1. No category

Millimeter Wave Wireless Networks:
Potentials and Challenges
Sundeep Rangan, NYU-Poly
Joint work with Felipe Gomez-Cuba,
Ted Rappaport, Elza Erkip
March 11, 2015
Rutgers Colloquium
1
NYU Wireless
Outline
 Millimeter Wave: A New Frontier for Cellular
 Can it Work? Measurements in NYC
 NYU WIRELESS
 Relaying Revisited
 Other Projects and Future Work
2
NYU Wireless
MmWave: The New Frontier for Cellular
 Massive increase in bandwidth
 Near term opportunities in LMDS and E-Bands
 Up to 200x total over long-time
 Spatial degrees of freedom from large antenna arrays
From Khan, Pi “Millimeter Wave Mobile Broadband:
Unleashing 3-300 GHz spectrum,” 2011
3
NYU Wireless
Commercial 64
antenna element
array
Millimeter Wave Before Cellular
Jagadis Bose at Royal Institution, London 1898
Demonstration of 60 GHz transmission
60 GHz Wireless LAN 802.11ad
Photo from WiLocity
4
NYU Wireless
mmWave backhaul
Image from http://mobilebackhaul.blog.com/
Key Challenges for Mobile Cellular
 All transmissions are directional:
 Friis’ Law:
𝑃𝑃𝑟𝑟
𝑃𝑃𝑡𝑡
=
𝜆𝜆 2
𝐺𝐺𝑡𝑡 𝐺𝐺𝑟𝑟
4𝜋𝜋𝜋𝜋
⇒ Path loss ∝ 𝜆𝜆−2
 Can be overcome with beamforming: 𝐺𝐺𝑡𝑡 , 𝐺𝐺𝑟𝑟 ∝ 𝜆𝜆−2
 But requires directional search, tracking to support mobility
 Shadowing
 Mortar, brick, concrete > 150 dB
 Human body: Up to 35 dB
 NLOS propagation relies on reflections and scattering
5
NYU Wireless
Millimeter Wave Cellular Vision
Uday Mudoi, Electronic Design, 2012
 Small cells
 Directional transmissions
 Relaying / mesh topology
6
NYU Wireless
http://www.miwaves.eu/
Outline
 Millimeter Wave: A New Frontier for Cellular
 Can it Work? Measurements in NYC
 NYU WIRELESS
 Relaying Revisited
 Other Projects and Future Work
7
NYU Wireless
NYC 28 and 73 GHz Measurements
 Focus on urban canyon
environment
 Likely initial use case
 Mostly NLOS
 “Worst-case” setting
 Measurements mimic microcell
type deployment:
 Rooftops 2-5 stories to street-level
 Distances up to 200m
All images here from Rappaport’s measurements:
Azar et al, “28 GHz Propagation Measurements for Outdoor
Cellular Communications Using Steerable Beam
Antennas in New York City,” ICC 2013
8
NYU Wireless
Isotropic Path Loss Models
 Standard linear path loss model
𝑃𝑃𝑃𝑃 𝑑𝑑 = 𝛼𝛼 + 𝛽𝛽 log 𝑑𝑑 + 𝜉𝜉
 Measures total power
 Aggregate across all directions
 Separate LOS and NLOS models
Akdeniz, Liu, Samimi, Sun, Rangan,
Rappaport, Erkip JSAC 2014
9
NYU Wireless
Isotropic Path Loss Comparison
 Isotropic NLOS path loss
measured in NYC
 ~ 20 - 25 dB worse than
20 dB loss
3GPP urban micro model
for fc=2.5 GHz
 But beamforming will
offset this loss.
 Bottom line:
mmW has no effective
increase in path loss
10
NYU Wireless
Hybrid LOS-NLOS-Outage Model
 mmW signals susceptible to
severe shadowing.
 Not incorporated in standard
3GPP models
 New three state link model:
LOS-NLOS-outage
 Form derived from random
shape theory
(Bai, Vaze, Heath 13)
 Outages significant only at
d>150m
 Will help small cell by
reducing interference
11
NYU Wireless
Angular Spread
 Measured powers at different
TX-RX angular pairs.
 Avg. of 2 clusters of paths
detected
 More likely with time
resolution
 Typical beamwidth in each
cluster:
 ~10 deg in AoA RX
RX power at different angles
 ~ 7 deg in AoA TX
 Typical 2-3 spatial DoFs
12
NYU Wireless
Simulations: SNR Distribution
 Simulation assumptions:
 200m ISD
 3-sector hex BS
 20 / 30 dBm DL / UL power
 8x8 antenna at BS
 4x4 (28 GHz), 8x8 (73 GHz) at UE
 A new regime:
 High SNR on many links
 Much better than current
macro-cellular
 Interference is non dominant
13
NYU Wireless
Comparison to Current LTE
 Initial results show significant gain over LTE
 Further gains with spatial mux, subband scheduling and wider bandwidths
System
antenna
mmW
Current
LTE
Duplex
BW
1 GHz
TDD
20+20
MHz
FDD
fc
(GHz)
Antenna
NYU Wireless
Cell edge rate
(Mbps/user, 5%)
DL
UL
DL
UL
28
4x4 UE
8x8 eNB
1514
1468
28.5
19.9
73
8x8 UE
8x8 eNB
1435
1465
24.8
19.8
2.5
(2x2 DL,
2x4 UL)
53.8
47.2
1.80
1.94
10 UEs per cell, ISD=200m,
hex cell layout
LTE capacity estimates from 36.814
14
Cell throughput
(Mbps/cell)
~ 25x gain
~ 10x gain
Outline
 Millimeter Wave: A New Frontier for Cellular
 Can it Work? Measurements in NYC
 NYU WIRELESS
 Relaying Revisited
 Other Projects and Future Work
15
NYU Wireless
NYU WIRELESS
• Exciting new center
 Ted Rappaport, founder
 Faculty across ECE, Courant and Med school
• Pioneering in mmWave for cellular
• First to demonstrate feasibility
 Measurements in NYC
• Research includes:
 Cellular design, capacity studies, prototyping
 Biological impact, medical applications
• Bringing technology to reality
16
NYU Wireless
Industrial Affiliates
NYU WIRELESS is Leading Industry
 14 industrial affiliates
 Vendors, carriers & test
 Leading contributions to industry
 Channel modeling groups
 FCC Notice of Inquiry
 Host: Brooklyn 5G Summit
 New entrants:
 SiBeam: Leader in mmWave RF Ics
 StraightPath: Nationwide holder of
38 GHz spectrum
17
NYU Wireless
Outline
 Millimeter Wave: A New Frontier for Cellular
 Can it Work? Measurements in NYC
 NYU WIRELESS
 Relaying Revisited
 Other Projects and Future Work
18
NYU Wireless
Multihop Relaying for mmWave
 Significant work in multi-hop
transmissions for cellular
 Gains have been minimal
 Why?
 Current cellular systems are
bandwidth-limited
 Millimeter wave may be different
 Overcome outage via macrodiversity
 Many degrees of freedom
19
NYU Wireless
Consider Two Possible Protocols
ISH: Infrastructure single hop
IMH: Infrastructure multi hop
 Direct transmissions to mobiles
 Multi-hop transmissions using UEs
 Similar to current cellular
 Rarely used today
[Gomez-Cuba, Rangan, Erkip, Gonzalez-Castano, ISIT 2014]
20
NYU Wireless
Scaling Laws Analysis
 Randomly drop 𝑛𝑛 mobiles.
 Area, bandwidth, number of antennas
scale with 𝑛𝑛
 Estimate scaling on 𝑅𝑅 𝑛𝑛
1
lim log 𝑅𝑅 𝑛𝑛
𝑛𝑛→∞ n
 Estimate rate to a constant factor.
 Assume EGoS
 Base stations have multiple antennas
 Mobiles have single antenna
21
NYU Wireless
Param
Scaling
Bandwidth
𝑊𝑊 = 𝑊𝑊0 𝑛𝑛𝜓𝜓
Num BS
antenans
Area
Num BS
ℓ = ℓ0 𝑛𝑛𝛾𝛾
𝐴𝐴 = 𝐴𝐴0 𝑛𝑛𝜈𝜈
𝑚𝑚 = 𝑚𝑚0 𝑛𝑛𝛽𝛽
Protocols for Downlink
 Infrastructure Single Hop
 BSs transmit using to UEs directly using MU-MIMO
 Frequency reuse one. Treat out-of-cell interference as noise
 Infrastructure Multi Hop
 Cell is divided into subcells
 Traffic is routed from BS to mobiles via multi-hop
 MU-MIMO on first hop from BS.
 Partial frequency reuse and time-division for half-duplex
constraint and interference
 Sub-cell size is optimized
22
NYU Wireless
Achievable Rates: IMH vs. ISH
Rate per user
Gain with widebands and
large num mobiles / BS.
Param
Scaling
Bandwidth
𝑊𝑊 = 𝑊𝑊0 𝑛𝑛𝜓𝜓
Num BS
antenans
Area
Num BS
Fixed DoFs
(current cellular)
Increasing DoFs
(e.g. mmW)
Path loss
Degs of freedom per BS (bandwidth * antenna)
23
NYU Wireless
ℓ = ℓ0 𝑛𝑛𝛾𝛾
𝐴𝐴 = 𝐴𝐴0 𝑛𝑛𝜈𝜈
𝑚𝑚 = 𝑚𝑚0 𝑛𝑛𝛽𝛽
𝐶𝐶𝐶𝐶−𝛼𝛼
Regimes
 Interference vs power limit regime:
 When DoFs per BS is large, rate scaling ceases to increase
 There is a limit to very large DoFs
 Multi-hop transmissions can better exploit high DoFs
 Shortens range ⇒ reduces power limit
 But, requires large scaling on number of mobiles / base station
24
NYU Wireless
What is the Right Protocol for 5G?
 Area constant. Increase mobile density
Technology
Scaling
Protocol
No action
BSs density fixed
DoF per BS is fixed
ISH or IMH Decreases with mobiles per cell
Densification
Increase BS per
mobile
DoF per BS is fixed
ISH or IMH Improves due to increased BS
density
Massive MIMO BS per mobile fixed
or mmWave
Increase DoF per BS
Rate per user
ISH
Increases until the SNR to furthest
mobile hits threshold
IMH
Increases until the SNR to first
mobile hits threshold
 IMH is not needed for densification
 But, IMH is required for Massive MIMO or mmWave
25
NYU Wireless
Link Layer Capacity
 Assume point-to-point MIMO link has capacity scaling:
Θ(𝑊𝑊 min ℓ𝑟𝑟 , ℓ𝑡𝑡 ) 𝑊𝑊 ≤ 𝑊𝑊 ∗
𝑅𝑅 = �
ℓ𝑟𝑟
𝑃𝑃
Θ
𝑊𝑊 ≥ 𝑊𝑊 ∗
min(ℓ𝑟𝑟 ,ℓ𝑡𝑡 ) 𝑁𝑁𝐼𝐼
 Critical bandwidth
𝑊𝑊 ∗
Bandwidth limited
Power limited
ℓ𝑟𝑟 𝑃𝑃
=Θ
𝑊𝑊𝑁𝑁𝐼𝐼 min(ℓ𝑟𝑟 , ℓ𝑡𝑡 )
 Applies to coherent and non-coherent channel
 For non-coherent, number of channel parameters can grow linearly
with bandwidth
 Assumes CSIT at transmitter
 Channel estimation and feedback overhead is a constant factor
26
NYU Wireless
Estimating the Network Capacity
 Assume path loss model 𝑃𝑃𝑟𝑟 = 𝐶𝐶𝐶𝐶𝑡𝑡 𝑑𝑑 −𝛼𝛼 , 𝛼𝛼 > 2
 Assume MU-MIMO can use full spatial DoFs
 Treat interference as noise
 Use worst case distances
 For IMH, use partial frequency reuse to separate neighbors in
adjacent sub-cells (most a constant capacity loss)
 For IMH, optimize subcell size / number of hops
 Ensure every sub-cell has at least one mobile
27
NYU Wireless
Can We Do Better than IMH?
 IMH is capacity achieving.
 Use a cut-set argument
Rate per user
 Hierarchical cooperation
not necessary
 Difference with
infrastructure vs ad hoc
 IRH: Infra relay hop
Fixed DoFs
(current cellular)
28
NYU Wireless
 Use if IMH is not practical
Increasing DoFs
(e.g. mmW)
 Multihop from fixed relays
not mobiles
Making Multihop Work
Interference-to-noise
 Many issues
 Network discovery
 Beamforming
synchronization &
tracking
 Directional isolation
 Dynamic duplexing
29
NYU Wireless
Qualcomm FlashLinQ
frame structure
Significant Potential Gains
Num
RNs
Capacity
Cell edge
DL
UL
DL
UL
0
2090
1890
10.0
3.40
2
2370
2280
28.2
5.99
4
2440
2330
238.6
244.5
 Significant gains possible
 Esp. cell edge
 Cell capacity limited by single stream
10 UEs per cell,
4x4 antenna array,
single stream
 Dynamic duplexing
 Multi-hop optimal scheduling
Garcia-Rois, Gomez-Cuba, Akdeniz, Gonzalez-Castano, Burguillo-Rial, Rangan,
Lorenzo, IEEE TWC, in revision
30
NYU Wireless
Outline
 Millimeter Wave: A New Frontier for Cellular
 Can it Work? Measurements in NYC
 NYU WIRELESS
 Relaying Revisited
 Other Projects and Future Work
31
NYU Wireless
Channel Measurements
Outage
 Current measurements:
 Outage from a single base station
 Next steps:
 Joint probability multiple base stations
 Changes over time, distance
 Combine with ray tracing?
 Needed to assess:
 Macro-diversity, handover
32
NYU Wireless
Channel Measurements
Deployment Models
 Current measurements:
 Rooftop antenna
 Traditional microcell
 Wide coverage, but NLOS
 Next set of measurements:
 “Street furniture”
 Lower range, but greater LOS paths
 More likely deployment model
33
NYU Wireless
Low-Power Beamforming Architectures
 A/D present major power bottleneck
 Wide bandwidths, large number of antennas
 Current solutions use analog BF via phase shifters
 Hybrid BF
 Many research questions
 Low-rate fully digital architectures?
 Synchronization, channel estimation?
 Implications for multiple access, cell search?
34
NYU Wireless
Directional Cell Search and Adaptation
 Synchronization may be key limiting factor
 Rapid changes from blockage
 Obstructions by buildings, hands, …
 Other challenges:
 Demands for very low latency
 Multi-flow connectivity
 Must discover many elements
35
NYU Wireless
Spatial Channel Estimation
 Estimating directions of arrival is essential for:
 Tracking, synchronization, …
 Analog BF, can “look” in only direction at a time
 Spatial covariance via non-negative matrix completion
min � 𝑝𝑝 𝑧𝑧𝑖𝑖 𝑤𝑤𝑖𝑖∗ 𝑄𝑄𝑤𝑤𝑖𝑖 + 𝜆𝜆𝜆𝜆𝜆𝜆 𝑄𝑄 ,
𝑄𝑄>0
𝑖𝑖
𝑄𝑄 = 𝐸𝐸 𝐻𝐻𝐻𝐻∗
Beamforming with a 8x8 array
Multipath NLOS channel
NYC measurements
[Amir-Eliasi, Rangan, Rappaport 2015]
36
NYU Wireless
Cell Search with Low-Rate
Digital Architectures
 Significant gains for low-rate digital architectures
 Enable searching in multiple directions
 Also beneficial for control messaging / multiple access
Detection in initial cell search
Digital BF vs. analog BF
12 dB gain
37
NYU Wireless
Barati, Hosseini, Rangan, SPAWC 2014
Other MAC Layer Work
 How to design LTE for mmW?
 Key concerns
 Modifications for directionality
 Intermittency in links
 Areas of focus for next year:
 Cell search, synchronization
 Dynamic duplexing
 Multi-flow, carrier aggregation
38
NYU Wireless
Qualcomm FlashLinQ
frame structure
Adaptive and MultiPath TCP
 Rapid changes in rate:
 Cells intermittently blocked.
 Current TCP is slow to adapt
UE
Gateway
 New research directions
 New stochastic TCP mechanisms
 Multi-path congestion control
 Tingting Lu, Subramanian, Panwar
39
NYU Wireless
Heterogeneous Networks
 Many aspects of heterogeneity
 4G and 5G
 Indoor / outdoor
 Third party ISP
vs Cellular operator
 Many questions:
 New spectrum license models?
 Load balancing?
 Pricing?
40
NYU Wireless
Prototyping
 This year:
 Demonstrated mmW
LTE-like signal
 Next steps:
 Integrated NI platform
with ns3 for current LTE
 Demonstrated at EuCNC,
June 2014
 Will modify for mmW
41
NYU Wireless
A Big Question
What is the killer app for mmWave?
 What applications can drive huge amounts of data?
 Many applications for human interaction are limited.
 Video < 20 Mbps. Much lower on mobile devices.
 Will data be driven by machine to machine?
 Many users bursty vs. few users continuous?
 What will drive very low latency (e.g. ~1ms)?
 What will be the network delays?
 What is the partition of mobile vs. cloud?
 Where will data be located?
 What about power, form factor, cost?
42
NYU Wireless
Summary
 MmWave offers tremendous potential
 A once in a generation technological advance
 But, many new challenges
 New regime where degrees of freedom are plentiful
 Dominant challenge is synchronization & intermittency
 Capacity tied closely with front-end capabilities
 Cellular will be re-designed
 Many opportunities for research, commercialization & theory
43
NYU Wireless
Thanks
 Faculty:
 Ted Rappaport, Elza Erkip, Shiv Panwar, Pei Liu
 Michele Zorzi (U Padova)
 Postdoc: Marco Mezzavilla
 Students:
 Felipe Gomez Cuba (U Vigo)
 Mustafa Riza Akdeniz, Parisa Amir Eliasi, Russell Ford,
Yuanpeng Liu, George McCartney, Oner Orhan, Matthew
Samimi, Shu Sun
44
NYU Wireless
References
 Khan, Pi, “Millimeter-wave Mobile Broadband (MMB): Unleashing 3-300GHz
Spectrum,” Feb 2011, http://www.ieee-wcnc.org/2011/tut/t1.pdf
 Rappaport et al. "Millimeter wave mobile communications for 5G cellular: It will
work!." Access, IEEE 1 (2013): 335-349.
 Rangan, Rappaport, Erkip, “Millimeter Wave Cellular Systems: Potentials and
Challenges”, Proc. IEEE, April 2014
 Akdeniz, Liu, Rangan, Rappaport, Erkip, “Millimeter Wave Channel Modeling and
Cellular Capacity Evaluation”, JSAC 2014
 Gomez, Rangan, Erkip, “Scaling Laws for Infrastructure Single and Multihop Wireless
Networks in Wideband Regimes”, ISIT 2014, http://arxiv.org/abs/1404.7022
 Barati, Hosseini, Rangan, et al, “Directional Cell Search for Millimeter Wave Cellular
Systems” http://arxiv.org/abs/1404.5068
 Eliasi, Rangan, and Rappaport. "Low-Rank Spatial Channel Estimation for Millimeter
Wave Cellular Systems." http://arxiv.org/abs/1410.4831
45
NYU Wireless