Millimeter Waves Millimeter Waves 1 GHz 10 GHz 100 GHz 1 THz 10 THz 100 THz 1PHz 30 GHz 300 GHz Frequency Wavelength Microwave mm- Wave THz Far IR Infrared Light UV 10 cm 1 cm 1 mm 100 µm 10 µm 1 µm Page 1
Millimeter Wave Frequencies (mmwave) 1 GHz 10 GHz 100 GHz 1 THz 10 THz 100 THz 1PHz 30 GHz 300 GHz Frequency Wavelength Microwave mm- Wave THz Far IR Infrared Light UV 10 cm 1 cm 1 mm 100 µm 10 µm 1 µm Potential 5G mmwave Bands 28-50 GHz multiple bands under consideration for 5G Radio Access 60 GHz 60 GHz (1) (Oxygen Absorption Band) Unlicensed, 802.11ad, Backhaul 70/80 GHz E-Band, Lightly Licensed. Primarily Backhaul, Fronthaul, Possible radio access band Note 1: 15dB/kM = 1.5dB/100m. Atmospheric absorption is not the main issue for mmwave Radio Access! Page 2
mmwave Technology Not completely new! HP Journal, 1991 HP Journal, 1988 Page 3
Increasing Frequencies: Challenge and Opportunity Free-space Path Loss 20 4-20 Frequency In words. For a given distance, as the frequency increases, the received power will drop unless offset by an increase in some combination of transmit power, transmit antenna gain, and receive antenna gain. The decrease in power as a function of frequency is caused by the decrease in the antenna aperture. Distance The Good News: Higher frequency antennas elements are smaller Easier to assemble into electronically steered arrays Reduced interference. Energy goes where it s needed Improve performance in dense crowds (5G goal) Higher frequencies wider bandwidths: faster (5G goal) Challenges: Increased complexity with more elements Multiple antenna arrays required for spherical coverage Discovery and Tracking (mobile devices) IBM 94 GHz Array Can Tile for Larger Arrays IBM Press Release, June 2013 Page 4
mmwave Radio Access Technology The future is already here it's just not very evenly distributed -- William Gibson WirelessHD has been available to consumers for several years 802.11ad ASIC s are available now and shipping in quantity Peraso Wilocity Silicon Image (SiBeam) The question is: How 5G will coexist with, adopt, and extend these technologies? Note: The 802.11AD standard will be 7 years old in 2020 Page 5
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Page 6
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Double the bandwidth, Double the noise Power. Drives requirements for antenna gain and receiver noise figure Integrated TX noise contributes to higher EVM Page 7
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Limits number of TX/RX channels, probably can t do full beamforming Page 8
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Connectors expensive, impractical Antenna In Package Difficult to route mmwave signals off-chip to separate antenna. Page 9
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Example: Station discovery requires search protocols plus very robust modulation and coding to establish communications with poorly aimed, wide beam (low gain), or omnidirectional antennas. MAC/PHY design needs to support wide range of device capabilities (small devices can t have large arrays) Page 10
Mobility and the Challenge of Directional Antennas Seach Strategies High Gain Large Volume to search Low Probability of both stations pointing in the same direction Low-Gain: Higher Probability of looking in the right direction, but much less energy to detect Connected High Gain Tracking Requires multiple antennas for coverage Page 11
Robust PHY Requirements Portable Devices don t have enough space for multiple high-gain steerable arrays (e.g. left, right, top bottom) Negative SNR s likely, especially during station discovery phase Low-Density Parity Check Codes with Spread Spectrum were used for the control channel in 802.11ad Though unrecognizable, the 802.11ad constellation shown here was decoded without LDPC codeword errors (as indicated by the green p s above). Page 12
mmwave Antenna Development and Validation Antenna Performance - Steerable: design, characterization (codebook), producibility - Beam forming: Reciprocity, Gain/Phase/TDD elements Traceable Measurements (affected by antenna pattern) - Measurements are challenging without facility for conducted measurements - Integrated Power (gain changes with pattern) - Receiver Sensitivity: - Spurious signals are also directional (possibly different directions) Interoperability with steerable antennas (MAC and PHY) Test Modes - Start or Stop steering/beamforming - Select test pattern - Use test-only DUT configurations to simplify parametric measurements, and to aid in isolating performance issues of individual antenna subsystems or elements. Page 13
mmwave PHY Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters Example: Adaptive Channel Estimation algorithms that work at current cellular symbol rates may be to expensive, or impossible to implement at GHz rates. 802.11ad makes extensive use of Golay sequences because they re efficient Page 14
Complementary Golay Codes Used extensively in 802.11ad; Synchronization and AGC Data Spreading Channel Estimation Gain and phase tracking Receive side - fast Golay correlator OR Ga Gb Important attributes of Golay codes are; Low side lobes and low DC content under π/2 rotation. Sum of Ga and Gb autocorrelations is perfect. Ga and Gb autocorrelations can be performed in parallel using a single correlator. Page 15
Golay Correlator Output for 802.11ad Preamble Synchronization Channel Estimation Gu 512 Ga 128 Ga 128 Ga 128 Ga 128 -Ga 128 -Gb 128 -Ga 128 Gb 128 -Ga 128 128 Gv 512 -Gb 128 Ga 128 -Gb 128 128 -Ga 128 -Gb 128 Page 16
Principle of Channel Estimation (application of Golay codes) Ga Gb Ga + Gb + = a a h( t) R ( a) = a a h( t) b b h( t) ( ) = b b h( t) R b Σ output = R ( a) + R( b) = a a h( t) + b b h( t) ( ) δ ( t) h( t) = a a + b b h( t) = = h( t) Page 17
mmwave Design Challenges High Frequency High Bandwidth High Path Loss High Data Rate Phase Stability Amplifier Efficiency High IF Converters (use 2 nd Nyquist) I and Q channel match over frequency Directional Antennas Usually Required Large codebook space for Beam Steering Output Power Integrated Noise Power Beam forming complexity Antenna Complexity IF/RF Flatness Robust Modulation and Coding (MCS) Quadrature Errors (Homodyne) A/D and D/A Converters (power consumption) Discovery and Tracking affect MAC and MCS Power consumption Algorithm Complexity Prototyping (FPGA s usually not fast enough) IO (memory, interfaces to CPU s etc) High sample-rate data to/from converters PHY Implementation Challenges Page 18
PHY Implementation Challenges Practical Considerations No Connector at RF Test Modes (for beam forming) Antenna Polarization Availability of Test Equipment Design Challenges Phase stability / frequency accuracy Quadrature errors DC/LO feedthrough Frequency Dependent I / Q Mismatch Transmit power Antenna Pattern and Performance Efficiency and Cost DAC Baseband ASIC LPF 0 90 RF ASIC with antenna array bonded directly on top of RFIC. DAC LPF Page 19
Early 802.11ad Prototype (single carrier 16 QAM) Measured Results 8 degrees of quadrature error 22.8% EVM Page 20
Early 802.11ad Prototype 60 Phase Error -60 Measured Results 28% EVM 4 Quadrature Error Page 21
Summary 4G to 5G: Evolution and Revolution New <6GHz Waveforms Heterogeneous Networks Network Densification Evolutionary 5G mmwave for Access Steerable Antenna Arrays Centralized & Nomadic RAN Applications Revolutionary Business Model Monetize Transparent Universal Access Models Supporting IoT RAN-Sharing Page 22
QUESTIONS? Copyright 2014 Agilent. Page 23
Summary (5G is not 4G++) There are many unanswered questions regarding massive MIMO (how many antennas, amplifier performance requirements, calibration requirements, suitable frequencies) MIMO Emulation systems will be expensive to build. Need to answer some questions with simulation and accurate channel models. 60 GHz mmwave access technology has been developed, but is still very new. Need to leverage what s been learned for 5G Emerging technologies can t be evaluated without the proper measurement tools - - tools which may not exist today. Agilent is proud to have been the principal supplier of mmwave tools critical to development of both WiHD and WiGig / 802.11ad. We look forward to working with you on 5G Page 24