is wireless (data) dead?bnrg.cs.berkeley.edu/~randy/talks/cc97/cc97.pdfis wireless (data) dead?...

1

Is Wireless (Data) Dead?Is Wireless (Data) Dead?

Randy H. KatzRandy H. KatzComputer Science Division, EECS DepartmentComputer Science Division, EECS Department

University of California, Berkeley, CA 94720-1776University of California, Berkeley, CA 94720-1776© 1997© 1997

2

Heilmeier’s VisionHeilmeier’s Vision

•• AccessAccess–– Internet connectivityInternet connectivity

•• Anytime, AnywhereAnytime, Anywhere–– Wide-Area CoverageWide-Area Coverage–– Scalable ProcessingScalable Processing–– Highly Available OperationHighly Available Operation

•• EasilyEasily–– Transparent AccessTransparent Access–– Localized ServiceLocalized Service

•• SecurelySecurely–– Global AuthenticationGlobal Authentication

•• Any MediumAny Medium–– Multimedia: AV/GraphicsMultimedia: AV/Graphics

•• TimelyTimely–– PerformancePerformance

•• Cost EffectiveCost Effective–– Heterogeneous Support via Heterogeneous Support via

ProxiesProxies

“People and their machines should be able to access information and communicate with each other easily and securely, in any medium or combination of media --

voice, data, image, video, or multimedia -- any time, anywhere, in a timely,cost-effective way.” G. Heilmeier, 1992

3

Marconi’s VisionMarconi’s Vision

“It is dangerous to put “It is dangerous to put limits on wireless.”limits on wireless.”

Guglielmo Marconi (1932)Guglielmo Marconi (1932)

4

Presentation OutlinePresentation Outline

•• Wireless Definitions and TrendsWireless Definitions and Trends

•• Wireless TelephonyWireless Telephony

•• Quo Vadis Wireless Data?Quo Vadis Wireless Data?

•• Access is the Killer App!Access is the Killer App!

•• Summary and ConclusionsSummary and Conclusions

5

The Wireless UniverseThe Wireless UniverseWireless Communications

Amateur Industrial Consumer Business Military/Aero Long-Haul

Automotive Monitoring

— IVHS— GPS

— AMR— Control

ResidentialCordless

Cellular Paging WPABX PMR/SMR Mobile Data

Analog Digital— CT-0— CT-1— CT-300

— DECT— CT-2— PHP— USCT— ISM

WLAN

Analog Digital— AMPS— ETACS— NMT450— NMT900— NMT-O— Comvik— JTACS

— GSM— IS-54— IS-95— RCR-27

— ARDIS— Mobitex— Omnitracs— Cellular/CDPD

— DECT— CT-2— PHP— USCT— ISM

— 802.11— DECT— HiPerLAN— ISM

— POSCAG— ERMES— SSB

PCN/PCS

— DCS1800— PHP— US??— LEO

— FPLMTS— UMTS— RACE— Others

Conv

ESMR— MIRS— TETRA

6

Radio BasicsRadio BasicsWavelength (m)

104 102 100 10-2 10-4 10-6 10-8 10-10 10-12 10-14 10-16

104 106 108 1010 1012 1014 1016 1018 1020 1022 1024

Frequency (Hz)

RadioSpectrum

IR X-Ray CosmicRays

UV

Visible LightR O Y G B I V

< 30 KHz30 - 300 KHz300 KHz - 3 MHz3 - 30 MHz30 - 300 MHz300 MHz - 3 GHz3 - 30 GHz> 30 GHz

VLFLFMFHFVHFUHFSHFEHF

Speed of Light = λ * ƒ = 3 x 108

AM RadioFM Radio, TV

Cell PhoneSatellite

1 MHz100 MHz

1 GHz10 GHz

300 m3 m

30 cm3 cm

≈1000 ft≈10 ft≈1 ft≈1 in

7

Absorption

Ionosphere

Reflected

HF Transmission

Line of Sight

Directional Antenna

VHF Transmission

Reflected waveinterferes with signal

Radio BasicsRadio Basics

8

Wireless SpectrumWireless Spectrum

CT-0(US)

CT-1(Japan)

LMR(US)

SMR(US)

NMT-450

CT-2(Eur, SEA)

ISM(US)

CT-1/CT-1+(Eur)

AMPS GSM

ETACSNMT-900RC2000

...

GPSDECT(Eur)

PHP (RCR-28)(Japan) ISM

PDC(RCR-27)

DCS-1800(Europe)

PCS(US)

0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

PCS BandsPCS Bands

9







10

Wireless Telephony not Dead!Wireless Telephony not Dead!

•• Exploding Cellular Voice ServicesExploding Cellular Voice Services–– Approximately 40 million subscribers in USApproximately 40 million subscribers in US–– Roughly half of the world’s cellular usersRoughly half of the world’s cellular users–– Similar number of paging subscribers in USSimilar number of paging subscribers in US

–– $7 billion in primary PCS license auctions$7 billion in primary PCS license auctions–– $10.2 billion in secondary PCS auctions$10.2 billion in secondary PCS auctions

11

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

0102030405060708090

100

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

PCSCellular+PCS

Proliferation of Wireless TelephonyProliferation of Wireless Telephony

Millions ofSubscribers

You are here!

Yankee GroupForecast

28000 new AMPS28000 new AMPSsubscribers per daysubscribers per day

12

Cellular Phone GrowthCellular Phone Growth

1990 1995 20000

5

10

15

20

25

30

35

40

45

1990 1995 2000

EuropeUnites StatesJapan

% of main linesthat are mobilephones

Year 2000: Mobile phones will reach 42% of fixed-line subscribers

Economist4 May 1996

13

Japan UnpluggedJapan Unplugged

86 87 88 89 90 91 92 93 94 95 96 Oct0

2

4

6

8

10

12

14

16

86 87 88 89 90 91 92 93 94 95 96 Oct

# SubscribPHS

Number of mobile phone subscribers in Japan, in March ofNumber of mobile phone subscribers in Japan, in March ofeach year — N.Y. Times, 20 January 1997each year — N.Y. Times, 20 January 1997

14

The Lessons of the Japanese The Lessons of the Japanese Cellular Phone SystemCellular Phone System

•• Unexpectedly Rapid GrowthUnexpectedly Rapid Growth–– 1995: ten years to 5.4 million subscribers1995: ten years to 5.4 million subscribers–– March 1996: 10 millionMarch 1996: 10 million–– October 1996: 21 millionOctober 1996: 21 million

•• Why?Why?–– Rapidly declining prices for cell phones & servicesRapidly declining prices for cell phones & services–– Deregulation of Japanese cellular telephone marketDeregulation of Japanese cellular telephone market–– Recent introduction of personal communication Recent introduction of personal communication

services: portable handiphone systemservices: portable handiphone system

15

PCS = “Pretty Crazy Situation”?PCS = “Pretty Crazy Situation”?Proposal

CDMA (IS-95)

PCS1900IS-54

OmnipointWACS

DECTPHS

5 MHz CDMA

MobilityHigh

HighHigh

High/LowLow

LowLowLow

SupportersQualcomm, AT&T, Motorola, ALPS,

GSIC, Samsung, Sony, US West, SprintBell Atlantic, Time Warner

Pac Bell, Nokia, MCI, Siemens, KycomAT&T, McCaw

Omnipoint, RockwellBellcore, Motorola, Panasonic, US

West, Sprint, Bell Atlantic, Time WarnerEricsson

NEC, Panasonic, Hitachi, Toshiba, PCSIInterdigital, Oki

•• PCS Primeco (Nynex/Bell Atlantic/US West/Airtouch): CDMA PCS Primeco (Nynex/Bell Atlantic/US West/Airtouch): CDMA•• Wirelessco (Sprint, Comcast, Cox, TCI): CDMA Wirelessco (Sprint, Comcast, Cox, TCI): CDMA•• AT&T Wireless: TDMA AT&T Wireless: TDMA•• Other RBOCs (e.g., PBMS): PCS 1900 (Upbanded GSM) Other RBOCs (e.g., PBMS): PCS 1900 (Upbanded GSM)

16







17

Mobile Computing is theMobile Computing is theNatural Evolution Natural Evolution of Computingof Computing

Single UserOS

Batch

Timesharing

Networking

LANs + WSs

Mobile Computing

Freedom from Collocation

MoreFlexibleResourceUsage

18

90 91 92 93 94 95 96 97 98 99 00

10

20

30

40

50

60

90 91 92 93 94 95 96 97 98 99 0

Global Markets for Portable ComputersGlobal Markets for Portable Computers$ Billions

You Are Here!Total PC Market: approx. $150 billion

19

Laptop Sales, Jan-Sept 1995Laptop Sales, Jan-Sept 1995

0 200,000 400,000 600,000

Toshiba

Compaq

IBM

Apple

NEC

Dell

Zenith

Texas Instr

AST

AT&T Unit Sales(Top 10)

From NYTimes17 Jan 96

556

331

301

203

137

97

75

72

68

46

Approx. 20%Approx. 20%of units areof units are

laptop computerslaptop computers

20

A Short History of Wireless DataA Short History of Wireless Data•• Wireless LANsWireless LANs

–– late 1970slate 1970s»» IBM Research--IR for factory apps (<<1 mbps)IBM Research--IR for factory apps (<<1 mbps)»» HP--RF for 100 kbps to distributed terminalsHP--RF for 100 kbps to distributed terminals»» Motorola--1.7 GHz prototype LAN extentionMotorola--1.7 GHz prototype LAN extention

–– 1985: ISM bands opened for low power SS1985: ISM bands opened for low power SS»» Evolves into the PCMCIA formfactor interfaceEvolves into the PCMCIA formfactor interface

–– Recent DevelopmentsRecent Developments»» 1990: Motorola--18 GHz Altair LAN extension1990: Motorola--18 GHz Altair LAN extension»» 1990: IEEE 802.11 Committee Formed1990: IEEE 802.11 Committee Formed»» 1992: Winforum Committee Formed1992: Winforum Committee Formed»» 1992: European ETSI HiperLAN Effort1992: European ETSI HiperLAN Effort»» 1993: Licensed and Unlicensed PCS Bands1993: Licensed and Unlicensed PCS Bands»» 1995: 60 GHz spectrum allocation1995: 60 GHz spectrum allocation

21

A Short History of Wireless DataA Short History of Wireless Data•• Wide Area DataWide Area Data

–– ArdisArdis»» 1983: IBM + Motorola; IBM sells out in 19941983: IBM + Motorola; IBM sells out in 1994

»» 1994: 35,000 users, 1300 base stations. 45 million msgs 1994: 35,000 users, 1300 base stations. 45 million msgs per month, 19.2 kbpsper month, 19.2 kbps

–– RAM Mobile Data/MobitexRAM Mobile Data/Mobitex»» Bell South/Ericsson/MotorolaBell South/Ericsson/Motorola

»» 1986: deployed in Sweden1986: deployed in Sweden

»» 1989: 840 base stations, 40 switching centers, in 100 1989: 840 base stations, 40 switching centers, in 100 metro areas, 6300 US cities and towns, 8000 kbps, metro areas, 6300 US cities and towns, 8000 kbps, nationwide roamingnationwide roaming

»» 1994: 12,000 subscribers in US1994: 12,000 subscribers in US

22

A Short History of Wireless DataA Short History of Wireless Data•• Wide Area DataWide Area Data

–– CDPDCDPD»» 1991: CDPD patents filed by 3 IBMers (Miller, Moore, 1991: CDPD patents filed by 3 IBMers (Miller, Moore,

Pate, IBM Boca Raton)Pate, IBM Boca Raton)

»» 1993: Initial CDPD specifications, carriers plus IBM. 1993: Initial CDPD specifications, carriers plus IBM. McCaw, IBM, PCSIMcCaw, IBM, PCSI

»» 1995: Deployment of CDPD infrastructure1995: Deployment of CDPD infrastructure

»» 19.2 kbps, make use of idle channels in analog cellular 19.2 kbps, make use of idle channels in analog cellular system via "channel hopping"; TCP/IP compliant with system via "channel hopping"; TCP/IP compliant with support for mobile-IP like mobility routingsupport for mobile-IP like mobility routing

»» Not viewed as a marketing success (yet): where are the Not viewed as a marketing success (yet): where are the subscribers? how pervasive is the coverage?subscribers? how pervasive is the coverage?

23

Some ObservationsSome Observations

•• Wireless Data ServicesWireless Data Services–– Have been around for more than 10 yearsHave been around for more than 10 years–– Messaging SystemsMessaging Systems

»» EMBARC, MobileComm, SkyTelEMBARC, MobileComm, SkyTel»» low speed (100s bps), short message, two-way availablelow speed (100s bps), short message, two-way available

–– Wide-Area Data NetworksWide-Area Data Networks»» 100,000 subscribers in US??100,000 subscribers in US??»» RadioMail: Internet E-mail gateway:RadioMail: Internet E-mail gateway:

a few thousand subscribers?a few thousand subscribers?–– Wireless Local Area NetworksWireless Local Area Networks

»» Large number of manufacturers yet small sales volumesLarge number of manufacturers yet small sales volumes»» Approximately 1.2 million units sold to dateApproximately 1.2 million units sold to date

24

Technical and Economic ChallengesTechnical and Economic Challenges•• PricingPricing

–– $/airtime, $/data transfer, flat rate?$/airtime, $/data transfer, flat rate?–– Transmitted packets or received packets?Transmitted packets or received packets?–– Artifact provider vs. pipe provider vs. service providerArtifact provider vs. pipe provider vs. service provider

•• PerformancePerformance–– 20 kbps but not quite fast enough for “multimedia”20 kbps but not quite fast enough for “multimedia”–– Coverage, non-interoperabilityCoverage, non-interoperability

•• Killer AppKiller App–– Vertical application vs. email, paging, web accessVertical application vs. email, paging, web access

•• FormfactorFormfactor–– Laptops vs. PDAs vs. Smart PhonesLaptops vs. PDAs vs. Smart Phones

25

Promising DevelopmentsPromising Developments

•• Converging PlatformsConverging Platforms–– Smart phones: cellular phone + built-in pager/PDASmart phones: cellular phone + built-in pager/PDA

»» e.g., Nokia/Geoworks “Internet” Phonee.g., Nokia/Geoworks “Internet” Phone

»» e.g., IP-addressable two-way pagers for Narrowband PCSe.g., IP-addressable two-way pagers for Narrowband PCS

•• PCS LicensesPCS Licenses–– Room for Room for allall of those cellular service providers? of those cellular service providers?–– Possible packet data services?Possible packet data services?

•• Emerging Industry StandardsEmerging Industry Standards–– CDPD, IEEE 802.11, WINFORUMCDPD, IEEE 802.11, WINFORUM–– Better interoperation or co-existance in the futureBetter interoperation or co-existance in the future–– 60 GHz allocation60 GHz allocation

26

Promising DevelopmentsPromising Developments

•• 60 GHz Allocation60 GHz Allocation–– FCC opens 5 GHz of spectrum at 59-64 GHzFCC opens 5 GHz of spectrum at 59-64 GHz–– Oxygen absorption leads to rapid fall off in signal Oxygen absorption leads to rapid fall off in signal

strength: makes allocation perfect for (very) high strength: makes allocation perfect for (very) high bandwidth WLAN bandwidth WLAN

–– MM Wave Communications Working Group is MM Wave Communications Working Group is defining sharing etiquettedefining sharing etiquette

–– Many technical challenges remain: multipath, Many technical challenges remain: multipath, antenna arraysantenna arrays

–– Many economic challenges remain: high volume Many economic challenges remain: high volume production and reduced costsproduction and reduced costs

27

MSC

Promising Developments: GPRSPromising Developments: GPRS

PDA + Modem +Digital Cell Phone

NokiaEricsson

Hewlett PackardGeoworks Software

PCS 1900PacBell Mobile Servicesdeployment in Bay Area

Fall 1996

BSS

ProxyServer

InternetGateway

PSTN

Internet

CellularInfrastructure

General Packet Radio Service (GPRS) of GSM SystemExplicit specification for proxy support services

28

Campus Network

Promising Developments: Promising Developments: Wireless CableWireless Cable

Dial-UpRouter

SparcStation

HEM 1110Enc/Mod

RF Upconv

Studio-to-Xmit Link

RLA

Dwinelle Hall“Hybrid POP”

Soda Hall

Router

Coax

“RemoteLink Adapter”

10 Mbps10 Mbpsin 6 MHz

@ 2.4 GHz

Metricom

DownstreamForward Channel

UpstreamReturn Channel

Hybrid RouterDownConverter

29

Wireless ATMWireless ATM

•• Uttered in same phrase as Uttered in same phrase as mobile multimediamobile multimedia

•• Switched scalable bandwidth is good thingSwitched scalable bandwidth is good thing

•• But ...But ...–– QoS guarantees incompatible with radio channels: QoS guarantees incompatible with radio channels:

still need to write applications that can adaptstill need to write applications that can adapt–– Mobility presents problems for ATM’s connection Mobility presents problems for ATM’s connection

orientation: connection tear-down and rebuildorientation: connection tear-down and rebuild–– No mechanisms for shared media: implications for No mechanisms for shared media: implications for

costs of the wireless networkcosts of the wireless network

30

What Every Happened to What Every Happened to Ubiquitous Computing?Ubiquitous Computing?

•• Computers are getting more ubiquitousComputers are getting more ubiquitous–– Multiple computers in my officeMultiple computers in my office–– Have you been to an airport lately?Have you been to an airport lately?–– Echelon style in-building controls: chip in a light Echelon style in-building controls: chip in a light

socketsocket

•• But it is information, in the form of the web, But it is information, in the form of the web, that has become ubiquitous, not the computer!that has become ubiquitous, not the computer!–– 47 million Americans on the Internet at end of 199647 million Americans on the Internet at end of 1996–– 12 million more than at start of 1996: 34% increase12 million more than at start of 1996: 34% increase

31

Wirless Data Success Story: UPSWirless Data Success Story: UPS

•• 60,000 widely dispersed mobile units 60,000 widely dispersed mobile units communicating with a central DB servercommunicating with a central DB server

•• Competitive need for real-time package trackingCompetitive need for real-time package tracking

•• “TotalTrack”: UPS + 100 cellular carriers“TotalTrack”: UPS + 100 cellular carriers–– Cellular modems, bulk service from an alliance of Cellular modems, bulk service from an alliance of

cellular service providers, interfaced to UPSnet, UPS’ cellular service providers, interfaced to UPSnet, UPS’ own packet switched networkown packet switched network

–– Integrated billing, roaming, redundancyIntegrated billing, roaming, redundancy–– ARDIS & RAM rejected due to insufficient coverageARDIS & RAM rejected due to insufficient coverage

32







33

AccessAccess is the Killer App is the Killer App

•• Killer CapabilitiesKiller Capabilities–– Access to your desktop environment while on the moveAccess to your desktop environment while on the move–– Be connected whenever possible via the “best” Be connected whenever possible via the “best”

available networkavailable network–– Support the application’s graceful adaptation to the Support the application’s graceful adaptation to the

available bandwidth and latencyavailable bandwidth and latency

•• Technical StrategiesTechnical Strategies–– Continuous Connectivity: Wireless OverlaysContinuous Connectivity: Wireless Overlays–– Proxies: Network- and Type-Aware Applications Proxies: Network- and Type-Aware Applications

Building BlocksBuilding Blocks

34

Overlay Network ChallengeOverlay Network ChallengeType of

NetworkBandwidth Latency Mobility Typ Video

PerformanceTyp Audio

PerformanceIn-Building >> 1 Mbps

Comm’l RF: 2 MbpsResearch IR: 50 Mbps

< 3 ms Pedestrian 2-Way ’ractiveFull Frame Rate

(Comp)

High Quality16-bit Samples22 KHz Rate

Campus-AreaPacket Relay

Network

≈ 64 Kbps ≈ 100 ms Pedestrian Med. QualitySlow Scan

Med. QualityReduced Rate

Metro-Area(Wireless

Cable)

10-30 mbps(one way, LOS)

< 10 ms Stationary 2-Way 'ractiveFull Frame Rate

(Compressed)

High Quality16-bit Samples22 KHz Rate

Wide-Area 19.2 Kbps > 100 ms Vehicular Freeze Frame Asynchronous“Voice Mail”

Regional-Area(LEO/DBS/VSAT)

4.8 kbps–10+ Mbps(asymmetric)

> 100 ms VehicularStationary

Seconds/FrameFreeze Frame

Asynchronous“Voice Mail”

Latency as critical as bandwidth in wireless networksWide diversity of network performance parameters

Competing infrastructure providers Pedestrian vs. vehicular mobility

35

Application Support ChallengeApplication Support Challenge

Client variation spans an order of magnitudeClient variation spans an order of magnitude

Device Bandwidth,bits/sec

CPU Mem/Disk

Screensize

Bits/pixel

High-endPC

Ethernet(10Mbits),

266 MhzPentium Pro

64/4G 1280x1024 16-24,color

Low-endPC

ISDN (128K) 150 MhzPentium

16/1G 1024x768 8-16,color

High-endnotebook

Cellular (9600)or wireline

800x600 8, color

Low-endnotebook

(28.8K) modem 100 Mhz 486 640x480 4, gray

PDA 2400-14.4Kmodem

20+ MhzRISC or x86

2/0 320x200 1–2, gray

36

Wireless Overlay NetworksWireless Overlay Networks

High-tier

Low-tier

Satellite

High Mobility Low MobilityWide Area

Regional Area

Local Area

•• Wireless internetwork: stacked networks Wireless internetwork: stacked networks providing wide area coverage and the best providing wide area coverage and the best possible bandwidth and latencypossible bandwidth and latency

37

Proxy ArchitectureProxy Architecture

Proxy

•• ProxyProxy–– Mediates between wireless and wireline environmentMediates between wireless and wireline environment–– Ideally executes at “well-connected” boundaryIdeally executes at “well-connected” boundary–– Manages caches & chooses representations on-the-flyManages caches & chooses representations on-the-fly–– Trade transcoding time against communications timeTrade transcoding time against communications time

Well Connected Poorly Connected

38

BARWAN ArchitectureBARWAN Architecture

IPInternet

GW

GW

GW

Server(Correspondent Host)

Proxy Host

Home AgentWireless Subnets

Basestations ForeignAgent

LocalProxy Host

LocalServices

HorizontalHandoff

VerticalHandoff

Overlay IP

•• Overlay IP extends Mobile IPOverlay IP extends Mobile IP•• Proxy/Forwarding Agent InteractionProxy/Forwarding Agent Interaction•• “Vertical” Handoffs between subnetworks“Vertical” Handoffs between subnetworks

39

Mobile Applications

“Ses

sio

ns”

Proxy AgentsTransport

(Asymmetric, Heterogeneous, Lossy Links)

LinkScheduling

Wireless LAN

PacketRadio

BARWAN ArchitectureBARWAN Architecture

Mobile IPBeacons

Net ConnectionMonitoring;

Net-ApplInterface;

ELN;

Overlay IP (Mobile IP + Overlays) Lo

cati

on

-Dep

end

ent

Ser

vice

s

Snoop Agent IP

Wire-LessCable

DBSSub-Net

CDPDSub-Net

ATM

40







41

SummarySummary

•• Wireless data is Wireless data is notnot dead dead–– Very successful vertical applications like UPSVery successful vertical applications like UPS–– Wide-Area: ISP model vs. “pipe” provider, GPRSWide-Area: ISP model vs. “pipe” provider, GPRS–– Local-Area: Higher b/w, 802.11 standards, etiquetteLocal-Area: Higher b/w, 802.11 standards, etiquette

•• Access is the killer appAccess is the killer app–– Connectivity through wireless overlaysConnectivity through wireless overlays–– Adaptivity through proxy servicesAdaptivity through proxy services

•• IP “Dial Tone”IP “Dial Tone”–– IP to the pager, phone, PDA, laptop, etc.IP to the pager, phone, PDA, laptop, etc.–– Internet service integrationInternet service integration

»» A phone is not a workstation: limited b/w, storage, displayA phone is not a workstation: limited b/w, storage, display

42

ConclusionsConclusions

•• Promising Developments:Promising Developments:–– Overlays and ProxiesOverlays and Proxies–– New spectrum allocations for serious bandwidthNew spectrum allocations for serious bandwidth

•• Many of these ideas apply equally well in Many of these ideas apply equally well in today’s heterogeneous network environmentstoday’s heterogeneous network environments–– Asymmetries in satellites and cable modem linksAsymmetries in satellites and cable modem links–– Transport over high bandwidth-delay networksTransport over high bandwidth-delay networks–– High loss links: bandwidth achieved?High loss links: bandwidth achieved?–– Adapt representations to the quality of the end Adapt representations to the quality of the end

device and its network connectivitydevice and its network connectivity

is wireless (data) dead?bnrg.cs.berkeley.edu/~randy/talks/cc97/cc97.pdfis wireless (data) dead?...

Documents