CAN CELLULAR NETWORKS
KEEP UP WITH THE GROWTH
      OF BIG DATA?
             David Goodman
               NYU WIRELESS
  Polytechnic Institute of New York University
             dgoodman@poly.edu


   IEEE VTS-UKRI CHAPTER MEETING
             July 26, 2012
                                      David Goodman July 26, 2012
Cellular Data Crisis 2010


                                                            January 28, 2010




                    @ 4G World: Carriers brace for data storm
                               October 20, 21 2010
Like the rest of the world, Japan is forecasting explosive growth in data traffic.
      Japanese end users are demanding high-definition digital cameras,
                 access to YouTube, e-books and online games.



                                                           David Goodman July 25, 2012
Cellular Data Crisis
             2012 Update
4G upgrade signals
threat and                           FT: Feb. 25, 2012
opportunity

10-fold surge in global mobile data traffic … by 2016

   Non-voice service revenue annual % growth:
          Italy 1.8, UK 4.9, Germany 7.0

                                       David Goodman July 25, 2012
Cellular Data Crisis
                                          COST
$$




                                         REVENUE




              Time    iPhone etc.
     SMS
                            David Goodman July 25, 2012
Cellular Data Crisis


• Revenue 2005-09     $8.5B →$37B       4x

• AT&T Data Usage 2006-09             70x

• Japan data usage 2007 – 2017      200x ??



                                 David Goodman July 25, 2012
Cellular Data Usage


• BlackBerry = Cellphone x 2

• iPhone = BlackBerry x 5

• Aircard = iPhone x 5

• iPad, future devices and services = ???



                                     David Goodman July 25, 2012
Cellular Service Plans
• Voice      $60      750 minutes          $600/GB

• SMS        $20      1000 texts*      $60000/GB

• Data       $60      5 GB                    $12/GB

• iPhone $30          unlimited         →$5/GB??
Pandora=60 MB/hour     Live video =300 MB/hour
*2011: 8T texts = $127B = $53,000/GB
                                        David Goodman July 25, 2012
What’s wrong with this picture?

            Revenue        Demand
     Voice Usage sensitive Limited
     SMS

     Data   Fixed          Unlimited




                               David Goodman July 25, 2012
Industry response $10-50/GB
                From June 28, 2012




                        David Goodman July 25, 2012
New picture (June 2012)

        Revenue         Demand
  Voice Usage sensitive Limited
  SMS Fixed
  Data Fixed            Unlimited
        Usage sensitive



                            David Goodman July 25, 2012
Increase Network Capacity


• Cell splitting

• More spectrum

• Advanced technology



                          David Goodman July 25, 2012
Increase Network Capacity
                   1985-2008

• Total                        x10,000

• Cell splitting               x400

• More spectrum                x3.4

• Efficient technology         x7


                                    David Goodman July 25, 2012
FCC Auction 2008
 $19,000,000,000




               David Goodman July 25, 2012
4G Spectrum Auctions
France    €3.8B

Germany   €4.4B

Italy     €3.9B

Spain     €1.6B

UK        Late 2012
                  David Goodman July 25, 2012
Cellular technology evolution
            wider transmission bands

  1G            2G       3G        4G



25-30 kHz     200 kHz   5 MHz   1.4-20 MHz        100 MHz


1980          1990      2000       2010              20??

AMPS          GSM       WCDMA      LTE               LTE-A




                                        David Goodman July 25, 2012
Cellular technology evolution
            Internet core network

 1G               2G, 3G       4G




Cellular
                SS7 - ISDN     IP
protocols




                             David Goodman July 25, 2012
Cellular technology evolution
              multiple access

  1G, “4G”          2G                   3G



 FDMA               TDMA                 CDMA
             1990                 2000




                           2010




                                         David Goodman July 25, 2012
Cellular technology evolution
              system names




    1G               2G                   3G
 1980 – NMT        1990 – GSM          2000 – UMTS
  NORDIC           GLOBAL          UNIVERSAL

                       4G

                    2010 – ???
                      LTE
                                 David Goodman July 25, 2012
Addressing the data crisis
                  Industry trends
• LTE
  –   Up to 20 MHz signal band
  –   Up to 300/75 (DL/UL) throughput
  –   Up to 4x4 MIMO
  –   IP core network
• Video compression
• Smaller cells
• Wi-Fi off-loading

                                        David Goodman July 25, 2012
NYU WIRELESS
Director: Prof. Ted Rappaport   nyuwireless.org




                                       David Goodman July 25, 2012
Addressing the data crisis
              Future directions

• Network optimization with small cells

• mm wave transmission

• Cooperative communications

• Cognitive radio

• Adaptive pricing

                                  David Goodman July 25, 2012
Addressing the data crisis
       NYU WIRELESS RESEARCH

• Network with small cells
  – Optimum cell selection Prof. Rangan
  – Handover for mobile terminals Prof. Panwar
• mm wave communication, Prof. Rappaport
  – Measurements, devices
• Cooperative communications
  Prof. Erkip, Knox, Panwar
  – Algorithms, signal processing, testbed


                                         David Goodman July 25, 2012
Femto and pico
     Femto- and Picocells:                             cells
     Small is Beautiful
                    Smaller, easily deployable, lower cost




Traditional                   Nokia MetroSite                Huawei femtocell
macrocell tower               UMTS base                      “personal base
                              station                        station”


5   Wireless Research Lab

                                                             David Goodman July 25, 2012
Femtocell
                  Use customer
                  broadband for
                  backhaul

                   Offload traffic from
                   Macro cells




            David Goodman July 25, 2012
Femtocell challenges
• Uncontrolled interference environment
  – Unplanned deployments.
  – Large power disparities between macro and femto.
  – Restricted association & closed subscriber groups

• Need for self-configuration & “plug & play”.
  – Must synchronize for macro-femto handover & to
    enable advanced interference mitigation
    techniques.

• Mixed backhaul: operator and subscriber’s ISP.
                                      David Goodman July 25, 2012
Small cells are coming
 Today's Top News (June 25, 2012)
Rumor mill: AT&T gears up to order
        100,000 small cells
(283,000 cells total in USA at end of 2011)

 Small cells=femto < pico < metro < micro
         www.smallcellforum.org



                                 David Goodman July 25, 2012
Backhaul becoming a dominant
             cost
• Backhaul represents significant cost to operators:
   – 30 to 50% of OPEX costs by various estimates

• Cost and management of airlink is decreasing
   – Lower power / smaller cells
   – Self-organization

• With greater bandwidth and larger numbers of cell
  sites, backhaul will become dominant expense

• Is there a cheaper way to deliver backhaul?

                                         David Goodman July 25, 2012
Cell selection with opportunistic
          backhaul: Prof S. Rangan
•   Cell selection now based on signal strength
•   Add backhaul as selection criterion
•   Provision for average (not maximum) traffic
•   Opportunistic backhaul challenges
    –   Variable traffic loads
    –   Distributed management
    –   Heterogenous cell sizes
    –   Load balancing
    –   Small numbers of users per cell

                                          David Goodman July 25, 2012
Cellular Handover, Prof. Panwar
• Handover between macrocells takes less than 100ms.



• Handover to indoor Femtocells
   – Near stationary user
   – Slow handover: Many messages going
     over the public internet.



• Handover among outdoor picocells
   – More Frequent: Users quickly passing through small cells.



                                                   David Goodman July 25, 2012
Faster Handover
• Prefetch user data to picocells in proximity.
• Add a fast over-the-air interface connecting picocells.
• Decouple control messages from actual handoff.

Handover Durations for Femtocells
• Legacy Handover               ~             1500ms.
  (applied as it is)
• Prefetch-based Fast Handover ~              800ms.
  (with all control messaging via backhaul)
• Near-Instantaneous Handover ~               30ms.
  (with OTA interface)




                                                David Goodman July 25, 2012
Background
 Spectrum Allocation for 60GHz – key
        mmWave frequency
                                                                                                                      • Spectrum
                                                                                                                        allocation similar
                                                                                                                        worldwide

                                                                                                                      • 5 GHz common
                                                                                                                        bandwidth
                                                                                                                        among several
                                                                                                                        countries


•Park, C., Rappaport, T.S. , “Short Range Wireless Communications for Next Generation Networks: UWB, 60 GHz Millimeter-Wave PAN, and ZigBee,”
Vol.14, No. 4, IEEE Wireless Communications Magazine, Aug. 2007, pp 70-78.
•G. L. Baldwin, “Background on Development of 60 GHz for Commercial Use,” SiBEAM, inc. white paper, May
2007, http://sibeam.com/whtpapers/Background_on_Dev_of_60GHz_for_Commercial%20Use.pdf


                                                                                                                         David Goodman July 25, 2012
mm Wave devices
                              mmWave/sub-THz on-chip antennas
                              Probe de-embedding and calibration
                              RFICs for 60 GHz, 77 GHz, sub-THz
                              Wireless propagation measurements
                               and modeling for mmWave/sub-THz
                              On-chip channel sounding system




                                                                        “PN Generator”
                      “Antenna” Dec. 2009                                 Oct. 2010 3
                               1,4



“Bevo 1” June 2009 1,4,5
                                                 “Bevo 4” July 2010 2

                                                                    David Goodman July 25, 2012
Cooperative communications
   research at NYU WIRELESS

•Antenna diversity (virtual MIMO)
•Cooperative coding
•Coop MAC – IEEE802
•Relay
•Video multicast
•P-to-P content sharing


                               David Goodman July 25, 2012
Cooperative communications

Portable devices communicate with each other

             S1




                             Base Station
                   S2




     Bandwidth consumers → suppliers
                                     David Goodman July 25, 2012
Adaptive Pricing Research



                 QuickTime™ and a
                  decompressor
          are needed to see this picture.




                                            David Goodman July 25, 2012
Beyond the data crisis

• Align theory, experiments, and systems
• Challenges become opportunities
  – Mobile (distribute information)
  – Portable (human energy source)
  – Free space (broadcasting, cooperation)
• Application specific networks
  –   Health
  –   Transportation
  –   Environment
  –   Entertainment
                                         David Goodman July 25, 2012

David Goodman - IEEE VTS UKRI - Can cellular networks keep up with the growth of big data?

  • 1.
    CAN CELLULAR NETWORKS KEEPUP WITH THE GROWTH OF BIG DATA? David Goodman NYU WIRELESS Polytechnic Institute of New York University dgoodman@poly.edu IEEE VTS-UKRI CHAPTER MEETING July 26, 2012 David Goodman July 26, 2012
  • 2.
    Cellular Data Crisis2010 January 28, 2010 @ 4G World: Carriers brace for data storm October 20, 21 2010 Like the rest of the world, Japan is forecasting explosive growth in data traffic. Japanese end users are demanding high-definition digital cameras, access to YouTube, e-books and online games. David Goodman July 25, 2012
  • 3.
    Cellular Data Crisis 2012 Update 4G upgrade signals threat and FT: Feb. 25, 2012 opportunity 10-fold surge in global mobile data traffic … by 2016 Non-voice service revenue annual % growth: Italy 1.8, UK 4.9, Germany 7.0 David Goodman July 25, 2012
  • 4.
    Cellular Data Crisis COST $$ REVENUE Time iPhone etc. SMS David Goodman July 25, 2012
  • 5.
    Cellular Data Crisis •Revenue 2005-09 $8.5B →$37B 4x • AT&T Data Usage 2006-09 70x • Japan data usage 2007 – 2017 200x ?? David Goodman July 25, 2012
  • 6.
    Cellular Data Usage •BlackBerry = Cellphone x 2 • iPhone = BlackBerry x 5 • Aircard = iPhone x 5 • iPad, future devices and services = ??? David Goodman July 25, 2012
  • 7.
    Cellular Service Plans •Voice $60 750 minutes $600/GB • SMS $20 1000 texts* $60000/GB • Data $60 5 GB $12/GB • iPhone $30 unlimited →$5/GB?? Pandora=60 MB/hour Live video =300 MB/hour *2011: 8T texts = $127B = $53,000/GB David Goodman July 25, 2012
  • 8.
    What’s wrong withthis picture? Revenue Demand Voice Usage sensitive Limited SMS Data Fixed Unlimited David Goodman July 25, 2012
  • 9.
    Industry response $10-50/GB From June 28, 2012 David Goodman July 25, 2012
  • 10.
    New picture (June2012) Revenue Demand Voice Usage sensitive Limited SMS Fixed Data Fixed Unlimited Usage sensitive David Goodman July 25, 2012
  • 11.
    Increase Network Capacity •Cell splitting • More spectrum • Advanced technology David Goodman July 25, 2012
  • 12.
    Increase Network Capacity 1985-2008 • Total x10,000 • Cell splitting x400 • More spectrum x3.4 • Efficient technology x7 David Goodman July 25, 2012
  • 13.
    FCC Auction 2008 $19,000,000,000 David Goodman July 25, 2012
  • 14.
    4G Spectrum Auctions France €3.8B Germany €4.4B Italy €3.9B Spain €1.6B UK Late 2012 David Goodman July 25, 2012
  • 15.
    Cellular technology evolution wider transmission bands 1G 2G 3G 4G 25-30 kHz 200 kHz 5 MHz 1.4-20 MHz 100 MHz 1980 1990 2000 2010 20?? AMPS GSM WCDMA LTE LTE-A David Goodman July 25, 2012
  • 16.
    Cellular technology evolution Internet core network 1G 2G, 3G 4G Cellular SS7 - ISDN IP protocols David Goodman July 25, 2012
  • 17.
    Cellular technology evolution multiple access 1G, “4G” 2G 3G FDMA TDMA CDMA 1990 2000 2010 David Goodman July 25, 2012
  • 18.
    Cellular technology evolution system names 1G 2G 3G 1980 – NMT 1990 – GSM 2000 – UMTS NORDIC GLOBAL UNIVERSAL 4G 2010 – ??? LTE David Goodman July 25, 2012
  • 19.
    Addressing the datacrisis Industry trends • LTE – Up to 20 MHz signal band – Up to 300/75 (DL/UL) throughput – Up to 4x4 MIMO – IP core network • Video compression • Smaller cells • Wi-Fi off-loading David Goodman July 25, 2012
  • 20.
    NYU WIRELESS Director: Prof.Ted Rappaport nyuwireless.org David Goodman July 25, 2012
  • 21.
    Addressing the datacrisis Future directions • Network optimization with small cells • mm wave transmission • Cooperative communications • Cognitive radio • Adaptive pricing David Goodman July 25, 2012
  • 22.
    Addressing the datacrisis NYU WIRELESS RESEARCH • Network with small cells – Optimum cell selection Prof. Rangan – Handover for mobile terminals Prof. Panwar • mm wave communication, Prof. Rappaport – Measurements, devices • Cooperative communications Prof. Erkip, Knox, Panwar – Algorithms, signal processing, testbed David Goodman July 25, 2012
  • 23.
    Femto and pico Femto- and Picocells: cells Small is Beautiful Smaller, easily deployable, lower cost Traditional Nokia MetroSite Huawei femtocell macrocell tower UMTS base “personal base station station” 5 Wireless Research Lab David Goodman July 25, 2012
  • 24.
    Femtocell Use customer broadband for backhaul Offload traffic from Macro cells David Goodman July 25, 2012
  • 25.
    Femtocell challenges • Uncontrolledinterference environment – Unplanned deployments. – Large power disparities between macro and femto. – Restricted association & closed subscriber groups • Need for self-configuration & “plug & play”. – Must synchronize for macro-femto handover & to enable advanced interference mitigation techniques. • Mixed backhaul: operator and subscriber’s ISP. David Goodman July 25, 2012
  • 26.
    Small cells arecoming Today's Top News (June 25, 2012) Rumor mill: AT&T gears up to order 100,000 small cells (283,000 cells total in USA at end of 2011) Small cells=femto < pico < metro < micro www.smallcellforum.org David Goodman July 25, 2012
  • 27.
    Backhaul becoming adominant cost • Backhaul represents significant cost to operators: – 30 to 50% of OPEX costs by various estimates • Cost and management of airlink is decreasing – Lower power / smaller cells – Self-organization • With greater bandwidth and larger numbers of cell sites, backhaul will become dominant expense • Is there a cheaper way to deliver backhaul? David Goodman July 25, 2012
  • 28.
    Cell selection withopportunistic backhaul: Prof S. Rangan • Cell selection now based on signal strength • Add backhaul as selection criterion • Provision for average (not maximum) traffic • Opportunistic backhaul challenges – Variable traffic loads – Distributed management – Heterogenous cell sizes – Load balancing – Small numbers of users per cell David Goodman July 25, 2012
  • 29.
    Cellular Handover, Prof.Panwar • Handover between macrocells takes less than 100ms. • Handover to indoor Femtocells – Near stationary user – Slow handover: Many messages going over the public internet. • Handover among outdoor picocells – More Frequent: Users quickly passing through small cells. David Goodman July 25, 2012
  • 30.
    Faster Handover • Prefetchuser data to picocells in proximity. • Add a fast over-the-air interface connecting picocells. • Decouple control messages from actual handoff. Handover Durations for Femtocells • Legacy Handover ~ 1500ms. (applied as it is) • Prefetch-based Fast Handover ~ 800ms. (with all control messaging via backhaul) • Near-Instantaneous Handover ~ 30ms. (with OTA interface) David Goodman July 25, 2012
  • 31.
    Background Spectrum Allocationfor 60GHz – key mmWave frequency • Spectrum allocation similar worldwide • 5 GHz common bandwidth among several countries •Park, C., Rappaport, T.S. , “Short Range Wireless Communications for Next Generation Networks: UWB, 60 GHz Millimeter-Wave PAN, and ZigBee,” Vol.14, No. 4, IEEE Wireless Communications Magazine, Aug. 2007, pp 70-78. •G. L. Baldwin, “Background on Development of 60 GHz for Commercial Use,” SiBEAM, inc. white paper, May 2007, http://sibeam.com/whtpapers/Background_on_Dev_of_60GHz_for_Commercial%20Use.pdf David Goodman July 25, 2012
  • 32.
    mm Wave devices  mmWave/sub-THz on-chip antennas  Probe de-embedding and calibration  RFICs for 60 GHz, 77 GHz, sub-THz  Wireless propagation measurements and modeling for mmWave/sub-THz  On-chip channel sounding system “PN Generator” “Antenna” Dec. 2009 Oct. 2010 3 1,4 “Bevo 1” June 2009 1,4,5 “Bevo 4” July 2010 2 David Goodman July 25, 2012
  • 33.
    Cooperative communications research at NYU WIRELESS •Antenna diversity (virtual MIMO) •Cooperative coding •Coop MAC – IEEE802 •Relay •Video multicast •P-to-P content sharing David Goodman July 25, 2012
  • 34.
    Cooperative communications Portable devicescommunicate with each other S1 Base Station S2 Bandwidth consumers → suppliers David Goodman July 25, 2012
  • 35.
    Adaptive Pricing Research QuickTime™ and a decompressor are needed to see this picture. David Goodman July 25, 2012
  • 36.
    Beyond the datacrisis • Align theory, experiments, and systems • Challenges become opportunities – Mobile (distribute information) – Portable (human energy source) – Free space (broadcasting, cooperation) • Application specific networks – Health – Transportation – Environment – Entertainment David Goodman July 25, 2012