Arogyaswami Paulraj Broadcom Corporation & Stanford University

1. Scaling Mobile Broadband Capacity – A Fresh LOOK IEEE-USA Presentation to FCC November 6, 2012 Arogyaswami Paulraj Broadcom Corporation & Stanford University

2. Broadcom Corp. and Presenter INFO • Sr. Advisor to Broadcom Corp. • Professor (Emeritus), Stanford University • Headquartered in Irvine, CA • A global leader in semiconductors for wired and wireless communications • ~ 11,200 staff worldwide • 75 design centers globally • 2011 net revenue of $7.39 billion • Broadcom Corporation • Arogyaswami Paulraj

3. Wireless – Connecting EVERYTHING

4. Mobile BROADBANDDrivers • Richer Media • Lower Latency • Greater User Density • Service Criticality • Higher Availability Expectation • Greater Geographic Penetration Higher Network Capacity Higher Areal Reliability

5. MOBILE BROADBANDDEMAND Growth Industry Forecasts of Mobile Data Traffic Cisco Coda Traffic Relative to 2009 Yankee Group Average 80% CAGR Source: FCC, October 2010

6. Growing Network Capacity Capacity (BPS / Sq. Mile) Spectrum(Hz) Spectrum Efficiency(BPS / Hz / Cell) No. of Cells/ Sq. Mile = x x

7. FCC Goal: 300 MHz by 2015 and additional 500 MHz by 2020 in the “mobile” bands – optimistic ?? Allocating spectrum Exclusively licensed Shared Rights Unlicensed Add Mobile Spectrum Capacity (BPS / Sq. Mile) Spectrum(Hz) Spectrum Efficiency(BPS / Hz / Cell) No. of Cells/ Sq. Mile = x x

8. What can SPECTRUM BUY? • At low SNRs (typical), we cannot increase BPS / User by adding bandwidth without proportionally adding power • Uplink: Power in handsets already at Max (SAR) – so more spectrum mainly buys more users, unless we shrink cell size or add antennas • Downlink: Power in BS is not at max (as yet) so more spectrum allows scaling BPS/user and number of users Bandwidth(Hz) Signal (Power per Hz) BPS / User Log (1 + ) = x Noise (Power per Hz)

9. More spectrum efficiency Capacity (BPS / Sq. Mile) Spectrum(Hz) Spectrum Efficiency(BPS / Hz / Cell) No. of Cells/ Sq. Mile = x x Better reuse, adaptive modulation Early MIMO

10. Add Cells Capacity (BPS / Sq. Mile) Spectrum(Hz) Spectrum Efficiency(BPS / Hz / Cell) No. of Cells/ Sq. Mile = x x Challenges in adding cells • High CAPEX + OPEX costs (US cell growth 7-8% recent years) • Zoning / access • Backhaul: Fiber is prohibitive in some areas, DSL/Cable do not scale (500 MBPS/Cell), small cell radio back haul faces many technology issues

11. MOBILE BROADBAND Capacity GROWTH TO 2017 • x10 capacity growth – optimistic

12. A New Approach

13. Frequency vs. Range (Fixed Aperture)

14. Frequency vs. CAPACITY (Fixed Aperture)

15. HYPER MIMO (H–MIMO) Adding Antennas Current (Small) Aperture Higher Frequency Band • Current (Low) Frequency Band Larger Aperture

16. Attenuation Loss – High Band Attenuation Loss from Atmosphere, Rain and Foliage is Acceptable in Small Cells • Atmospheric • Rain Foliage • Weissberger’sMode where, L = The loss due to foliage (dB) f = The transmission frequency (GHz) d = The depth of foliage ‘’’along’’’ the path (m)

17. H-MIMO Propagation – High Band • High shadowing and low diffraction – so only LOS or Near LOS propagation will work – ball parks • Foliage loss is manageable with “look down” rather than “look through” deployments • Small cells keeps atmospheric and rain loss manageable • Handset antennas can be switched to avoid hand / head shadowing

18. H-MIMO: Antenna and RF Hardware • Low cost high band RF array technology is possible • Long history in military radar applications • Wi-Fi 802.15c and .11ad already building multi element arrays • Broadcom, IBM, Cybeam, … have built low-cost integrated array modules • Low band large aperture arrays need multi-use models – bill boards, architectural structures 16 element Active Array 802.15c

19. H-MIMO: Baseband and network Complexity • With Multi-User(MU)-MIMO, a 50 user baseband complexity is lower than Single User 4x4 MIMO • Other layer 2 and 3 technologies are no different from current systems – data rates are just scaled up • But strong shadowing will require rapid multi-cell coordination

20. H-MIMO: Small Aperture – High Band • Use Case: Elevated base station antenna array, small cell, for high density outdoor pedestrian subscribers • Scale up frequency by Kf • No. of BTS antennas up by Kf2 • No of handset antennas 1 (or more) • Technology – MU-MIMO • Capacity scales by Kf2 • Range unaffected(ignoring absorption loss)

21. H-MIMO: LOW BAND – large aperture • Use Case: Large array base stations for nomadic and vehicular subscribers • Aperture = Ka times wider and taller • No. of BS antennas up by Ka2 • No. of handset antennas 1 (or more) • Technology – MU-MIMO • Capacity scales by Ka2 • Range “doublings” = log (Ka2)

22. LOW Band – large aperture (Mobile station) • Use Case: Vehicular subscribers in macro cell environment • No of vehicular antennas = N • Technology – Up Link and Down Link beam forming • Range “doublings” = log (N) • Technology – MU-MIMO / COMP • Capacity scale up by x10 ?

23. H-MIMO Regulatory ISSUES • New spectrum allocation in high bands for “mobile” use • Spectrum licensing models in high bands – exclusive, shared, unlicensed • RF bio-safety in high and low bands • High gain beam forming increases EIRP • Multi-User MIMO – multiple beams • Near field • Base station zoning in low bands (for physically large arrays)

24. MOBILE BROADBAND Potential Capacity GROWTH 2017 – 2022 • x500 capacity

25. Summary • Mobile broadband capacity will severely lag demand if we rely only on current approaches • Huge capacity gains can come from H-MIMO • Needs significant spectrum, regulatory and technology initiatives

26. Thank You