Cellular (WAN)

3G Cellular

2.5 G Cellular

802.11 (LAN)

802.1a

Bluetooth (PAN)

Sensor networks

Metropolitan

More bit/secMore bit/secMore bit/($·More bit/($·nJnJ))

Range

Dat

a R

ate

1m 10m 100m 1km 10km

1Kb

10Kb

100Kb

1Mb

10Mb

100Mb

Cellular (WAN)

3G Cellular

2.5 G Cellular

802.11 (LAN)

802.1a

Bluetooth (PAN)

Sensor networks

Metropolitan

More bit/secMore bit/secMore bit/($·More bit/($·nJnJ))

How to Get More bits/sec in aHow to Get More bits/sec in aBandBand--Limited Environment? Limited Environment? The Shannon Bound:In an AWGN channel, the best bandwidth-efficiency (in bits/sec/Hz)that can be achieved with arbitrarily low bit-error rate is given by

bweff ≤≤ log2 (1 + bweff * Eb/N0)

0 5 10 15 20 25 30 35 40 45 500

1

2

3

4

5

6

7

8

9

Energy/bit over Noise Spectral Density Eb/N0

Ban

dwid

th e

ffici

ency

(bi

ts/s

ec/H

z)

Bandwidth EfficientBandwidth Efficient

Energy EfficientEnergy Efficient

1

10

100

1000

10000

100000

0 2 4 6 8 10 12

SNR (db)

Rel

ativ

e C

om

ple

xity

The Cost of Approaching Shannon’s BoundThe Cost of Approaching Shannon’s Bound

Courtesy Engling Yeo, UCB

The Bliss and Challenge of Error CodingThe Bliss and Challenge of Error Coding

8/9 Turbo, νν=4, N=4k

2/3 Turbo, νν=4, N=64k 1,2, and 3 iterations

1/2 Turbo, νν=4, N=64k 1,2, and 3 iterations

8/9 LDPC, N=4k 1,3, and 5 iterations

8/9 Conv. Code, νν=3, N=4k

1/2 Conv. Code, νν=4, N=64k

2/3 Conv. Code, νν=4, N=64k

8/9 Capacity Bound2/3 Capacity Bound

1/2 Capacity Bound

1/2 LDPC, N=107, 1100 iterationsfor BER of 10-5

Dealing with NonDealing with Non--ideal Channels (e.g., fading)ideal Channels (e.g., fading)

• Multi-antenna approach exploits multi-path fading by sending data along good channels

• Results in large theoretical improvements in bandwidth efficiency for fading channels

• But…computationally hungry

)(ty

ArrayProcessing

)(tx

ArrayProcessing

1st path, α1 = 1

2nd path, α2 = 0.6

SNR (dB)0 5 10 15 20 250

5

10

15

20

25

30

Cap

acity

(bi

ts/s

/Hz)

(4,4) With Feedback(4,4) No Feedback(4,1) Orthogonal Design(1,1) Baseline

The Cost of Dealing with NonThe Cost of Dealing with Non--ideal Channelsideal Channels

0

1000

2000

3000

4000

5000

6000

7000

Performance

MIPS

0.8 Mb/s0.9 b/s/Hz

1.6 Mb/s1.8 b/s/Hz

1.9 Mb/s2.1 b/s/Hz

3.8 Mb/s4.2 b/s/Hz

5.6 Mb/s6.3 b/s/Hz

Data rate per userSpectral efficiency

* Assume 25 MHz bandwidth and 28 users

MatchedFilter

Multi-UserDetector

OFDM +Coding

OFDM +Multi-Antenna

OFDM + Multi-Antenna + Coding

Source: Ning Zhang, UCB

Compelling Wireless Implementation IssuesCompelling Wireless Implementation Issues

•• A Ferocious Quest for PerformanceA Ferocious Quest for Performance–– Driven by the hunger for bits/secDriven by the hunger for bits/sec

–– Outstripping the technology evolutionOutstripping the technology evolution

•• With a Premium on Reduction in Energy With a Premium on Reduction in Energy ConsumptionConsumption–– The compelling argument behind wireless is its The compelling argument behind wireless is its

untethered untethered naturenature

–– Power consumption key impediment to penetration of Power consumption key impediment to penetration of new servicesnew services

–– Energy sources on slow evolutionary path (5%/year)Energy sources on slow evolutionary path (5%/year)

What Technology Offers UsWhat Technology Offers UsGene (Frantz)’s LawGene (Frantz)’s Law

Gene’s Law

DSP Power

1,000

100

10

1

0.1

0.01

0.001

0.0001

0.00001

mW

/MIP

S

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

YearSource: Gene Frantz (TI)

Energy Efficiency of DSPs Doubles Every

18 Months

Shannon beats Moore beats ChemistsShannon beats Moore beats Chemists

1

10

100

1000

10000

100000

1000000

10000000

1980

1984

1988

1992

1996

2000

2004

2008

2012

2016

2020

Algorithmic Complexity(Shannon’s Law)

Processor Performance (~Moore’s Law)

Battery Capacity1G

2G

3G

Courtesy: Ravi Subramanian (Morphics)

The Need for FlexibilityThe Need for Flexibility• Cost-of-design issues point towards component

reuse– Design complexity impacts time-to-market– Physical effects increase verification costs and design risk

– NRE of new designs is increasing significantly (mask making, fab cost)

• Multi-standard has become a must in the diverse wireless landscape

• Adaptive solutions lead to better spectral utilization• A wide variety of unpredictable services

An Attractive Option:An Attractive Option:MultiMulti--Processor SystemProcessor System--onon--aa--ChipChip

Courtesy: Chris Rowen, Tensilica

Copyright Tensilica, Inc 2001

“The New NAND gate”

Flexibility Comes at a Huge CostFlexibility Comes at a Huge Cost

µP

Prog Mem

MACUnit

AddrGenµP

Prog Mem

µP

Prog Mem

SatelliteProcessor

DedicatedLogic

Satellite

Processor

SatelliteProcessor

GeneralPurpose

µP

Software

DirectMapped

Hardware

HardwareReconfigurable

Processor

ProgrammableDSP

Fle

xibi

lity

Fle

xibi

lity

InefficiencyInefficiency

3 orders of magnitude!3 orders of magnitude!

500 pJ/op

0.5 pJ/op

The OpportunityThe Opportunityof Configurable Architecturesof Configurable Architectures

Power (mW)

0

50

100

150

200

250

DSP

(T

IC

54)

DSP

Ext

ensi

on

Con

figu

rabl

e

Ded

icat

edH

ardw

are

Execution

Memory

Control

Clocking

Area (mm^2)

DSP

(T

IC

54)

DSP

Ext

ensi

on

Con

figu

rabl

e

Ded

icat

edH

ardw

are

* Based on the implementations of a multi-user detector

Source: N. Zhang, UCB

The Opportunity of ReconfigurableThe Opportunity of Reconfigurable——When Does it Work? When Does it Work?

* All results are scaled to 0.18µm

101

102

103

103

104

105

106

107

108

FFT sizeT

rans

form

s/se

c/m

m2

2 ) Function-specific reconfigurable hardwareData-path reconfigurable processor FPGA Low-power DSP High-performance DSP

101

102

103

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

FFT size

J/T

rans

form

Lower limit Function-specific reconfigurable hardwareData-path reconfigurable processor FPGA Low-power DSP High-performance DSP

Source: N. Zhang, UCB

Energy and Area Efficiencyof Various FFT Implementations

ReconfigurableDataPath

ReconfigurableState Machines

Embedded uP+ DSPs

FPGA

DedicatedDSP

The Ideal “RadioThe Ideal “Radio--onon--aa--Chip” PlatformChip” Platform

• Heterogeneous• Matches the computational

model• Provides flexibility only where

needed and desirable and at the right granularity

• Supports massive concurrency

• Operates at minimum supply voltage and clock frequency

Combines performance, flexibility and energy-efficiency

An Orthogonal Approach to Bit/secAn Orthogonal Approach to Bit/sec——UltraUltra--Wide Band RadioWide Band Radio

Traditional Sinusoidal, Narrowband

Frequency

Time

Time

Frequency

Impulse, Ultra-Wideband

Splurge on Bandwidth (> 5 GHz) and Punt on Bit/sec/HzPossible advantages: easy co-existence, low-power, simple

Digital PulseDigital Pulse--Based RadioBased RadioSimple Digital Architecture:

• Transmit Only Narrow Pulses (No Carrier Frequency)• Spread Energy Over Existing Noise Floor

A/DV

-+ V

-+

The Architectural Challenge:Providing accurate timing resolution without high-frequency clocks!

Predicted performance: 100 kbits/sec @ < 10-4 bit/sec/Hz and ~ 10 nJ/bit

Range

Dat

a R

ate

1m 10m 100m 1km 10km

1Kb

10Kb

100Kb

1Mb

10Mb

100Mb

Cellular (WAN)

3G Cellular

2.5 G Cellular

802.11 (LAN)

802.1a

Bluetooth (PAN)

Sensor networks

Metropolitan

More bit/secMore bit/secMore bit/($·More bit/($·nJnJ))

More Bits/(More Bits/(nJnJ·$·mm·$·mm33):):Wireless Sensor NetworksWireless Sensor NetworksPushing the Bounds in Pushing the Bounds in Ultra [Small, Cheap, LowUltra [Small, Cheap, Low--Power]Power]

The Smart BuildingThe Smart BuildingIntegrated Sensor/Actuator/Control System• Improves quality-of-living• Saves energy• Provides security• Helps localizing items• Extends building-human interface

Berkeley Berkeley PicoRadio’sPicoRadio’sMeso-scale low-cost radio’s for ubiquitous wireless data acquisition in sensor/actuator networks that are fully integratedand consume less than 100 µµW enabling energy scavenging

The EnergyThe Energy--Scavenging OpportunityScavenging Opportunity

0

1

10

100

1000

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5Years

µµW

atts

Lithium

Zinc-Air

NiMH

Li (rechargeable)

Solar

Vibrations

Alkaline

Battery size: 0.5 cm3

Vibration: 1 cm2 piezo-electricSolar: 1 cm2 single-crystal

Courtesy: S. Roundy (UCB)

Opportunity: Metcalfe’s LawOpportunity: Metcalfe’s Law

“The true value of a network increases as the square of the number of users on the network”

A Variant: Jan’s LawA Variant: Jan’s Law“The power efficiency of a wireless sensor node increases as the square of the number of nodes in the network (or is proportional to the node density).”

Addressing the Communication CostAddressing the Communication Cost

-100

-80

-60

-40

-20

0

20

1 10 100

Distance (in m)

Tran

smit

Pow

er (i

n db

M) f c

= 2.4 GHz

f c= 0.43 MHz

Bluetooth

(assuming d4 path loss of and 10 kB/sec data rate)

• Minimum required transmit power increases as d4 due to ground wave and multi-path

• Increasing carrier frequency costs 20db/decade

16 pJ/bit

Adding a Single Relay Point in a Adding a Single Relay Point in a Wireless LANWireless LAN

Source: M. Kubisch and H. Karl, TU Berlin

Bit-rate: 6 Mb/secPacket Error Rate: 1%Fixed Receiver Power: 100 mWPath Loss Exp.: 3.8d1/d2 =1

Reduces energy/bitup to 4 times!

node range

TradingTrading--off Latency for Energy:off Latency for Energy:SelfSelf--configuring Multiconfiguring Multi--hop Networkshop Networks

sensoractuator

controller

• 1 hop over 50 m1.25 nJ/bit

• 5 hops of 10 m each5 × 2 pJ/bit = 10 pJ/bit

• Multi-hop reduces transmission energy by 125!

But …how to ensure fairness?But …how to ensure fairness?

EnergyEnergy--Conscious NetworkingConscious NetworkingSimulated Energy DissipationSimulated Energy Dissipationin Sensor Networks (BWRC)in Sensor Networks (BWRC)

Performance-oriented

Energy-Conscious

Po

wer

Co

nsu

mp

tio

n

Source: R. Shah (UCB)

The SensorThe Sensor--Node RF ChallengeNode RF Challenge

• Increasing carrier frequency increases power dissipation– Mostly due to higher speed

active components (synthesizers, mixers, A/Ds)

• But enables higher integration– Smaller sizes of passives

ands antennas0.1

1

100.1 1 10 100 1000

Receiver Power Consumption (mW)

Car

rier

Fre

quen

cy

Rx power consumption versus carrier frequencyfor a number of low-data rate, small-distance RF implementations(all operate in Shannon’s “energy-efficient zone”)

Eliminating most HighEliminating most High--Speed ComponentsSpeed ComponentsSubSub--sampling receiver with passive sampling receiver with passive frontendfrontend

RF Filter LNA RF Filter AD

fbase

Image Rejection

Provides diversityShapes LNA thermal noise

RF Filter

RF Filter

RF Filter

• Does not requireany high-frequencyactive components

• Receive energy reduced by 1-2 ordersof magnitude to ~ 5 nJ/bit

Enabled by HighEnabled by High--Q Integrated FiltersQ Integrated Filters

-

RF-MEMS: Poly Si-Ge Resonator

Thin-Film Bulk Acoustic Resonators Q > 1000 @ 2 GHz(FBAR – Agilent)

The Importance of Power Management The Importance of Power Management •• Activity in sensor (data) networks is low and random (< 1%)Activity in sensor (data) networks is low and random (< 1%)• Receiving a bit is computationally more expensive than

transmitting one • Most Media Access protocols assume that the receiver is

always on and listening!

Why not power transceiver up for real events only Why not power transceiver up for real events only (incoming data, sensor event, network maintenance)?(incoming data, sensor event, network maintenance)?

Reactive MediaReactive Media--Access ControlAccess Control

Truly Reactive Messaging at the Physical and Media-Access Levelv Power Down the Whole Data Radiov Reduces Monitoring Energy Consumption by 103 Timesv Wa keup Radi o Power’ s Up Dat a Radi o f or Dat a Recept i on

Sleeping nodes

Communicating nodes

The WakeThe Wake--up (Reactive) Radioup (Reactive) Radio

PAPulse

Generator

RF Filter

RF Filter

RF Filter

Transmit

RF Filter LNA RF Filter AD

RF Filter

RF Filter

RF Filter Receive

EnergyDetector

PulsePositionDetector

Wake-up

• Always running– Super low power:

10-4~ 10-3 active mode power

• Data radio shut down when idle, and powered up by wake-up radio

• Receiver response time: < 10ms

Receive SessionWakeup

Tone

What it Ultimately Boils Down ToWhat it Ultimately Boils Down To——PowerPower--Profile of PicoRadio Profile of PicoRadio (Projected)(Projected)

Clock distribution

14%

Reconfigurable protocol stack

25%

Reactive receiver 1%

Radio receiver12%

Radio transmitter

5%

Baseband13%

Embedded microprocessor

30%

Clock distribution

14%

Reconfigurable protocol stack

25%

Reactive receiver 1%

Radio receiver12%

Radio transmitter

5%

Baseband13%

Embedded microprocessor

30%

Summary/PerspectiveSummary/Perspective• Both the bits/sec/Hz and the bits/nJ quests create

formidable energy challenges• Keep your eyes open for innovative, orthogonal

approaches that re-stack the cards– There is a whole lot of unexplored land available > 10 GHz

• In the end, it are the laws of physics that provide the ultimate bounds