From Cognitive Radios over From Cognitive Radios over Millimeter Wave to Millimeter Wave to PicoRadiosPicoRadiosJan M. RabaeyScientific Co-Director BWRCEECS Dept.Univ. of California, Berkeley

Berkeley Wireless Research CenterBerkeley Wireless Research Center• A partnership of UC researchers, industry,

and government • Participating Members:

Agilent TechnologiesInfineon TechnologiesIntel CorporationSTMicroelectronicsHitachi LtdHewlett PackardSun Microelectronics

• Associate Members:Atmel CorporationCadenceEricsson Radio Systems SamsungNECQualcomm Incorporated

• Other Funding: DARPA, NSF, ONR, MARCO, MURI

• Operational since Feb.1999• Downtown Berkeley, 1 block from

campus• 11,000 sq. feet• 55 graduate students,

5 technical staff, 8 faculty

Overall budget: > 5 M$/year

The Center GoalsThe Center Goals• Pre-competitive research – beyond 5 years• In close collaboration with industrial partners• Focus on silicon (CMOS) implementation

– Explore novel and disruptive paradigms – Determine relationship between theoretical and

algorithmic advances and implementation– Understand tradeoffs between various implementation

architectures with respect to performance, power and cost

– Realize concept-proving prototypes using rapid design flow from algorithm to implementation

The BWRC Research AgendaThe BWRC Research AgendaDa

ta Ra

te

1Kb

10Kb

100Kb

1Mb

Cellular (WAN)

3G Cellular

2.5 G Cellular

802.11 (LAN)

802.1a

Bluetooth (PAN)

Sensor networks

Metropolitan

Zigbee (PAN)

More bit/secMore bit/secThe quest for spectral capacity• Improving spectral utilization: MIMO• Exploring new spectrum: 60 GHz• Overlaying spectrum: UWB• Re-cycling spectrum: Cognitive

The quest for spectral capacityThe quest for spectral capacity• Improving spectral utilization: MIMO• Exploring new spectrum: 60 GHz• Overlaying spectrum: UWB• Re-cycling spectrum: Cognitive

Ubiquitous embeddedwireless• Ultra-low cost• Ultra-low power • Small size

Ubiquitous embeddedUbiquitous embeddedwirelesswireless• Ultra-low cost• Ultra-low power • Small size

Cheaper bitsCheaper bits100Mb

10Mb

1m 10m 100m 1km 10km

Range

The Wireless ArenaThe Wireless Arena

Designing Radios to Improve Spectrum UtilizationDesigning Radios to Improve Spectrum UtilizationThe target: Ubiquitous wireless multimedia delivery

• Exploit spatial dimension• Improve utilization of spectrum

– UWB – use low transmit power– Cognitive – dynamically find spectra that

isn’t being used• Exploit new frequencies – 60 GHz

Exploiting Spatial Diversity Exploiting Spatial Diversity –– Multiple Antenna ArraysMultiple Antenna ArraysExample: 802.11n

Network Improve range

MAC

A B C A B C D

Minimize interference

TX RX

PHY

Increase data rateImprove reliabilityReduce multipath

But it’s costly …But it’s costly …H11

H41

Σ

H14

H44

Σ

TX RX

Demands intensive real-timematrix computations.

SVD22.6mm

19.3mm

E.g. a division-free, deflation-type and LMS-based SVD algorithm

Direct-mapped area estimates: > 200 mm2

Based on 1 V, 1 MHz, 16 bits and 0.25 µm CMOS technology.

How Many Antenna’s Do We Really Need?How Many Antenna’s Do We Really Need?A testbed for wireless experimentation• FPGA prototyping engines (4)• 2.4 GHz RF Front-ends (32)• Scalable multiple antenna transmission system

Combining Theory and ExperimentsCombining Theory and ExperimentsTotal Capacity for Time/Freq/Space [Poon03]:2 A Ω W T λ-2 log SNR

scatteringantenna area

Scattering angle measurements

Designing Radios to Improve Spectrum UtilizationDesigning Radios to Improve Spectrum Utilization

• Exploit spatial dimension• Improve utilization of spectrum

– UWB – use low transmit power– Cognitive – dynamically find spectra that isn’t

being used• Exploit new frequencies – 60 GHz

Window of OpportunityWindow of Opportunity

Time (min)

Freq

uenc

y (H

z)

Recent measurements by the FCC in the US show 70% of the allocated spectrum is not utilized

Time scale of the spectrum occupancy varies from msecs to hours

Existing spectrum policy forces spectrum to behave like a fragmented disk

Bandwidth is expensive and good frequencies are taken

Unlicensed bands – biggest innovations in spectrum efficiency

What is a Cognitive Radio?What is a Cognitive Radio?

Easement User

2nd-aryUser

2nd-aryUser

Licensee

Not-to-Interfere Basis

Below the Acceptable “Interference Temperature”

FCC diagram

Increase User spectrum efficiencies

• Cognitive radio requirements– co-exists with legacy wireless systems– uses their spectrum resources – does not interfere with them

• Cognitive radio properties– RF technology that "listens" to huge swaths of spectrum – Knowledge of primary users’ spectrum usage as a function of location and time– Rules of sharing the available resources (time, frequency, space)– Embedded intelligence to determine optimal transmission (bandwidth, latency, QoS)

based on primary users’ behavior

Cognitive RadiosCognitive RadiosC

onfig

urab

le a

rray RF

RF

RF

Sensor

Optimizer

ReconfigurableBaseband D

imen

sion

2

Feasibleregion

Constraints

• Sensor finds the feasible region• Optimizer selects the best waveform …

– when, how long, frequency, bandwidth, array configuration.• Reconfigurable baseband adapts to the optimal schemes.• Major challenges:

– Broadband RF– Intelligent flexible radios– Massive processing

Dimension 1

Designing Radios to Improve Spectrum UtilizationDesigning Radios to Improve Spectrum Utilization

• Exploit spatial dimension• Avoid interfering with primary users

– UWB – use low transmit power– Cognitive – dynamically find spectra that

isn’t being used• Exploit new frequencies – 60 GHz

6060--GHz Unlicensed AllocationGHz Unlicensed AllocationBandwidth and

channel properties suitable for ~1-Gbps

wireless link

If there is so much opportunity, why is there so little use of this band?

Power handling, linearity, and noise performance of circuits at 60 GHzTraditional mm-wave solutions are very expensive (InP, GaAs).System design challenges, baseband analog interface ……

CMOS is Capable of 60 GHz OperationCMOS is Capable of 60 GHz Operation

VGS = 0.65 V

VDS = 1.2 V

IDS = 30 mA

W/L = 80u/0.13u

C. H. Doan, S. Emami, A. M. Niknejad, and R. W. Brodersen, “Design of CMOS for 60GHz Applications,” in IEEE Int. Solid-State Circuits Conf. Dig. Tech. Papers, Feb. 2004.

mmmm--Wave Challenges: ModelingWave Challenges: Modeling

Measured and modeled IDS vs. VDS.

Conclusion:lumped transistor modelswork well into the 60 GHz range!

RF-LO coupler

Transmission-lineterminations

mmmm--Wave Challenges: DesignWave Challenges: Design

Dual-gate mixer

Single-gate mixer

60 GHz Wireless LAN System 60 GHz Wireless LAN System

10-100 m

• Objective: Enable a fully-integrated low-cost Gb/s data communication using 60 GHz band.

• Approach: Employ emerging, standard CMOS and SiGe technology for the radio building blocks. Exploit antenna array for improved gain and resilience.

Ubiquitous Embedded WirelessUbiquitous Embedded Wireless

Meso-scale low-cost wireless transceivers for ubiquitous wireless data acquisition that• are fully integrated

–Size smaller than 1 cm3

• minimize power/energy dissipation– Limiting power dissipation to 100 µW

enables energy scavenging

• support low data-rates (< 100 kBit/sec)• and form self-configuring, robust, ad-hoc networks containing 100’s to 1000’s of nodes

Meso-scale low-cost wireless transceivers for ubiquitous wireless data acquisition that• are fully integrated

–Size smaller than 1 cm3

• minimize power/energy dissipation– Limiting power dissipation to 100 µW

enables energy scavenging

• support low data-rates (< 100 kBit/sec)• and form self-configuring, robust, ad-hoc networks containing 100’s to 1000’s of nodes

Berkeley PicoRadio ProjectBerkeley PicoRadio Project

“Ambient Intelligence” (The Concept)“Ambient Intelligence” (The Concept)• An environment where technology is embedded, hidden in the

background • An environment that is sensitive, adaptive, and responsive to

the presence of people and object• An environment that augments activities through smart non-

explicit assistance• An environment that preserves security, privacy and

trustworthiness while utilizing information when needed and appropriate

Fred Boekhorst, Philips, ISSCC02

The road to lowThe road to low--energy, lowenergy, low--cost, smallcost, small--sizesizeNot wireless as usual!Not wireless as usual!• Simplicity rules!

– Advanced techniques used in traditional wireless links are not necessarily relevant

– Looking back at the techniques of old• Standby power the greatest enemy

– Monitoring connectivity dominates overall power– Leakage dominates digital power

• Redundancy and randomness as a means to create robustness – Elements and links can and will fail– The environment and its conditions change rapidly

LowLow--Power RF: Back to The FuturePower RF: Back to The Future(Courtesy of Brian Otis)(Courtesy of Brian Otis)

© 1949 - superregenerativefc= 500MHz2 active deviceshigh quality off-chip passives - hand tuning D. Yee, UCB

© 2000 - Direct Conversionfc= 2GHz>10000 active devicesno off-chip components

RFRF--MEMS: The OpportunityMEMS: The Opportunity

Thin-Film Bulk Acoustic Resonator (FBAR [Ruby,ISSCC01])Q > 1000 @ 2 GHz

RadialBulk Acoustic Resonator (RBAR)Q > 500; f > 1 GHz

CMOS Metallization Stack

ri rog

Sense Electrode

Drive Electrode

Passive micro-resonators• High Q (> 1000)• High Frequency (> 1 GHz)• Very Small• Potential for integration

OSC1

OSC2

Back to the FutureBack to the Future

MatchingNetwork

MOD1

MOD2

Preamp PA

RF Filter A

fclockRF Filter Env

Det ∫

fclockRF Filter Env

Det ∫

Receiver

RF Amp Test

LNATest

Diff Osc

PA Test

TX1

TX2Env DetTest

Passive Test Structures

4 mm

• Minimizes use of active components –exploits new technologies such as RF-MEMS

• Uses the simple modulation scheme (OOK)• Allows efficient non-linear PA• Down-conversion through non-linearity (Diode)• Tx and Rx in 1-2 mW range (when on)

FBARFBAR--basedbased

Thin-Film Bulk Acoustic Resonator

The Return of The Return of SuperregenerativeSuperregenerative1500µm

1200

µm

• Fully Integrated Receiver Front-end

• 400µA when active(~200µW) with 50% quench duty cycle

RF In

Osc Out

BB Out

(Currently in fab -prototype expected early January)

A

fclockRF Filter Env

Det ∫

fclockRF Filter Env

Det ∫

Dealing with Standby Power:Dealing with Standby Power:The Reactive or WakeThe Reactive or Wake--Up RadioUp Radio

Information Burst

-

LCARF Filter

RF FilterDetector

Wake-up Radio:Low Gain, Low Sensitivity, Low BER

Wake-up Signal

RF Filter

Shifts Burden to TransmitterReduces monitoring power to < 50 µW

Realizing subRealizing sub--100 100 µµW carrierW carrier--sensesensePrototyping building blocks

-100dBc/Hz@1MHz offset

Phase noise

150mV(Vdd=500mV)

Differential output swing

1.5GHzOscillation frequency

150µACurrent consumption

0.5 – 1.2VSupply voltage

Another example: 140 µW 26 MHz wake-up receiver(R. Banna, N. Weste, UNSW)

Example: sub-threshold RF oscillatorusing integrated LCs (in fab)

Simulated Performance

Wireless Sensor Network Protocol ProcessorWireless Sensor Network Protocol Processor

In fab (Jan 04)

µWsStandby Power

< 1 mWOn_Power

3mm x 2.75mm =8.2 mm2

Chip Size

0.13µ CMOSTechnology

1V(High) –0.3V(Low)Core Supply Voltages

68KbytesOn Chip memory

16MHz(Main), 1MHz(BB)

Clocks Freqs

62.5K gatesGate Count

3.2MTransistor Count

64Kmemory DW8051

µc

BaseBand

SerialInterface

GPIOInterface

LocationingEngine

Neighbor List

SystemSupervisor

DLL

NetworkQueues

VoltageConv

Integrates all digital protocol and applications functions ofwireless sensor nodeMain Features:

- As simple as can be- Aggressive power management

EnergyEnergy--Scavenging becoming a RealityScavenging becoming a Reality• A self contained 1.9GHz transmitter - powered by Solar & Vibrational energy only

Light Level Duty CycleLow Indoor Light 0.36%

Fluorescent Indoor Light 0.53%Partly Cloudy Outdoor Light 5.6%

Bright Indoor Lamp 11%High Light Conditions 100%

Vibration Level Duty Cycle2.2m/s2 1.6%5.7m/s2 2.6%

Front

FBAR

TXBack

10 µFcap

Extrapolating towards the futureExtrapolating towards the future• How far can we push cost, size, and power? Can

we make “real” smart dust?• Absolutely! By going completely non-traditional!

– Ultra-dense networks: get the nodes closer, use lots of them, and make their energy consumption absolutely minimal (this is < 10 µW).

– Use non-tuned mostly passive radio’s – center carrier frequency randomly distributed

– Use statistical distribution to ensure reliable data propagation

• Leads to “smart surfaces”: sensitive plane wings, adaptive fabrics, intelligent walls

The Roadmap to UltraThe Roadmap to Ultra--dense Networksdense Networks

NEMS?

Super-regenerative:< 500 µW

Resonant bodyFinfet

Untuned mostlypassive< 5 µW

Untuned Subthreshold< 50 µW

Powering UltraPowering Ultra--Dense NetworksDense Networks

Anchor Spring flexure Comb fingers

Energy generation and conversion network

Energy Source 1

Energy Source 2

ConversionNetwork 1

ConversionNetwork 2

Reservoir 1(capacitor)

Reservoir 2(microbattery)

Needs integrated meso-scale energy train

Micro-battery

Electrostatic MEMSvibration converters

Summary and PerspectivesSummary and Perspectives• Wireless a key enabler to realization of truly

ubiquitous and disappearing electronics• Opening the door for amazing new

applications– Sensor networks, ambient intelligence, home

gateways, …• Making these come true requires new

paradigms, novel architectures, aggressive and alternative technologies, and tight integration of all components

• The change is truly “in the air”