Millimeter Wave Mobile Broadband: Unleashing 3-300 GHz Spectrum

2. Outline • Introduction – Mobile broadband growth – The myth of traffic and revenue gap – The national broadband plan • mmW spectrum – History of millimeter wave communications – Unleashing 3-300GHz spectrum – LMDS and 70/80/90 GHz bands • mmW Propagation characteristics – Free Space Propagation • MMB air-interface design – Duplex and multiple access schemes – Frame Structure – Channel coding and modulation • Dynamic beamforming with miniature antennas – Beamforming fundamentals – Baseband beamforming – Analog beamforming – RF beamforming – Beamforming in fading channels– Free Space Propagation – Material penetration loss – Oxygen and water absorption – Foliage absorption – Rain absorption – Diffraction – Ground reflection • mmW Mobile Broadband (MMB) network architecture – Stand-alone MMB system – MMB base station grid – Hybrid MMB + 4G systems – Deployment and antenna configuration – Beamforming in fading channels • Radio frequency components design and challenges – RF transceiver architecture – MMB RF transceiver requirement – mmWave Power amplifier – mmWave LNA • MMB system performance – Link budget analysis – Link Level performance – Geometry distribution – System throughput analysis • Summary 2Copyright © 2011 by the authors. All rights reserved.

4. Mobile Subscriber Growth 30 40 50 60 70 80 90 100 Per100inhabitants Global ICT developments, 2000-2010* Mobile cellular telephone subscriptions Internet users Fixedtelephone lines Mobile broadband subscriptions Fixedbroadband subscriptions Year 2010 • 6.9 billion world population • 5.3 billion mobile users • 2.1 billion Internet users • 940 million mobile broadband users 0 10 20 30 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010* *Estimates Source: ITU World Telecommunication /ICT Indicators database 4

5. Mobile Internet traffic pushes the limit of mobile broadband networks … 5

6. The exciting (and yet concerning) future A much larger (and growing) portion of the IP traffic to become mobile 6

7. The myth of mobile data traffic revenue gap 1000 1500 2000 2500 3000 3000 4000 5000 6000 7000 8000 mobiledatarevenue($B/yr) Mobiledatatraffic(PB/mo) 2000 3000 4000 5000 6000 400 600 800 1000 1200 1400 Computerpriceindex CPUgatecount(millions) The yawning gap between mobile data traffic and revenue But, we have seen this before … 0 500 1000 0 1000 2000 3000 2010 2011 2012 2013 2014 2015 mobiledatarevenue($B/yr) Mobiledatatraffic(PB/mo) Mobile data traffic Mobile data revenue 0 1000 0 200 400 2000 2002 2004 2006 2008 2010 Computerpriceindex CPUgatecount(millions) CPU gate count Computer price index Expon. (CPU gate count) Also available in many other occasions: •Transistor counts per IC – Moore’s Law •The hard disk storage – Kryder’s Law •Network capacity – Butter’s Law of Photonics, Nielsen’s Law •The Great Moore’s Law Compensator – Wirth’s Law 7Copyright © 2011 by the authors. All rights reserved.

8. 1000 1000 10000 mobiledatarevenue($B/yr) Mobiledatatraffic(PB/mo) 100 1000 100 1000 10000 Computerpriceindex CPUgatecount(millions) Looking from another perspective, our gap seems to be actually quite small, if any The myth of mobile data traffic revenue gap 100100 2010 2011 2012 2013 2014 2015 mobiledatarevenue($B/yr) Mobiledatatraffic(PB/mo) Mobile data traffic Mobile data revenue 1010 2000 2002 2004 2006 2008 2010 CPU gate count Computer price index Expon. (CPU gate count) The moral of the story • Your eyes may fool you • “Yawning gap” already exists in many (yet thriving) ICT industries • If there is a gap between mobile data traffic and revenue, get used to it. Need technology to provide >100x capacity with similar cost as cellular 8Copyright © 2011 by the authors. All rights reserved.

9. US Frequency Allocation Chart 9

10. National Broadband Plan Part of the American Recovery and Reinvestment Act passed in February 2009 Published on March 15, 2010 Ambitious goals •Goal 1: At least 100 million U.S. homes should have affordable access to actual download speeds of at least 100 megabits per second and actual upload speeds of at least 50 megabits per second. •Goal 2: The United States should lead the world in mobile innovation, with the fastest and most extensive wireless networks of any nation. •Goal 3: Every American should have affordable access to robust broadband service, and the means and skills to subscribe if they so choose. •Goal 4: Every community should have affordable access to at least 1 Gbps broadband service to anchor institutions such as schools, hospitals and government buildings. •Goal 5: To ensure the safety of Americans, every first responder should have access to a nationwide public safety wireless network. •Goal 6: To ensure that America leads in the clean energy economy, every American should be able to use broadband to track and manage their real-time energy consumption. 10Copyright © 2011 by the authors. All rights reserved.

11. The gist of NBP - 500MHz more spectrum Make 500 megahertz of spectrum newly available for broadband within 10 years 300 megahertz (between 225MHz and 3.7GHz) should be made available for mobile use within five years • 20 MHz of Wireless Communications Spectrum (WCS) in the 2.3GHz range• 20 MHz of Wireless Communications Spectrum (WCS) in the 2.3GHz range • 10 MHz D Block of 700MHz band re-auction for "commercial use” that is compatible with public safety services • 60 MHz of Advanced Wireless Spectrum (AWS) in the 1900MHz and 2000MHz range • 90 MHz of Mobile Satellite Spectrum to be used for terrestrial purposes in regions where it is difficult to deploy cellular • 120 MHz of reallocated TV broadcast spectrum (TV broadcasters are rallying against it) 11Copyright © 2011 by the authors. All rights reserved.

13. Wave propagation Eletric field Magnetic field The Electromagnetic Waves A changing electric field generates an oscillating magnetic field A changing magnetic field generates another changing electric field These oscillating fields forms electromagnetic waves that λλλλ electromagnetic waves that propagates at speed of light 13Copyright © 2011 by the authors. All rights reserved.

14. What is Millimeter-wave? Micro-wave •Frequency 300 MHz ~ 300 GHz (Wavelength 1mm ~ 1m) Millimeter-wave (the high-end of the micro-wave frequencies) •Frequency 30 GHz ~ 300GHz (Wavelength 1mm ~ 1cm) •In this tutorial, we refer to frequencies between 3GHz and 300GHz as millimeter-wave in our effort to explore new spectrum for mobile broadband communication Current applications of millimeter-wave •Radio astronomy, wireless backhaul, inter-satellite link, high resolution radar, security screening, amateur radio 14Copyright © 2011 by the authors. All rights reserved.

15. The History of Millimeter-wave 1864 : • The existence of electromagnetic waves was predicted by James Clerk Maxwell from his equations. 1888: • Heinrich Hertz was the first to demonstrate the existence of electromagnetic waves by building an apparatus that produced and detected microwaves in the UHF region. The design necessarily used horse-and-buggy materials, including a horse trough, a wrought iron point spark, Leyden jars, and a length of zinc gutter whose parabolic cross-section worked as a reflection antenna.1888: length of zinc gutter whose parabolic cross-section worked as a reflection antenna. 1894: • J. C. Bose publicly demonstrated radio control of a bell using millimeter wavelengths, and conducted research into the propagation of microwaves. 1931: • the first, documented, formal use of the term microwave: "When trials with wavelengths as low as 18 cm were made known, there was undisguised surprise that the problem of the micro-wave had been solved so soon." Telegraph & Telephone Journal XVII. 179/1 15Copyright © 2011 by the authors. All rights reserved.

16. Unleashing the 3-300GHz Spectrum With a reasonable assumption that about 40% of the spectrum in the mmW bands can be made available over time, we open the door for possible 100GHz new spectrum for mobile broadband • More than 200 times the spectrum currently allocated for this purpose below 3GHz. 16Copyright © 2011 by the authors. All rights reserved.

17. LMDS and 30/40 GHz bands A total of 1.3GHz spectrum available in the LMDS bands Possibly, a total of 3.4GHz spectrum available in the 38GHz (38.6–40GHz) and 40GHz (40.5-42.5) bands 17Copyright © 2011 by the authors. All rights reserved.

19. Spectrum Identified for MMB Band Frequency range [GHz] Available Spectrum [GHz] 23GHz 22.55-23.55 1.0 LMDS 27.50-28.35, 29.10-29.25, 31.075-31.225 1.3 38GHz 38.6-40.0 1.4 40GHz 40.50-42.50 2.040GHz 40.50-42.50 2.0 46GHz 45.5-46.9 1.4 47GHz 47.2-48.2 1.0 49GHz 48.2-50.2 2.0 E-band 71-76 81-86 92-95 12.9 Others -- Total 23.0 19Copyright © 2011 by the authors. All rights reserved.

21. Propagation Loss Transmission loss of mmW is accounted for principally by the free space loss Free-space loss (Isotropic antennas) • LFSL, dB = 92.4 + 20logf + 20logR • f is the carrier frequency in GHz• f is the carrier frequency in GHz • R is distance in km Other factors • Atmosphere Gaseous Losses • Precipitation attenuation • Foliage blockage • Scattering • Bending 21Copyright © 2011 by the authors. All rights reserved.

22. Solid Angle 2 Area r= rr r • One radian is defined as the plane angle with its vertex at the center of a circle of radius r that is subtended by an arc of length r • 2π radians in a full circle • One square radian (or steradian) is defined as the solid angle with its vertex at the center of a sphere of radius r that is subtended by a spherical surface of area r2 • 4π steradians in a closed sphere Copyright © 2011 by the authors. All rights reserved. 22 2 4A rπ=2C rπ=

23. Beam Area (or Beam Solid Angle) [1/2] z 0θ = θ θ sin r θ sinr dθ φ rdθ 2 sindA r d dθ θ φ= θ x y φ 0φ = φ dφ r rdφ θ dθ 2 ( )( sin ) sin dA rd r d r d d d d θ θ φ θ θ φ = = Ω Ω = Where Solid Angle subtended by the areadΩ = dA 23Copyright © 2011 by the authors. All rights reserved.

24. Beam Area (or Beam Solid Angle) [2/2] • The beam area (or beam solid angle) of an antenna is given by the integral of the normalized power pattern over a sphere: ( ) ( ) 2 0 0 , sin , A nP d d P d φ π θ π φ θ θ φ θ θ φ θ φ = = = = Ω = Ω = Ω ∫ ∫ ∫∫ 2 sin 4d d φ π θ π θ θ φ π = =  =     ∫ ∫ • The beam area is the solid angle through which all of the power radiated by the antenna would stream if P(θ, φ) maintained its maximum value over ΩΑ and was zero elsewhere. ( ) 4 ,A nP d π θ φΩ = Ω∫∫ 0 0φ θ= =   ∫ ∫ Power radiated = ( , ) AP θ φ Ω 24Copyright © 2011 by the authors. All rights reserved.

25. Directivity and Gain of an Antenna ( ) ( ) ( ) ( ) ( ) ( ) max max 2 0 0 max max , , , 1 , sin 4 , , av P D P P D P d d P P D φ π θ π φ θ θ φ θ φ θ φ θ φ θ θ φ π θ φ θ φ = = = = = = = = ∫ ∫ ( ) ( ) ( ) ( ) ( ) ( ) ( ) max max 4 4 4 4 max , , 1 1 , , 4 4 4 4 4 ,, , An P P D P d P d D P dP d P π π π π θ φ θ φ θ φ θ φ π π π π π θ φθ φ θ φ = =     Ω Ω        = = = Ω  Ω Ω     ∫∫ ∫∫ ∫∫ ∫∫ ( ) Antenna Gain where k= efficiency factor 0 Gain is less than directivity due to ohmic losses in the antenna G kD k= ≤ ≤ 25Copyright © 2011 by the authors. All rights reserved.

26. Antenna Aperture AΩ aE rE eA r 2 a e r a E A E r E P A λ = = 4 A D π = Ω 0 2 2 0 2 2 2 0 0 2 2 2 2 2 2 0 0 2 a e r A a r e A a a e e A A e E P A Z E P r Z E E A r Z Z E E A A r Z r Z A λ λ = = Ω = Ω = Ω Ω = 2 2 2 4 4 4 A A A e e e D A D A A D π λ π λ π λ Ω Ω = Ω = = = 2 2 4 1 Isotropic Antenna 4 e e A D A π λ λ π = = = 26Copyright © 2011 by the authors. All rights reserved.

27. Pathloss with Isotropic Antennas 2 2 2 Area of receive antenna of a sphere 4 Area of an ideal isotropic antenna, 4 Formula, 4 r r t r r P A P Area r A P Friis P r π λ π λ π = = =   ⇒ =     rA r 2 4A rπ= 4tP rπ    • For isotropic antennas, path loss increases with frequency 27Copyright © 2011 by the authors. All rights reserved.

28. Pathloss with directional antennas 4 effeff eff AP A Aπ  eff tA eff rA tP rPt tPG 2 2 2 2 2 4 4 4 1 effeff eff tr r r t t eff eff t rr t AP A A G P r r A AP P r π π λ π λ   = =     = ⋅ • Receive power improves with frequency!! • 31.5 dB more gain at 90GHz compared to 2.4GHz if antenna size is kept constant 28Copyright © 2011 by the authors. All rights reserved.

29. Foliage Losses At 80GHz frequency and 10 meters foliage penetration, the loss can be about 23.5dB which is about 15dB higher compared to the loss at 3GHz frequency. Foliage losses for millimeter waves are significant and can be a limiting impairment for propagation in some cases. ( ) ( ) 0.3 0.6 5 fol f D L = 29Copyright © 2011 by the authors. All rights reserved.

30. Rain Attenuations Light rain at a rate of 2.5mm/hour yields just over 1 dB/km attenuation while heavy rain at the rate of 25 mm/hour can result in over 10 dB/km attenuation at E-band (70/80/90 GHz) frequencies More severe rain such as Monsoon at a rate of 150mm/hour can jeopardize communications with up to tens of dBs of loss per km at millimeter wave frequencies. Therefore, a mechanism such as supporting emergency communications over cellular bands when mmW communications are disrupted by heavy rains should be considered as part of the MMB system design. 30Copyright © 2011 by the authors. All rights reserved.

31. Attenuations for different materials Attenuation [dB] Material Thickness [cm] 2.5GHz [7] 40GHz [8] 60GHz [7] Drywall 2.5 5.4 - 6.0 Office Whiteboard 1.9 0.5 - 9.6 Clear Glass 0.3/ 0.4 6.4 2.5 3.6 Mesh Glass 0.3 7.7 - 10.2 Chipwood 1.6 - 8.6 - Wood 0.7 - 3.5 - Plasterboard 1.5 - 2.9 - Mortar 10 - 160 - Brick wall 10 - 178 - Concrete 10 - 175 - 31Copyright © 2011 by the authors. All rights reserved.

32. Diffraction (Fresnel Zones) 1 2 1 2 n n d d r d d λ = + Higher frequencies have smaller Fresnel zone radii and hence can pass through narrow gaps with lower loss 32Copyright © 2011 by the authors. All rights reserved.

33. Ground Reflection [1/3] ( ) ( ) 2 22 2 2 1 2 2 2 2 1 1 21 1 1 1 2 2 t r t r t r t r t r t r t r d d h h d h h d h h h h d d d h h h h h h d d d d ∆ = − = + + − − +  + −    ∆ = + − +            + −    ∆ ≈ + − + =          22 , 2 t r d c h h d c f π θπ θ τ λ λ π ∆ ∆ ∆ ∆ = = = = The “Free-space” path loss exponent at mmW frequencies is likely going to be smaller than the lower RF frequencies for coverage distances of interest 33Copyright © 2011 by the authors. All rights reserved.

34. Ground Reflection [2/3] The field strength can be calculated by adding the direct and reflected ray accounting for phase differenceθ∆ ( ) ( ) ( ) ( ) ( )( ) 2 2 2 1 1 1 1 1 cos sin j T i j j E E e P e P e P j θ θ θ ρ ρ ρ θ θ ∆ ∆ ∆ − − − = + ∝ + ∝ − = − ∝ − + For grazing incidence ρ = -1 ( ) ( )( ) ( ) ( ) ( ) ( ) 2 2 2 2 2 2 2 2 2 2 2 2 1 cos sin 1 cos sin 1 cos 4 sin 2 4 sin 4sin cos 2 2 2 4sin sin cos 2 2 2 P j P P P P θ θ θ θ θ θ θ θ θ θ θ θ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∝ − +  ∝ − +   − ∝ +           ∝ +                     ∝ +             2 4sin 2 P θ∆    ∝     34Copyright © 2011 by the authors. All rights reserved.

37. MMB Network Packet Data Server Gateway #1 F G K L Cell Site 2 Cell Site 3 Cell Site 4 MS3 MS2 TX/RX R X/TX TX High-gain Tx/Rx beamforming • significantly suppress other-cell interference and • creates overlapping Packet Data Server Gateway #2 B C DE A H I J K Cell Site 1 Cell Site 5 Cell Site 6 Cell Site 7 MS1 MS2 TX/RX RX/TX • creates overlapping coverage of MMB base stations mmW in-band inter-base station backhauling • enable deployment without fixed-line high-speed backhaul 37Copyright © 2011 by the authors. All rights reserved.

38. An interference limited cellular network Clear cell boundaries and cellular footprint for each cell due to inter-cell interference Many base stations are notMany base stations are not visible to mobile station due to strong interference from a few dominant serving/interfering base stations Mobile station only communicate with one serving cell (or few serving cells) at a time (All other cells generate interference that works against the communication) 38Copyright © 2011 by the authors. All rights reserved.

39. An MMB base station grid Inter-cell interference significantly suppressed due to strong Tx/Rx beamforming (Large spatial multiplexing capability) Base stations are visible toBase stations are visible to the mobile station as long as link budget permits Large overlap of base station coverage area (No clear cell boundary) 39Copyright © 2011 by the authors. All rights reserved.

40. SINR comparison An interference limited cellular network 2 2 0 ik i ik jk j j i h P N h P ρ ≠ = + ∑ 2 0jk j j i h P N ≠ >>∑ Copyright © 2011 by the authors. All rights reserved. 40 An MMB network ( ) ( ) ( ) ( ) 2 2 0 ik i i ik k ik ik jk j j jk k jk j i h P N h P α θ β φ ρ α θ β φ ≠ ⋅ ⋅ = + ⋅ ⋅∑ Serving cell TxBF enhances signal strength RxBF enhances signal strength RxBF suppresses interferences Other cells TxBF increases interference variation Increased channel capacity

41. MMB Overlay on Cellular Users are served by MMB opportunistically Fall back to cellular when mmWave channel conditions are not suitable Achieved via dual-mode MMB and 4GAchieved via dual-mode MMB and 4G deployment •Separate MMB and 4G systems share cell sites •Inter-RAT handover to achieve fallback Or heterogeneous MMB+4G networks •MMB integrated into existing 4G system (e.g., as an extension carrier) •Users are dynamically scheduled in the MMB carriers when suitable 41Copyright © 2011 by the authors. All rights reserved.

42. Heterogeneous MMB+4G Networks Packet Data Server Gateway MS1 MBS_0 MBS_1 MBS_2 MBS_3 MBS_4 MBS_5 MBS_6 CBS_A CBS B 42Copyright © 2011 by the authors. All rights reserved. MS2 MS3 MBS_9 MBS_7 MBS_8 MBS_10 MBS_11 MBS_12 MBS_13 MBS_14 Cellular Base Station (CBS) MMB Base Station (MBS) MMB communication link cellular communication link CBS_B CBS_C Fixed line between MBS and CBS Fixed line between CBS / MBS and Packet Data Server / Gateway MS4

43. MMB In-band Dynamic Backhauling Packet Data Server Gateway MS2 BS_A mmWave Link 1 mmW Beam 1 Beam 0 Beam5 Base stations are connected via mmWave link Dynamic beamforming enables dynamic backhauling via multiple MS1 MS3 Cellular Base Station mmWave communication link cellular communication link BS_B BS_C Fixed line between BS and Packet Data Server / Gateway mmWave Link 2 WaveLink3 Beam 2 Beam 3 Beam4 43Copyright © 2011 by the authors. All rights reserved. backhauling via multiple routes •Increased throughput •Congestion mitigation •Improved robustness Backhaul link and access link share the same mmWave band •Maximize spectrum utilization •Minimize infrastructure cost

44. MMB Deployment Scenario Carrier frequency • Can be deployed between 3GHz and 300GHz • Achievable transmission power and efficiency decrease as carrier frequencies increase (limited by existing mmWave integrated circuit technology) Deployment scenariosDeployment scenarios • Urban Macro (Site-to-site distance 1km – 5km, BS Tx Power ~ 50dBm) • Urban Micro (Site-to-site distance 200m – 1km, BS Tx Power ~40dBm) • Urban Pico (Site-to-site distance 50m – 200m, BS Tx Power ~30dBm) • Femto (Site-to-site distance < 50m, BS Tx Power ~20dBm) Mobility • Support mobility up to 350kmph for carrier frequencies between 3GHz and 30GHz • Support mobility up to 120kmph for carrier frequencies between 30GHz and 90GHz 44Copyright © 2011 by the authors. All rights reserved.

45. Antenna Configurations Base station antenna configurations • 12 horn antennas per cell with each horn covers 30o field of view • 17 – 23 dB antenna gain • 6 antenna arrays per cell with each array Base station horn antenna • 6 antenna arrays per cell with each array covers 60o field of view • 64 – 1024 antenna elements for each array • 18 – 30 dB antenna gain Mobile station antenna configurations • Antenna arrays with 4 – 64 antenna elements each • 6 – 18 dB antenna gain Mobile station phase antenna array Base station phase antenna array 45Copyright © 2011 by the authors. All rights reserved.

46. Base Station with Horn Antennas 12 horn antennas per base station Each horn antenna serves one mobile station at a time Mobile stations perform antenna selection Mobile stations employ multi- antenna array Mobile stations perform Tx and Rx beamforming 46Copyright © 2011 by the authors. All rights reserved.

47. Base Station with Horn Antennas 17dB Tx Antenna Gain ~12dB Rx Antenna Gain MMB Sector Antenna Gain 16 Elements (4x4) array 2cm 47Copyright © 2011 by the authors. All rights reserved. Horn antennas at base station Antenna array at mobile station Mobile station Tx beamforming for uplink Mobile station Rx beamforming for downlink ( ) ( ) 2 2 0 ik i k ik ik jk j k jk j i h P N h P β φ ρ β φ ≠ ⋅ = + ⋅∑ Downlink SINR

48. Base Station with Antenna Arrays Antenna Arrays arranged in a hexagon structure with each array covering 600 Each antenna array can serve multiple mobile stations at a time Mobile stations perform antenna array selection Dynamic joint Tx-Rx beamforming with narrow beams within each sector 48Copyright © 2011 by the authors. All rights reserved.

49. Base Station with Antenna Arrays 49Copyright © 2011 by the authors. All rights reserved. Antenna array at base station Antenna array at mobile station MS TxBF and BS RxBF for uplink MS RxBF and BS TxBF for downlink Downlink SINR ( ) ( ) ( ) ( ) 2 2 0 ik i i ik k ik ik jk j j jk k jk j i h P N h P α θ β φ ρ α θ β φ ≠ ⋅ ⋅ = + ⋅ ⋅∑

51. Duplex schemes FDD (Frequency Division Duplex) • Requires DL and UL carriers (DL and UL can have different bandwidth) • Good coverage and link throughput • More RF/mmW hardware (separate antennas/arrays or frequency duplexers/filters) higher cost TDD (Time Division Duplex) • DL and UL transmissions occur in the same carrier but in different time slots• DL and UL transmissions occur in the same carrier but in different time slots • Tradeoff between downlink and uplink (in terms of both coverage and throughput) • Less RF/mmW hardware (DL and UL share antennas/arrays) lower cost • For MMB, the delay performance is expected to be comparable to FDD because of the short slot duration (125 us) SDD (Spatial Division Duplex) • DL and UL transmissions (to and from different mobile stations) occur in the same time- frequency resources but in different beams • Require separation of Tx/Rx circuits at the base station • SDD from the base station point of view • TDD from the mobile station point of view 51Copyright © 2011 by the authors. All rights reserved.

52. Spatial Division Duplex Tx R x R x Base station transmit and receive in the same time-frequency resource The transmission and reception are spatially separated (via Tx and Rx beamforming) R x 52Copyright © 2011 by the authors. All rights reserved. beamforming) Base stations coordinate the downlink and uplink transmissions to minimize interferences Both uplink and downlink channel state information required at the base station for proper scheduling and coordination

53. Multiple Access Flexible choice of multiple access schemes • Supports TDMA, SDMA, OFDMA, and SC-FDMA • Configurable based on hardware and deployment scenarios MMB DownlinkMMB Downlink • TDMA at slot level (OFDM or single-carrier transmission) • SDMA/OFDMA within a slot MMB Uplink • TDMA at slot level • SDMA/OFDMA/SC-FDMA within a slot 53Copyright © 2011 by the authors. All rights reserved.

54. Time-domain Channel Selectivity 3 10 30 60 90 3 8.33 27.78 83.33 166.67 250.00 30 83.33 277.78 833.33 1666.67 2500.00 120 333.33 1111.11 3333.33 6666.67 10000.00 350 972.22 3240.74 9722.22 19444.44 29166.67 3 10 30 60 90 3 120.00 36.00 12.00 6.00 4.00 Doppler (Hz) Carrier Frequency (GHz) Mobile speed (kmph) Coherence Time (ms) Carrier Frequency (GHz) Doppler calculated based on worst case scenario •The direction of velocity aligned with the direction of beamforming Design choice: MMB to support Doppler up to 10kHz (Coherent time ≥≥≥≥ 100us) •Different numerology can be developed to support mobility with higher Doppler (e.g., due to either higher speed or higher frequency) 3 120.00 36.00 12.00 6.00 4.00 30 12.00 3.60 1.20 0.60 0.40 120 3.00 0.90 0.30 0.15 0.10 350 1.03 0.31 0.10 0.05 0.03 Mobile speed (kmph) 54Copyright © 2011 by the authors. All rights reserved.

55. Frequency-domain Channel Selectivity Millimeter wave propagation characteristics in a mobile communication environment is not well understood • Most literature reports RMS delay ~ 10ns and maximum delay spread ~ 100ns • For cellular waves in a 1km cell, the maximum delay spread should mostly be within a few microseconds (1km ~ 3.3us) • High gain antennas and beamforming of mmW should significantly reduce the delay spread • As the cell size gets smaller, LOS will occur more frequently, which also reduce the delay• As the cell size gets smaller, LOS will occur more frequently, which also reduce the delay spread and frequency selectivity Best guesstimate (at this point) is the maximum delay spread of mmW in a mobile environment should mostly be less than 1us for cell radius < 1km • Equivalent to 333-meter difference among the travel distance of the rays that arrive at the receiver via the narrow beams formed by the transmitter antennas and receiver antennas Design choice: MMB to support delay spread up to 1us (Coherent bandwidth ≥ 1MHz) 55Copyright © 2011 by the authors. All rights reserved.

56. OFDM/Single-Carrier numerology Subcarrier spacing and Cyclic Prefix • Two important design parameters that pertain to many issues • Wide subcarrier spacing leads to sensitivity to frequency selectivity, and high CP overhead • Narrow subcarrier spacing leads to sensitivity to Doppler, frequency offset, and require more expensive VCOs and VCXOs • Larger CP provides protection against inter-symbol-interference due to multipath channel • Shorter CP incurs less overhead and leads to a more efficient system design• Shorter CP incurs less overhead and leads to a more efficient system design Subcarrier spacing = 270 kHz • OFDM/SC symbol length is ~3.7 us • Possible to support a coherence time of 100 us (350kmph @ 30 GHz, 120kmph @ 90 GHz) 1/8 and 1/4 CP • Both 1/8 and 1/4 CP are supported • 1/8 CP = 463 ns, low overhead • 1/4 CP = 926 ns, ensure good performance in deployment with larger cells (1~5 km) 56Copyright © 2011 by the authors. All rights reserved.

57. OFDM/SC numerology (short CP) Short CP Configurations System Bandwidth (MHz) 62.5 125 250 500 1000 Sampling rate (MHz) 69.12 138.24 276.48 552.96 1106 Subcarrier spacing (kHz) 270 OFDM symbol length (FFT size) 256 512 1024 2048 4096 OFDM symbol duration (us) 3.70OFDM symbol duration (us) 3.70 CP length 32 64 128 256 512 CP duration (us) 0.46 Slot duration (us) 125 Number of OFDM symbols per slot 30 Subframe duration (ms) 1 Number of slots per subframe 8 Frame duration (ms) 10 Number of subframes per frame 10 57Copyright © 2011 by the authors. All rights reserved.

58. OFDM/SC numerology (long CP) Long CP Configurations System Bandwidth (MHz) 62.5 125 250 500 1000 Sampling rate (MHz) 69.12 138.24 276.48 552.96 1106 Subcarrier spacing (kHz) 270 OFDM symbol length (FFT size) 256 512 1024 2048 4096 OFDM symbol duration (us) 3.70OFDM symbol duration (us) 3.70 CP length 64 128 256 512 1024 CP duration (us) 0.93 Slot duration (us) 125 Number of OFDM symbols per slot 27 Subframe duration (ms) 1 Number of slots per subframe 8 Frame duration (ms) 10 Number of subframes per frame 10 58Copyright © 2011 by the authors. All rights reserved.

60. Resource Channelization – Subbands Each subband consists of 4 RBs • Bandwidth = 4 × 4.86 = 19.44 MHz • The number of subbands for different system bandwidth • 3 (62.5 MHz), 6 (125 MHz), 12 (250 MHz), 24 (500 MHz), 48 (1 GHz) Number of occupied subcarriers • 216 / 58.6 MHz (for 256-point FFT / 62.5 MHz) • 432 / 116.9 MHz (for 512-point FFT / 125 MHz) • 864 / 233.6 MHz (for 1024-point FFT / 250 MHz) • 1728 / 466.8MHz (for 2048-point FFT / 500 MHz) • 3456 / 933.4MHz (for 4096-point FFT / 1 GHz) 60Copyright © 2011 by the authors. All rights reserved.

61. Resource Channelization – Resource Block 8 9 10 11 12 13 14 15 16 17 Each RB includes 18 subcarriers •1 RB = 18 × 270kHz = 4.86 MHz = 540 REs •24 symbols used Time Freq 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 0 1 2 3 4 5 6 7 8 Control channel RE Control channel RS Data channel RE Data channel RS •24 symbols used for data, 1 RB carries 432 data REs •The number of RBs for different system bandwidth •12 (for 62.5MHz) •24 (for 125MHz) •48 (for 250MHz) •96 (for 500MHz) •192 (for 1GHz) 61Copyright © 2011 by the authors. All rights reserved.

62. Channel Coding Regular protograph LDPC codes • Highly structured simple decoder design • High degree of parallelism high throughput • low decoding complexity low power consumption • Gbps decoder with power consumption on the order of 100mW• Gbps decoder with power consumption on the order of 100mW Data channel FEC • Length-1728 and Length-432 code with code rate (1/2, 5/8, 3/4,13/16) • Length-1728 codes lifted from the length-432 codes Control channel FEC • Length-432 LDPC code with code rate (1/2, 5/8, 3/4,13/16) • Lower code rates achieved by shortening 62Copyright © 2011 by the authors. All rights reserved.

63. LDPC 0 1 0 1 1 1 0 0 1 1 1 0 0 0 1 0 0 0 1 0 0 1 1 1 1 0 0 1 1 0 0 1 H      =       1 4, 2, 2 c r c r w w w r w = = = = Low Density Parity Check codes • Linear block codes • Row and column weights << dimension of the parity check matrix • “Capacity-approaching” • Iterative decoding with linear decoding complexity (e.g., belief propagation) 63Copyright © 2011 by the authors. All rights reserved.

64. HARQ A combination of FEC and ARQ Support both incremental redundancy (IR) and chase combining (CC) Effective in combating fast fading, interference variation, and increasing throughput Buffer management is critical for effective HARQ in MMB •1ms buffering = 1M information bits @ 1Gbps, = 16M information bits @ 16Gbps 64Copyright © 2011 by the authors. All rights reserved.

65. Modulation and Coding Schemes MCS Modulation Code rate Repetition Data rate (Mbps) Spectral Efficiency 0 π/2−BPSK 1/2 4 415 0.08 1 π/2−BPSK 1/2 2 829 0.17 2 π/2−BPSK 1/2 1 1659 0.33 3 π/2−BPSK 5/8 1 2074 0.41 4 π/2−BPSK 3/4 1 2488 0.50 5 π/2−BPSK 13/16 1 2696 0.54 6 QPSK 1/2 1 3318 0.66 *Data rate is calculated assuming 5 carriers with 1GHz bandwidth each, 1/8 CP, and 20% control channel overhead 6 QPSK 1/2 1 3318 0.66 7 QPSK 5/8 1 4147 0.83 8 QPSK 3/4 1 4977 1.00 9 QPSK 13/16 1 5391 1.08 10 16QAM 1/2 1 6636 1.33 11 16QAM 5/8 1 8294 1.66 12 64QAM 1/2 1 9953 1.99 13 64QAM 5/8 1 12442 2.49 14 64QAM 3/4 1 14930 2.99 15 64QAM 13/16 1 16174 3.23 65Copyright © 2011 by the authors. All rights reserved.

67. MMB Beamforming The “most critical” technology for MMB Beamforming Options • Baseband Digital Beamforming • Baseband Analog Beamforming (phase shifters)• Baseband Analog Beamforming (phase shifters) • RF Beamforming (phase shifters) Challenges • Beam pattern/codebook design • Beam training signal/procedure/algorithm design • Beam tracking signal/procedure/algorithm design 67 Copyright by the authors, all rights reserved

68. Array of 2 Point Sources [1/3] θ r cos 2 d θ cos 2 d θ 1 2 2 2 cos cos 2 2 0 0 2 2 cos cos 2 2 02 2 d d j j d d j j E E E E E e E e e e E E π π θ θ λ λ π π θ θ λ λ − − = + = + + = Same amplitude, same phase 20 0 0 2 2 2 cos cos 2 Assuming 2 1 and / 2 cos cos 2 d E E E d E π θ λ λ π θ   =     = =   =     Copyright by the authors, all rights reserved

69. Array of 2 Point Sources [2/3] θ r cos 2 d θ cos 2 d θ 1 2 2 2 cos cos 2 2 0 0 2 2 cos cos 2 2 02 2 d d j j d d j j E E E E E e E e e e E E π π θ θ λ λ π π θ θ λ λ − − = − = − − = Same amplitude, opposite phase 20 0 0 2 2 2 sin cos 2 Assuming 2 1 and / 2 sin cos 2 d E jE jE d E π θ λ λ π θ   =     = =   =     Copyright by the authors, all rights reserved

70. Array of 2 Point Sources [2/3] 1 2 2 2 cos cos 2 2 2 2 0 0 2 2 cos cos 2 2 2 2 02 2 d d j j d d j j E E E E E e E e e e E E π δ π δ θ θ λ λ π δ π δ θ θ λ λ     − + +            − + +        = + = + + = Same amplitude, arbitrary phase difference θ r cos 2 d θ 0 0 0 2 2 2 2 cos cos 2 2 Assuming 2 1 and / 2 cos cos 2 2 E E d E E E d E π δ θ λ λ π δ θ =   = +    = =   = +    Copyright by the authors, all rights reserved cos 2 d θ

71. Dynamic Antenna Array [1/2] ) ) +∆ ) 3 cosθ+∆ ) 4 cosθ+∆ θ ( ) cos d θ+∆ d 2 2 cosj d e π θ λ −2 cosj d e π θ λ − ( )s t( )s t( )s t( )s t 2 2 cosj d e π θ λ + 2 cosj d e π θ λ + ( )s t ∆ (2 cos d θ+∆ (3 cos d θ (4 cos d Copyright by the authors, all rights reserved

72. Dynamic Antenna Array [2/2] A dynamic antenna array can point in any direction to maximize the received signal Enhanced receiver/transmitter antenna gain (reduced PA power, LNA gain) Improved diversity Reduced multi-path fadingReduced multi-path fading Null/ Suppress interfering signals Spatial power combining: •Less power per PA •Simpler PA architecture Copyright by the authors, all rights reserved

73. BB Digital Beamforming ( ) ( )1M s t− ( ) ( )1N r t− ( )1s t 1 t w 1 r w ( )2r t ( )0s t 0 t w 0 r w ( )1r t M Optimal capacity for all channel conditions: min(M,N) data streams Can support a variety of MA schemes (TDMA / OFDMA / SC-FDMA / SDMA) High hardware complexity: M(N) full transceivers High system power consumption ( ) ( )1M − ( ) ( )1N r t− ( )1 t M w − ( )1 r N w − Copyright by the authors, all rights reserved

74. BB Analog Beamforming 1 t w 1 r w( )s t 0 t w 0 r w ( )r t M 1 data stream, antenna weights applied at “analog” baseband Achieves high antenna gain in an arbitrary direction Intermediate hardware complexity: M(N) RF mixers Intermediate power consumption ( )1 t M w − ( )1 r N w − Copyright by the authors, all rights reserved

75. RF Beamforming 1 r w ( )s t 0 r w ( )r t M 1 t w 0 t w Σ 1 data stream, RF phase shifters only, digitally controlled Achieves high antenna gain in an arbitrary direction Low hardware complexity: N RF phase shifters Low system power consumption No (or limited) support simultaneous transmissions to multiple users (e.g. OFDMA / SC-FDMA / SDMA) ( )1 r N w −( )1 t M w − Copyright by the authors, all rights reserved

76. Long Term Fading of mmWave Channel Beamforming is a transmission / reception scheme that can be used to adapts to the long-term statistics of the channel Question to answer: Will the long-term channel fading for mmWave varies much faster than that of cellular waves?mmWave varies much faster than that of cellular waves? Let’s look at the determinants of the long-term fading of a wireless channel • Pathloss • Penetration loss • Absorption due to atmosphere and/or precipitation • Reflection, diffraction, and blocking 76Copyright © 2011 by the authors. All rights reserved.

77. Long Term Fading of mmWave Channel Pathloss • Change depends on change of the location of the MS and the distance travelled – not the carrier frequency Penetration loss • Change depends on change of the location of the MS and blockages in the path travelled – again not depend on the carrier frequency AbsorptionAbsorption • Change depends on change of the location of the MS and distance travelled, not the carrier frequency Reflection, diffraction, blocking • Change depends on change of the location of the MS and the path travelled, not the carrier frequency Observation • The long term fading of mmWave channel induced by communication medium and environment should not be significantly faster than that of the cellular waves 77Copyright © 2011 by the authors. All rights reserved.

78. Beamforming and Channel Fading Beamforming enhances the gain of the strongest paths while suppresses others The good Beamforming increases the sensitivity of channel condition to channel fading The bad 78Copyright © 2011 by the authors. All rights reserved.

79. Beamforming and Channel Fading The spatial selectivity of beamforming • Use horn antenna as an example for analysis • Gain of a horn antenna G ≈ 10log10(4.5AeλAhλ) • 3 dB beamwidth of a horn antenna • θ = 56 / A ,• θ3dB, E-plane = 56 / Aeλ, • θ3dB, H-plane = 67 / Ahλ • Assume θ3dB = θ3dB, E-plane = θ3dB, H-plane • G ≈ 42.27 - 20log10θ3dB Tradeoff between beamforming gain and beam adaptation overhead and robustness of the channel • The larger the Tx/Rx beamforming gain, the narrower the beamwidth • Increased sensitivity to mobility • More frequent beamforming adaptation needed 79Copyright © 2011 by the authors. All rights reserved.

80. Beamforming and Channel Fading For BS Tx/Rx beamforming with < 30 dB antenna gain • θ3dB ≈ 4.1 degree • 7.2 meters wide @ 100 meter distance • It takes 73.7 ms (589 MMB slots) for an MS moving at 350kmph to travel 7.2 meters For MS Tx/Rx beamforming with < 20 dB antenna gainFor MS Tx/Rx beamforming with < 20 dB antenna gain • θ3dB ≈ 13 degree • 22.5 meters wide @ 100 meter distance • It takes 231.1 ms (1849 MMB slots) for an MS moving at 350kmph to travel 22.5 meters Observation • The increased long term channel variation due to Tx/Rx beamforming in MMB is on the order of tens of milliseconds or slower (hundreds of MMB slots or more) 80Copyright © 2011 by the authors. All rights reserved.

82. RF Transceiver Transceiver key RF components • Antenna, Filters, Power Amplifier (PA), Low-Noise Amplifier (LNA), Oscillator (VCO), Mixer and Data converters (DAC/ADC) Copyright by the authors, all rights reserved

83. Nonlinear Device In the most general sense, the output response of a nonlinear circuit can be modeled as a Taylor series in terms of the input signal voltage 2 3 0 0 1 2 3 where the Taylor Coefficients are defined as i i iv a a v a v a v= + + + +L ( )0 0 0 1 2 0 2 2 where the Taylor Coefficients are defined as 0 0 0 and higher order terms ii ii a v dv a vdv d v a vdv = = = = = iv ov Copyright by the authors, all rights reserved

84. Gain Compression Consider the case where a single frequency sinusoid is applied to the input of a nonlinear device such as a power amplifier 0 0 2 2 3 3 0 0 1 0 0 2 0 0 3 0 0 cos cos cos cos iv V t v a aV t a V t a V t ω ω ω ω ω = = + + + + + L 2 0 0 0 1 0 0 2 0 3 3 0 0 3 0 0 3 0 2 3 0 0 2 0 1 0 3 0 0 2 2 0 0 3 1 cos2 cos 2 cos cos21 cos 2 4 1 3 cos 2 4 1 1 cos2 2 4 t v a aV t a V t t a V t a V v a a V aV a V t a V t a ω ω ω ω ω ω ω +  = + + +    +  + +        = + + + +        + L 3 0 0cos3V tω +L 0 3 1 0 3 0 2 1 3 0 0 3 Voltage gain at frequency 3 34 4 is negative in most practical amplifiers aV a V G a a V V a ω + = = + Copyright by the authors, all rights reserved

85. Intermodulation Distortion Consider two-tone input voltage consisting of two closely spaced frequencies 2 1ω ω− 1 22ω ω− 2 12ω ω− 1ω 2ω 12ω 22ω 2 1ω ω+ 13ω 23ω 2 12ω ω+1 22ω ω+ ( ) ( ) ( ) ( ) ( ) ( ) 0 1 2 2 32 3 0 1 0 1 2 2 0 1 2 3 0 1 2 2 2 0 1 0 1 1 0 2 2 0 1 2 0 2 2 2 3 2 0 1 2 2 0 1 2 3 0 1 cos cos cos cos cos cos cos cos 1 1 cos cos 1 cos2 1 cos2 2 2 3 1 cos( ) cos( ) cos cos3 4 4 i o o v V t t v a aV t t a V t t a V t t v a aV t aV t a V t a V t a V t a V t a V t ω ω ω ω ω ω ω ω ω ω ω ω ω ω ω ω ω ω = + = + + + + + + + = + + + + + + + − + + + + L 3 1 3 0 2 2 3 3 3 0 2 1 2 1 2 3 0 1 2 1 2 1 1 2 3 1 cos cos3 4 4 3 3 3 3 3 3 cos cos(2 ) cos(2 ) cos cos(2 ) cos(2 ) 2 4 4 2 4 4 Output spectrum consists of harmonics of the form , t a V t t a V t t t a V t t t m n m n ω ω ω ω ω ω ω ω ω ω ω ω ω ω     + +            + + − + + + + − + + +       + = L 0, 1, 2 3, These combinations of the two input frequencies are called intermodulation products ± ± ± L 2 1 1 22ω ω− 2 12ω ω− 2 1 2 12ω ω+1 22ω ω+ Copyright by the authors, all rights reserved

86. Third-Order Intercept Point (IP3) 1 1 2 2 2 1 0 2 2 2 6 2 3 0 3 0 1 2 1 3 9 2 4 32 These two powers are equal at the third-order IP P a V P a V a V ω ω ω− =   = =    1 0 2 2 2 6 1 3 1 2 3 2 2 1 3 1 3 These two powers are equal at the third-order IP Let input signal voltage at the IP be 1 9 2 32 4 3 21 2 3IP IP IP IP IP V V IP V a V a V a V a a P P a V a ω = = = = = = Copyright by the authors, all rights reserved

87. Mixer IFf LOf RF LO IFf f f= ± RFf LOf IF RF LOf f f= ± IFf LOf LO IFf f+LO IFf f− 0 RF LO IFf f f− = RF LOf f+LOf0 RFf ( ) ( ) ( ) ( ) ( ) ( ) cos2 cos2 cos2 cos2 2 RF LO IF LO IF RF LO IF LO IF v t K v t v t K f t f t K v t f f t f f t π π π π  = =   = − + +  ( ) ( ) ( ) ( ) ( ) ( ) cos2 cos2 cos2 cos2 2 IF LO RF LO RF IF RF LO RF LO v t K v t v t K f t f t K v t f f t f f t π π π π  = =   = − + +  Copyright by the authors, all rights reserved

88. Image Frequency IFf LOf RF LO IFf f f= + 0 IM LO IFf f f= − IFf LOf IM LO IFf f f= + 0 RF LO IFf f f= − 2 IFf 2 IFf -fIF is mathematically identical to fIF because the frequency spectrum of any real signal is symmetric about zero frequency , and thus contains negative as well as positive frequencies A received RF signal at the image frequency fIM is indistinguishable at the IF stage from the desired RF signal of frequency fRF ( ) ( ) IF RF LO LO IF LO IF IF IM LO LO IF LO IF f f f f f f f f f f f f f f = − = + − = = − = − − = − ( ) ( ) IF RF LO LO IF LO IF IF IM LO LO IF LO IF f f f f f f f f f f f f f f = − = − − = − = − = + − = Copyright by the authors, all rights reserved

89. Homodyne (Zero-IF) Receiver ( ) ( ) ( ) 2 2 1 cos2 cos cos cos 2 After low pass filtering 1 2 1 cos2 sin sin sin 2 After low pass filtering 1 RF RF RF RF RF RF RF RF t I t t t t I t t Q t t t t ω ω ω ω ω ω ω ω + = = = = − = = = ( ) 1 2 Q t = Copyright by the authors, all rights reserved Benefits Drawbacks Less hardware LO Leakage Low power consumption DC offset errors No IF stage and hence no image filter I/Q mis-match Flicker (or 1/f) noise

90. Super-heterodyne Receiver LOf IFf Copyright by the authors, all rights reserved Benefits Drawbacks Good sensitivity High Q filter Good selectivity High performance oscillator LNA output impedance matched to 50 ohm is difficult Integration of HF image reject filter is a major problem

91. Wideband-IF Receiver ( )sin IF tω ( )cos IF tω LOω ( )I t ( )Q t ( )'I t ( )cos IF tω ( )Q t ( )'Q t Copyright by the authors, all rights reserved Benefits Drawbacks Image cancellation by IR mixer IR Mixer Image rejection from the RF front-end pre- selection filter Good phase noise performance

92. Wideband-IF Receiver (Image Rejection) ( ) ( ) ( ) ( ) ( ){ } ( ) ( ){ } ( ) ( ) ( ) cos cos cos 1 1 cos cos cos cos 2 2 cos cos sin RF RF IM IM LO RF IF RF LO IM IF IM LO RF RF IM IM LO I t x t x t t x t t x t t Q t x t x t t ω α ω β ω ω α ω ω α ω β ω ω β ω α ω β ω  = − + −     = − + + − + + + + −     = − + −  Low-side injection Signal of interest Image RF LO LO IM IFω ω ω ω ω− = − = ( )cos cos cos sin sinRF RF RF RF RF RFx t x t x tω α α ω α ω− = + ( )cos cos cos sin sinIM IM IM IM IM IMx t x t x tω β β ω β ω− = + ( ) ( ) ( ) ( ) ( ){ } ( ) cos cos sin 1 1 sin sin sin sin 2 2 RF RF IM IM LO RF IF RF LO IM IF IM Q t x t x t t x t t x t ω α ω β ω ω α ω ω α ω β ω  = − + −   = − − + + − + + +  ( ){ } ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) After low-pass filtering 1 1 cos cos , sin sin 2 2 1 1 ' cos sin cos cos 2 2 2 1 1 ' sin cos sin sin 2 2 2 LO RF IF IM IF RF IF IM IF IF IF RF IM IF IF IF RF IM t I t x t x t Q t x t x t I t I t t Q t t x x t Q t I t t Q t t x x ω β ω α ω β ω α ω β ω ω α ω β ω ω α ω  + −     = − + + = − − + +    = − = + + = + = + ( ) ( ) ( ) After low-pass filtering 1 1 ' cos , ' sin 2 2 IF RF RF t I t x Q t x β α α + = = Copyright by the authors, all rights reserved

93. Low-IF Receiver ( )2sin 2 LOf tπ ( )2cos 2 LOf tπ ( )2cos 2 LOf tπ Copyright by the authors, all rights reserved Benefits Drawbacks Potential advantages of both heterodyne and homodyne receivers. ADC dynamic range The IF frequency is just one or two channels bandwith away from DC, which is just enough to overcome DC offset problems. Image reject mixer which is implemented in digital baseband

94. Downlink RF Transceiver Requirement Base station transmitter •Transmit antennas / antenna arrays • 20 – 30 dB antenna gain, horn antennas or phase antenna arrays (64 – 1024 elements) •Power amplifier • 20 – 50 dBm, >20% efficiency, EVM < 5% for OFDM waveform •Packaging • Integrated solution of antenna array / PA / MMIC / RFIC to minimize transmission loss• Integrated solution of antenna array / PA / MMIC / RFIC to minimize transmission loss Mobile station receiver • Receive antenna arrays • 6 – 18 dB antenna gain, phase antenna arrays (4 – 64 elements) • Receiver sensitivity < -80dBm • Total Rx chain Noise Figure < 7dB • Similar solutions exist today! • 60GHz CMOS RFIC with phase antenna array (BWRC) • 60GHz Single-chip integrated antenna and RFIC (GEDC) • Packaging • Integrated solution of antenna array / LNA / MMIC / RFIC to minimize transmission loss 94 Copyright by the authors, all rights reserved

95. Uplink RF Transceiver Requirement Mobile station transmitter • Transmit antenna arrays • 6 – 18 dB antenna gain, phase antenna arrays (4 – 64 elements) • Power amplifier • 20 – 23 dBm, >20% efficiency, EVM < 5% for 16QAM single-carrier waveform • Packaging• Packaging • Integrated solution of antenna array / PA / MMIC / RFIC to minimize transmission loss • Power consumption on the order of 100mW ~ 1W Base station receiver • Receiving antennas / antenna arrays • 20 – 30 dB antenna gain, horn antennas or phase antenna arrays (64 – 1024 elements) • Receiver sensitivity < -95 dBm • Total Rx Noise Figure < 5dB • Packaging • Integrated solution of antenna array / PA / MMIC / RFIC to minimize transmission loss 95 Copyright by the authors, all rights reserved

96. Travelling Wave Tube (TWT) Power Amplifier TWT amplifiers have been extensively used for high power applications at millimeter wave frequencies • Provides KWs to MWs power for satellite and radar • Cost in 10K’s of US$ (too expensive for cellular) Need to consider solid-state amplifier design for MMB Copyright by the authors, all rights reserved

97. Solid-state Power Amplifier Gallium-Nitride based power amplifier •Wide bandgap materials such as gallium nitride (GaN) or silicon carbide (SiC) have much larger bandgaps than conventional semiconductors •Gallium-nitride High Electron Mobility Transistor (GaN HEMT) devices have breakdown voltages 10 times higher than GaAs HEMT devices, allowing GaN HEMT devices to operate with much higher voltages 97 Source: “Gallium Nitride (GaN) Microwave Transistor Technology For Radar Applications”, Microwave Journal, January 2008

98. Solid-state Power Amplifier State-of-the-art for solid-state mmWave PAs •11 Watts at 34 GHz (D. C. Streit, et. al., “The future of compound semiconductors for aerospace and defense applications”, CSIC 2005) •842 mW at 88 GHz (M. Micovic, et. al., “W-Band GaN MMIC with 842mW output power at 88 GHz”, IMS 2010) •5.2 Watts at 95 GHz with a 12-way radial-line combiner (James Schellenberg, et. al., “W-Band, 5W solid- state power amplifier/combiner”, IMS 2010) 98 Source: “Gallium Nitride (GaN) Microwave Transistor Technology For Radar Applications”, Microwave Journal, January 2008

99. Cascaded Constructive Wave Amplifier • Forward wave is amplified as it propagates along the transmission line • Backward wave is attenuated as it propagates • Distribution of N cascaded traveling wave stages • Active devices along the transmission line provide feedback • Relative phase of transmission line and active device determines amplification/ attenuation. Source: J. Buckwalter and J. Kim, ISSCC 2009

100. Low-Noise Amplifier [1/2] Single Stage 60 GHz LNA Source: Javier Alvarado, PhD thesis, May 2008 Gain 12 dB Noise Figure 5 dB over 57 – 64 GHz Power Consumption 4.5mA from a 1.8 V source 1-dB compression point +1.5dBm Efficiency 17.4% Process IBM0.12 μm, 200 GHz fT, SiGe technology.

101. Low-Noise Amplifier [2/2] Two Stage 23–32GHz LNA Source: El-Nozahi et al, IEEE JOURNAL OF SOLID-STATE CIRCUITS, FEB 2010 Gain 12 dB Noise Figure 4.5–6.3dB over 23–32 GHz Power Consumption 13mW from a 1.5 V source IP3 -4.5dBm to -6.3dBm [stage1=-2dBm, stage2=7dBm] Efficiency NA Process Jazz Semiconductor 0.18 m BiCMOS 1 3, 3,1 3,2 1 1 tot G IP IP IP = +

103. MMB downlink budget MMB link downlink budget analysis Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Transmit Power (dBm) 35.00 40.00 35.00 40.00 35.00 40.00 35.00 40.00 Transmit Antenna Gain (dBi) 17.00 17.00 23.00 23.00 17.00 17.00 23.00 23.00 Carrier Frequency (GHz) 28.00 28.00 28.00 28.00 28.00 28.00 28.00 28.00 Distance (km) 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 Key system configuration parameters •Base station Tx power: 35dBm – 40dBm •Base station Tx antenna gain: 17 dB – 23 dB •Mobile station Rx antenna gain: 3 dB – 10 dB Distance (km) 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 Propagation Loss (dB) 115.32 115.32 115.32 115.32 115.32 115.32 115.32 115.32 Other losses 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00 Receive Antenna Gain (dB) 3.00 3.00 3.00 3.00 10.00 10.00 10.00 10.00 Received Power (dBm) -80.32 -75.32 -74.32 -69.32 -73.32 -68.32 -67.32 -62.32 Bandwidth (MHz) 500 500 500 500 500 500 500 500 Thermal Noise PSD (dBm/Hz) -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 Noise Figure 7.00 7.00 7.00 7.00 7.00 7.00 7.00 7.00 Thermal Noise (dBm) -80.01 -80.01 -80.01 -80.01 -80.01 -80.01 -80.01 -80.01 SNR (dB) -0.31 4.69 5.69 10.69 6.69 11.69 12.69 17.69 Implementation loss (dB) 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 Spectram Efficiency 0.37 0.95 1.12 2.23 1.31 2.50 2.78 4.29 Data rate (Mbps) 186.08 474.53 559.37 1117.08 653.70 1250.93 1390.35 2145.23 103Copyright © 2011 by the authors. All rights reserved. Path loss formula: PL = 141.3 + 20log10d with d in km (free-space loss + 20dB)

104. MMB uplink budget MMB uplink budget analysis Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Transmit Power (dBm) 20.00 23.00 20.00 23.00 20.00 23.00 20.00 23.00 Transmit Antenna Gain (dBi) 3.00 3.00 3.00 3.00 10.00 10.00 10.00 10.00 Carrier Frequency (GHz) 28.00 28.00 28.00 28.00 28.00 28.00 28.00 28.00 Distance (km) 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 Key system configuration parameters •Mobile station Tx power: 20dBm – 23dBm •Mobile station Tx antenna gain: 3 dB – 10 dB •Base station Rx antenna gain: 17 dB – 23 dB Distance (km) 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 Propagation Loss (dB) 115.32 115.32 115.32 115.32 115.32 115.32 115.32 115.32 Other losses 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00 Receive Antenna Gain (dB) 17.00 17.00 23.00 23.00 17.00 17.00 23.00 23.00 Received Power (dBm) -95.32 -92.32 -89.32 -86.32 -88.32 -85.32 -82.32 -79.32 Bandwidth (MHz) 50 50 50 50 50 50 50 50 Thermal Noise PSD (dBm/Hz) -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 -174.00 Noise Figure 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 Thermal Noise (dBm) -92.01 -92.01 -92.01 -92.01 -92.01 -92.01 -92.01 -92.01 SNR (dB) -3.31 -0.31 2.69 5.69 3.69 6.69 9.69 12.69 Implementation loss (dB) 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 Spectram Efficiency 0.20 0.37 0.67 1.12 0.80 1.31 1.98 2.78 Data rate (Mbps) 9.92 18.61 33.32 55.94 39.92 65.37 98.96 139.03 104Copyright © 2011 by the authors. All rights reserved. Path loss formula: PL = 141.3 + 20log10d with d in km (free-space loss + 20dB)

105. Link Budget Analysis Summary MMB downlink budget • Low end: 35 dBm Tx power, 17 dB Tx antenna gain, 3 dB Rx antenna gain, 5 dB implementation loss 180 Mbps on 500 MHz bandwidth at 500 meters • High end: 40 dBm Tx power, 23 dB Tx antenna gain, 10 dB Rx antenna gain, 5 dB implementation loss) 2145 Mbps on 500 MHz bandwidth at 500 meters MMB uplink budgetMMB uplink budget • Low end: 20 dBm Tx power, 3 dB Tx antenna gain, 17 dB Rx antenna gain, 5 dB implementation loss 9.92 Mbps on 50 MHz bandwidth at 500 meters • High end: 23 dBm Tx power, 10 dB Tx antenna gain, 23 dB Rx antenna gain, 5 dB implementation loss 139 Mbps on 50 MHz bandwidth at 500 meters Conclusion • Assuming free-space plus 20dB path loss, MMB can provide 100 Mbps ~ 2 Gbps cell-edge throughput on the downlink and 10 Mbps ~ 100 Mbps cell-edge throughput on the uplink at 28 GHz for cell radius of 500 meters. 105Copyright © 2011 by the authors. All rights reserved.

107. System Level Performance 19 cells wrap-around 12 sectors per cell 1 horn antenna per sector 20 dB antenna gain • 17.5o 3-dB beamwidth in azimuth domain • 10o 3-dB beamwidth in elevation domain • 30 dB front-to-back ratio Base station Tx power = 13, 16, 19, 22 dBm/MHz Mobile station uniformly dropped in the coverage area 107Copyright © 2011 by the authors. All rights reserved.

108. Geometry with Single Rx Antenna Site-to-site distance = 500 meters Single Rx antenna with -1 dB antenna gain 0.7 0.8 0.9 1 Interference limited 3dB worse 5%-tile geometry than cellular No Rx beamforming PLF1: PL = 141.3 + 20log10d with d in km 12dB Lognormal shadowing •8dB Lognormal shadowing for cellular 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -12.00 -9.00 -6.00 -3.00 0.00 3.00 6.00 9.00 12.00 15.00 18.00 Cellular,33dBmPerMHz 13dBmPerMHz, PLF1, NoBF 16dBmPerMHz, PLF1, NoBF 19dBmPerMHz, PLF1, NoBF 22dBmPerMHz, PLF1, NoBF 108Copyright © 2011 by the authors. All rights reserved.

109. Geometry with Single Rx Antenna 0.7 0.8 0.9 1 Site-to-site distance = 500 meters Single Rx antenna with -1 dB antenna gain 2 – 6 dB worse 5%-tile geometry than cellular 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -12.00 -9.00 -6.00 -3.00 0.00 3.00 6.00 9.00 12.00 15.00 18.00 Cellular,33dBmPerMHz 13dBmPerMHz, PLF2, NoBF 16dBmPerMHz, PLF2, NoBF 19dBmPerMHz, PLF2, NoBF 22dBmPerMHz, PLF2, NoBF 109Copyright © 2011 by the authors. All rights reserved. No Rx beamforming PLF2: PL = 157.4 + 32log10d with d in km 12dB Lognormal shadowing •8dB Lognormal shadowing for cellular

110. 1 2 3 4 30 60 90 120 150 RxBeam 1 RxBeam 2 RxBeam 3 RxBeam 4 Mobile station Rx beamforming             = += 2 sin 2 sin cos ϕ ϕ θφϕ N AF kd 210 240 270 300 330 180 04-element uniform linear array N=4, d=λ/2, k=2π/λ 4 fixed beams (φ=0, π/2, π, 3π/2) Mobile station orientation is random ~ U[0, 2π) Mobile station selects the beam that maximizes geometry  2 110Copyright © 2011 by the authors. All rights reserved.

111. Geometry with Rx Beamforming 0.7 0.8 0.9 1 Site-to-site distance = 500 meters 4-element antenna array with - 3 dB antenna gain per element Interference limited The same 5%-tile geometry as cellular 0 0.1 0.2 0.3 0.4 0.5 0.6 -12.00 -9.00 -6.00 -3.00 0.00 3.00 6.00 9.00 12.00 15.00 18.00 Cellular,33dBmPerMHz 13dBmPerMHz, PLF1, RxBF 16dBmPerMHz, PLF1, RxBF 19dBmPerMHz, PLF1, RxBF 22dBmPerMHz, PLF1, RxBF 111Copyright © 2011 by the authors. All rights reserved. Rx beamforming PLF1: PL = 141.3 + 20log10d with d in km 12dB Lognormal shadowing •8dB Lognormal shadowing for cellular

112. Geometry with Rx Beamforming 0.7 0.8 0.9 1 Site-to-site distance = 500 meters 4-element antenna array with - 3 dB antenna gain per element 0-3dB worse 5%-tile geometry than cellular 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -12.00 -9.00 -6.00 -3.00 0.00 3.00 6.00 9.00 12.00 15.00 18.00 Cellular,33dBmPerMHz 13dBmPerMHz, PLF2, RxBF 16dBmPerMHz, PLF2, RxBF 19dBmPerMHz, PLF2, RxBF 22dBmPerMHz, PLF2, RxBF 112Copyright © 2011 by the authors. All rights reserved. Rx beamforming PLF1: PL = 157.4 + 32log10d with d in km 12dB Lognormal shadowing •8dB Lognormal shadowing for cellular

113. System Throughput Analysis MMB system performance analysis assumptions Number of cells 19 Number of sectors per cell 12 Site to site distance 500 meters Carrier frequency 28 GHz System bandwidth 500 MHz Path loss model 141.3 + 20log10d, or 157.4 + 20log10d Base station Tx power 40, 43, 46, or 49 dBmBase station Tx power 40, 43, 46, or 49 dBm Base station Tx antenna configuration Single horn antenna Base station Tx antenna gain 20 dB Log-normal shadow fading STD 12 dB Mobile station Rx noise figure 7 dB Mobile station Rx antenna configuration Single antenna, or Rx beamforming with 4-element ULA Mobile station Rx antenna gain -1 dB for single antenna case, -3 dB for ULA case System overhead (cyclic prefix, control channels, etc.) 40% Transceiver implementation loss 3 dB 113Copyright © 2011 by the authors. All rights reserved.

114. System Throughput 4000 4500 5000 5500 6000 Cellthroughput(Mbps) PLF1,NoBF PLF2 provides higher system throughput than PLF1 30x LTE cell throughput (4Tx MIMO per sector, 3-sector cell, 20MHz system bandwidth) 2000 2500 3000 3500 4000 40 43 46 49 Cellthroughput(Mbps) Base station transmit power per sector (dBm) PLF1,NoBF PLF2,NoBF PLF1,RxBF PLF2,RxBF 114Copyright © 2011 by the authors. All rights reserved. RxBF significantly improves system throughput system bandwidth)

115. Cell-edge performance 80 100 120 140 edgethroughput(Mbps) PLF1,NoBF PLF1 allows better cell- edge throughput than PLF2 0 20 40 60 40 43 46 49 cell-edgethroughput(Mbps) Base station transmit power per sector (dBm) PLF1,NoBF PLF2,NoBF PLF1,RxBF PLF2,RxBF 115Copyright © 2011 by the authors. All rights reserved. RxBF significantly improves system throughput

117. Summary Millimeter wave spectrum (3-300GHz) can potentially provide the bandwidth required for mobile broadband applications for the next few decades and beyond. •Opportunity to open 200 times the spectrum currently allocated for cellular below 3GHz. Propagation and other losses due to rain, foliage and penetration through building materials needs better understandingbuilding materials needs better understanding Millimeter waves are also attractive for mobile application due to small component sizes such as antennas. •Further research is needed towards components and devices that meets mobile application demand of higher power and efficiency Wireless community should take on the growing data demand by exploiting millimeter wave spectrum paving the way for multi-Gbps mobile broadband. 117Copyright © 2011 by the authors. All rights reserved.