Characterizing mmWave Channels Greg VanWiggeren, Ph.D. Hongwei Kong - PowerPoint PPT Presentation

Characterizing mmWave Channels Greg VanWiggeren, Ph.D. Hongwei Kong Zhu Wen for 5G Keysight Laboratories Nov 16 th , 2015

Outline 11/16/2015 IEEE 5G Summit Page 2 – Why mmWaves? – Properties of mmW channels – Channel sounding techniques – Early experimental results from Keysight Labs – Summary

Historical Context 5G vision represents revolutionary change 1000X 100X Data Rates Capacity 100X Densification 1ms Latency 5G 3G 4G 1G 2G Reliability 99.999% 100X First Mobile Analog Digital Mobile Ubiquitous Energy Efficiency Voice Broadband Voice Internet Connectivity 11/16/2015 IEEE 5G Summit Page 3

Achieving the 5G Vision More capacity is needed • ~ 700 MHz total BW 100X • < 3 GHz Data Rates 1000X Capacity 0 20 40 60 80 100X Frequency (GHz) Densification Worldwide Data Usage 1ms Latency Exabytes / Month 30 Reliability 20 99.999% 10 100X Energy Efficiency 0 2014 2015 2016 2017 2018 2019 Source: Cisco VNI Mobile, 2015 11/16/2015 IEEE 5G Summit Page 4

Achieving the 5G Vision More capacity is needed • ~ 700 MHz total BW 100X • < 3 GHz Data Rates 1000X Capacity 0 20 40 60 80 100X Frequency (GHz) Densification 1ms Latency “FCC rakes in $45 billion from Reliability wireless spectrum auction…” 99.999% Headline from cnet.com, January 2015 100X Energy Efficiency 11/16/2015 IEEE 5G Summit Page 5

mmWaves for 5G A key enabler of the 5G vision 57-66 64-71 71-76 81-86 37,39 100X 28 Data Rates 1000X Capacity 100X Densification 0 20 40 60 80 Frequency (GHz) 1ms Latency mmWave band properties: Reliability 99.999% • Wider bandwidths • Higher path losses 100X • Different channel properties Energy Efficiency 11/16/2015 IEEE 5G Summit Page 6

Outline 11/16/2015 IEEE 5G Summit Page 7 – Why mmWaves? – Properties of mmW channels – Channel sounding techniques – Early experimental results from Keysight Labs – Summary

Challenges with mmW Channels Higher Path Loss Report ITU-R M.2376-0 (06/2015) For a dense network, atmospheric absorption is a minor issue 11/16/2015 IEEE 5G Summit Page 8

Challenges with mmW Channels Primary Source of Higher Path Loss From the Friis equation: 𝑄𝑝𝑥𝑓𝑠𝑆𝑦 = 𝑄𝑝𝑥𝑓𝑠𝑈𝑦 + 𝐻𝑏𝑗𝑜𝑈𝑦 + 𝐻𝑏𝑗𝑜𝑆𝑦 − 20log 𝑒 2 − 20log 𝑔 2 Frequency • Higher frequencies  smaller antenna elements • Higher directivity overcomes path losses • Massive MIMO enabled in small size • Some challenges • Discovery and tracking • Added complexity 11/16/2015 IEEE 5G Summit Page 9

mmW Channels Dramatically different than channels below 6 GHz mmW Channels: Industry needs for mmW models: • • Higher path losses Indoor and outdoor • • Less diffraction Urban canyons, stadium, shopping mall • Objects are more reflective • MIMO (array) behavior • More Doppler • Consistency with < 6 GHz models • Greater penetration losses to allow system performance (indoor/outdoor) comparisons 11/16/2015 IEEE 5G Summit Page 10

Outline 11/16/2015 IEEE 5G Summit Page 11 – Why mmWaves? – Properties of mmW channels – Channel sounding techniques – Early experimental results from Keysight Labs – Summary

Basic Channel Sounding h ( t )   y t h t x t ( ) ( ) ( ) 11/16/2015 IEEE 5G Summit Page 12

MIMO Channel Sounding Conceptually similar but more subscripts     y j t x i t   h ij t h ( t )       y t h ( t ) x t j ij i   y ( t ) h ( t ) x ( t ) MIMO mmW Channel Model 11/16/2015 IEEE 5G Summit Page 13

Channel Sounding Options Three valid approaches Techniques Frequency Swept Sliding Correlator Wideband Correlation • Low speed • Low speed • Fast measurements measurement measurement • Amp./phase • Challenging for widely • Some temporal information supports separated Tx and Rx aspects not captured AoD, AoA meas. RF Channel x ( t ) y ( t ) h ( t ) 11/16/2015 IEEE 5G Summit Page 14

Channel Sounding Options Three valid approaches Techniques Frequency Swept Sliding Correlator Wideband Correlation • Low speed • Low speed • Fast measurements measurement measurement • Amp./phase • Challenging for widely • Some temporal information supports separated Tx and Rx aspects not captured AoD, AoA meas. y ( t ) x ( t ) RF Channel h ( t ) Sliding correlator receiver T. S. Rappaport et al. , “Millimeter wave mobile communications for 5G cellular : It will work!” IEEE Access , May 2013. 11/16/2015 IEEE 5G Summit Page 15

Channel Sounding Options Three valid approaches Techniques Frequency Swept Sliding Correlator Wideband Correlation • Low speed • Low speed • Fast measurements measurement measurement • Amp./phase • Challenging for widely • Some temporal information supports separated Tx and Rx aspects not captured AoD, AoA meas. x ( t ) y ( t ) RF Channel h ( t )   y ( t ) h ( t ) x ( t )       y ( t ) x ( t ) h ( t ) x ( t ) x ( t )    y ( t ) x ( t ) h ( t ) 11/16/2015 IEEE 5G Summit Page 16

Techniques for angular channel characterization Rotating horn antennas n 2 measurements Horn Antenna Horn Antenna 1 st Line-of-Site Path 𝜄 𝐵𝑝𝐸 𝜄 𝐵𝑝𝐵 Rx Tx 2 nd Path 11/16/2015 IEEE 5G Summit Page 17

Techniques for angular channel characterization MIMO Sounding 1 measurements MIMO Array MIMO array 1 st Line-of-Site Path 𝜄 𝐵𝑝𝐸 𝜄 𝐵𝑝𝐵 Rx Tx 2 nd Path 11/16/2015 IEEE 5G Summit Page 18

Outline 11/16/2015 IEEE 5G Summit Page 19 – Why mmWaves? – Properties of mmW channels – Channel sounding techniques – Early experimental results from Keysight Labs – Summary

Keysight MIMO mmW Sounding Architecture Preferred approach   h ij t     y j t x i t RF Channel Wideband Wideband Multichannel     Tx   RX y t h ( t ) x t j ij i MIMO mmW Channel Model 11/16/2015 IEEE 5G Summit Page 20

MIMO mmW Channel Sounder Up to 44 GHz ANT 85332B 50GHz LNA ANT SP4T Switch PA E8257D 67GHz LO RF Channel BPF M9362 50 GHz 4CH Digital Coherent Down Convertor L4450A Digital I/O Switch Controller E8267D mmWave Signal Source M9703A 8CH Digitizer M8190A 5-GHZ AWG Rubidium Clock Rubidium 33512 Clock HJ5418B AWG 11/16/2015 IEEE 5G Summit Page 21

MIMO mmW Channel Sounder Calibration and synchronization Antenna Calibration ANT 85332B 50GHz LNA ANT SP4T Switch PA E8257D 67GHz LO RF Channel BPF M9362 50 GHz 4CH Digital Coherent Down Convertor L4450A Digital I/O Switch Controller E8267D mmWave Signal Source M9703A 8CH Digitizer M8190A 5-GHZ AWG Rubidium Clock Rubidium 33512 Clock HJ5418B AWG Synchronize 11/16/2015 IEEE 5G Summit Page 22

MIMO mmW Sounding Signal Autocorrelation Experimental vs. simulation Sounding Sequence Autocorrelation Sounding Sequence Autocorrelation Sounding Sequence Autocorrelation (Meas.) 0 0 0 -10 Amplitude (dB) Amplitude (dB) -20 -20 -100 -30 Amplitude (dB) -40 -40 -200 -50 -60 -60 -300 -70 -80 -80 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Simulation Measured by Channel Sample -6 x 10 Sounder RX (M9703A) 11/16/2015 IEEE 5G Summit Page 23

MIMO mmW Sounder Data Handling Broadband channel sounding leads to huge data sets MIMO mmW Channel Model M9703A: 4 Channels @ 3.2 GS/s = 3.2 GS/s * 2B/S*4 = 25.6 GB/s Very difficult to store or stream out in real-time!!! = 200 Gb/s 11/16/2015 IEEE 5G Summit Page 24

MIMO mmW Sounder Data Handling Broadband channel sounding leads to huge data sets Fortunately, the M9703A has four FPGAs for streaming data reduction DSP includes: ADC FPGA ADC • Digital down-conversion to I/Q ADC FPGA • Real-time autocorrelation ADC • Data-reduced complex impulse response ADC FPGA ADC ADC FPGA 25.6 GB/s  1.6 GB/s ADC ( e.g. for 5-us delay spread ) Customer-programmable FPGA “sandboxes” for Real-time streaming to custom DSP storage now feasible. 11/16/2015 IEEE 5G Summit Page 25

MIMO mmW Channel Modeling Parameter extraction using Keysight SystemVue MIMO mmW Channel Model Channel Sounding Model Data • Angle of Arrival/Departure I/Q data from • Power Delay Profile the M9703A Parameter extraction • Rician K factor using well-known • Doppler Shift SAGE algorithm 11/16/2015 IEEE 5G Summit Page 26

MIMO mmW Channel Modeling Parameter extraction using Keysight SystemVue MIMO mmW Channel Model Channel Sounding Model Data • Angle of Arrival/Departure I/Q data from • Power Delay Profile the M9703A Parameter extraction • Rician K factor using well-known • Doppler Shift SAGE algorithm 11/16/2015 IEEE 5G Summit Page 27

MIMO mmW System Simulation Enabled by SystemVue and channel measurements Channel Sounding Model Data I/Q data from • Angle of Arrival/Departure the M9703A Parameter extraction • Power Delay Profile using well-known • Rician K factor SAGE algorithm • Doppler Shift SystemVue Simulation Flow Channel I/Q TX Channel RX EVM & Waveform System Model System BER Model 11/16/2015 IEEE 5G Summit Page 28