A quantitative approach to spectrum...
Transcript of A quantitative approach to spectrum...
A quantitative approach to spectrum regulation
Kate Harrison Dissertation talk
May 8, 2015
1
Outline
• Introduction to whitespaces
• Quantifying whitespaces
• Whitespace software
• Whitespace policy
2
Wireless spectrum today
AM RADIO
TV TV FM
TV TV
TV
Amateur
3 kHz
300 kHz
3 MHz
30 MHz
300 MHz
3 GHz
30 GHz
300 kHz
3 MHz
30 MHz
300 MHz
3 GHz
30 GHz
300 GHz
3
Wireless spectrum today
AM RADIO
TV TV FM
TV TV
TV
Amateur
3 kHz
300 kHz
3 MHz
30 MHz
300 MHz
3 GHz
30 GHz
300 kHz
3 MHz
30 MHz
300 MHz
3 GHz
30 GHz
300 GHz
3
Whitespaces correct a mismatch
2011 US spectrum allocation chart
4
Whitespaces correct a mismatch
Free!
Used
Measurement from Berkeley Wireless Research Center
(BWRC) in 2004 (Cabric, et al.)
2011 US spectrum allocation chart
Broadcast TV
70% unused
4
Whitespaces correct a mismatch
Free!
Used
Measurement from Berkeley Wireless Research Center
(BWRC) in 2004 (Cabric, et al.)
2011 US spectrum allocation chart
Broadcast TV
70% unused
“Long-term Spectral Occupancy Findings In
Chicago” (Taher, et al. 2011)
“14% of spectrum in 30-3000 MHz is
occupied”
4
Whitespaces correct a mismatch
Free!
Used
Measurement from Berkeley Wireless Research Center
(BWRC) in 2004 (Cabric, et al.)
2011 US spectrum allocation chart
Broadcast TV
70% unused
“Long-term Spectral Occupancy Findings In
Chicago” (Taher, et al. 2011)
“14% of spectrum in 30-3000 MHz is
occupied”
4
Availability of TV channel 10
Used (TV station)
Free (no TV)
5
Availability of TV channel 10
Used (TV station)
Free (no TV)
We call this whitespace
5
Availability of TV channel 10
Used (TV station)
Free (no TV)
We call this whitespace
Focus on TV whitespaces
5
Current regulations
Plus… • Power limits • Height limits
TV station
TransmittingsecondariesSpatial buffer
Protected TV receivers
2007 2008 2009 2010 2011 2012 2013 2014 2015
Brief history of WS
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
Brief history of WS
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
Firs
t US
devic
es c
ertifi
ed
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
Firs
t US
devic
es c
ertifi
ed
US
PCAS
T re
port
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
Firs
t US
devic
es c
ertifi
ed
US
PCAS
T re
port
Lega
l in
Sing
apor
e
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
Firs
t US
devic
es c
ertifi
ed
US
PCAS
T re
port
Lega
l in
Sing
apor
e
Whitespace trials in Africa, UK (Microsoft, Google)
7
2007 2008 2009 2010 2011 2012 2013 2014 2015
Firs
t Ofc
om c
onsu
ltatio
n
ECC
draf
ts r
ules
Lega
l in
UK
Lega
l in
Cana
da
Lega
l in
US
* *
* Updates to US rules published or proposed
Brief history of WS
Firs
t US
devic
es c
ertifi
ed
US
PCAS
T re
port
Lega
l in
Sing
apor
e
Whitespace trials in Africa, UK (Microsoft, Google)
7
WS access methodsSensing Geolocation + databases
8
WS access methodsSensing Geolocation + databases
• Favored in 2008 regulations
• Conservative threshold
• Shown infeasible in 2009 paper [Mishra, et al.]
• No longer required for all devices in 2010
8
WS access methodsSensing Geolocation + databases
• Favored in 2008 regulations
• Conservative threshold
• Shown infeasible in 2009 paper [Mishra, et al.]
• No longer required for all devices in 2010
• Currently preferred method
• Used for rest of talk
Location
Available channels
Whitespace database
WS device
8
WS access methodsSensing Geolocation + databases
“How much whitespace has the FCC opened up?”
(Mishra, et al. 2009)
8
WS access methodsSensing Geolocation + databases
“How much whitespace has the FCC opened up?”
(Mishra, et al. 2009)
8
Outline
• Introduction to whitespaces
• Quantifying whitespaces
• Whitespace software
• Whitespace policy
9
How much white space is there?
• FCC TV transmitter data • ITU propagation model
“How much white space capacity is there?” (Harrison, Mishra, and Sahai 2009)10
Single-link data rates
• 1 km link • Noise from TV stations only
“How much white space capacity is there?” (Harrison, Mishra, and Sahai 2009)11
Single-link data rates
• 1 km link • Noise from TV stations only
1 Blu-ray disc (40 Mbps)
125 Blu-ray discs
25 Blu-ray discs
“How much white space capacity is there?” (Harrison, Mishra, and Sahai 2009)11
Secondary interference modelAssume nearby cells same size
Freq. reuse-1 (CDMA, OFDMA)
12
Secondary interference modelAssume nearby cells same size Users scattered
randomly
Downlink only (computational
limitations)
Freq. reuse-1 (CDMA, OFDMA)
12
Secondary interference modelAssume nearby cells same size Users scattered
randomly
Downlink only (computational
limitations)
Cell size depends on population Simple economic model: 2000 people per tower
e.g. 10% of people paying $20/month → $50k/year
Freq. reuse-1 (CDMA, OFDMA)
12
Rates with interference
• Cellular model • 2000 people/tower
• Per-tower rates • Rural areas suffer “How much white space capacity is there?”
(Harrison, Mishra, and Sahai 2009)13
Rates with interference
DSL• Cellular model
• 2000 people/tower • Per-tower rates
• Rural areas suffer “How much white space capacity is there?” (Harrison, Mishra, and Sahai 2009)
Blu-ray disc (40 Mbps)
13
A paradigm shift(before)
“Measuring the TV “White Space” Available for Unlicensed Wireless Broadband”
(New America Foundation, 2006)
14
A paradigm shift
“Aggregate interference with FCC and ECC white space usage rules: case study in
Finland” (Jäntti, et al. 2011)
(after)
15
“UHF white space in Europe — A quantitative study into the potential of the 470–790
MHz band”(Beek, et al. 2011)
A paradigm shift
“Aggregate interference with FCC and ECC white space usage rules: case study in
Finland” (Jäntti, et al. 2011)
(after)
15
“UHF white space in Europe — A quantitative study into the potential of the 470–790
MHz band”(Beek, et al. 2011)
A paradigm shift
“Aggregate interference with FCC and ECC white space usage rules: case study in
Finland” (Jäntti, et al. 2011)
Ofcom 2013 consultation
(after)
15
“UHF white space in Europe — A quantitative study into the potential of the 470–790
MHz band”(Beek, et al. 2011)
A paradigm shift
“Aggregate interference with FCC and ECC white space usage rules: case study in
Finland” (Jäntti, et al. 2011)
Ofcom 2013 consultation
(after)
“Opportunities for white space usage in Australia”
Freyens and Loney, 201115
Outline
• Introduction to whitespaces
• Quantifying whitespaces
• Whitespace software
• Whitespace policy
16
Motivating example
Public Policy AM/FM whitespace
+ →
17
Motivating example
Public Policy AM/FM whitespace
+ →
• Easy to use
• Free
• Reliable
• Flexible/extensible
17
Existing tools
18
Existing tools
“Software Tool for Assessing Secondary System
Opportunities in Spectrum Whitespaces”
(WoWMoM 2013)
19
Existing tools
• Many tunable parameters but still targeting TVWS
“Software Tool for Assessing Secondary System
Opportunities in Spectrum Whitespaces”
(WoWMoM 2013)
19
Existing tools
• Many tunable parameters but still targeting TVWS
• Based on Matlab
“Software Tool for Assessing Secondary System
Opportunities in Spectrum Whitespaces”
(WoWMoM 2013)
19
Existing tools
• Many tunable parameters but still targeting TVWS
• Based on Matlab
• Not available
“Software Tool for Assessing Secondary System
Opportunities in Spectrum Whitespaces”
(WoWMoM 2013)
19
Existing tools
• Written in Matlab
• Grew organically
20
WEST (Whitespace Evaluation SofTware)
Written in...
Posted on...
west.kateharrison.net
21
Key features
• Support for other bands
• Free, easier to integrate (e.g. AWS)
• Modular, extensible
• Open-source with GPLv2
22
Use case: explore new regions
23
Use case: explore new rulesets
Industry Canada 2015
24
Use case: explore new rulesets
Industry Canada 2015
24
Use case: explore new rulesets
Canada
25
Use case: explore new rulesets
Canada United States
25
Use case: number of contiguous channels
26
Other uses
• Comments to regulators
• Reproducible research
• New economic models
• Cloud-based for greater impact, engagement
27
DesignDevice: fixed
28
DesignDevice: fixed Region: United States
28
DesignDevice: fixed Region: United States
DataMap2D
28
DesignRegion: United States
DataMap2D
Device: fixed
29
Design
Ruleset: FCC
Region: United States
DataMap2D
Device: fixed
30
Design
Ruleset: FCC
DataMap2D
Channel number: 21
Region: United States
DataMap2D
Device: fixed
31
Design
Whitespace channel count
Whitespace delta map
CCDFs by area, population
Pareto plots 2D histograms
Ruleset: FCC
DataMap2D
Channel number: 21
Region: United States
DataMap2D
Device: fixed
32
Generating a whitespace map
from west.data_management import * from west.data_map import * from west.boundary import BoundaryContinentalUnitedStates, \ BoundaryContinentalUnitedStatesWithStateBoundaries from west.region_united_states import RegionUnitedStates from west.ruleset_fcc2012 import RulesetFcc2012 from west.device import Device !test_device = Device(is_portable=False, haat_meters=30) !datamap_spec = SpecificationDataMap(DataMap2DContinentalUnitedStates, 200, 300) !region_map_spec = SpecificationRegionMap(BoundaryContinentalUnitedStates, datamap_spec) !is_whitespace_map_spec = SpecificationWhitespaceMap(region_map_spec, RegionUnitedStates, RulesetFcc2012, test_device) !is_whitespace_map = is_whitespace_map_spec.fetch_data() !total_whitespace_channels = is_whitespace_map.sum_all_layers() !is_in_region_map = region_map_spec.fetch_data() !plot = total_whitespace_channels.make_map(is_in_region_map=is_in_region_map) !plot.add_boundary_outlines(boundary=BoundaryContinentalUnitedStatesWithStateBoundaries()) plot.set_boundary_color('k') plot.set_boundary_linewidth('1') !plot.save("Number of TVWS channels in the United States.png")33
Outline
• Introduction to whitespaces
• Quantifying whitespaces
• Whitespace software
• Whitespace policy
34
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
“What’s better, whitespaces or refarming?”
35
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
“How can we make devices/deployments
cheaper, easier, and faster?”
“What’s better, whitespaces or refarming?”
35
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
“How can we increase data rates while protecting the primaries?”
“How can we make devices/deployments
cheaper, easier, and faster?”
“What’s better, whitespaces or refarming?”
35
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
“How can we increase data rates while protecting the primaries?”
“How can we make devices/deployments
cheaper, easier, and faster?”
“How can we make WS access more flexible, robust without losing secondary QoS?”
“What’s better, whitespaces or refarming?”
35
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
36
Spectrum sharing scenarios
Single-use band1
37
Spectrum sharing scenarios
Single-use band1
Whitespace (WS)2
37
Spectrum sharing scenarios
Single-use band1
Whitespace (WS)2
Partial clearing + WS3
37
Spectrum sharing scenarios
Single-use band1
Whitespace (WS)2
Efficient clearing + WS4
Partial clearing + WS3
37
Impact on incumbents
Efficient clearing allows more incumbents to stay with same clearing target38
14 channels repurposed
lost with naive clearing
lost with efficient clearing
Impact on incumbents
Efficient clearing allows more incumbents to stay with same clearing target38
Team + tools
Anant Sahai Vidya MuthukumarAngel Daruna Vijay Kamble
PycoSAT (PicoSAT) WEST
39
Whitespace policy
• Spectrum reallocation [ICC 2015]
• Architecture (briefly) [ICC 2015]
• Variable power limits [DySpAN 2011, DySpAN 2012]
• Alternative methods for localization [DySpAN 2014]
40
WS architectureCurrent model
M Master
WSDB Whitespace database
41
WS architectureCurrent model
M Master
WSDB Whitespace database
Device ID, location
41
WS architectureCurrent model
M Master
WSDB Whitespace database
Device ID, location
Operating parameters
41
C Client
Dev
ice
ID
Ope
ratin
g pa
ram
eter
s
WS architecture
M Master
WSDB Whitespace database
Current model
42
M Master
C Client
L Localization
G Gateway
P Packager
Dev
ice
ID
Ope
ratin
g pa
ram
eter
s
WSDB Whitespace database
WS architectureCurrent model
42
C Client
L Localization
G Gateway
P Packager
Dev
ice
ID
Ope
ratin
g pa
ram
eter
s
Localization information
Req. for loc. info.
Access request
Access response
WSDB Whitespace database
WS architectureProposed model
42
Main benefits
43
Main benefits
• Leverages existing infrastructure (e.g. iPhones)
43
Main benefits
• Leverages existing infrastructure (e.g. iPhones)
• Shares infrastructure across whitespaces, reduces burden to O(1) in some cases (scalability)
43
Main benefits
• Leverages existing infrastructure (e.g. iPhones)
• Shares infrastructure across whitespaces, reduces burden to O(1) in some cases (scalability)
• Increases flexibility for manufacturers, deployers (modularity)
43
Main benefits
• Leverages existing infrastructure (e.g. iPhones)
• Shares infrastructure across whitespaces, reduces burden to O(1) in some cases (scalability)
• Increases flexibility for manufacturers, deployers (modularity)
• Upgrade/interchange components without recertification (upgradability)
43
Main benefits
• Leverages existing infrastructure (e.g. iPhones)
• Shares infrastructure across whitespaces, reduces burden to O(1) in some cases (scalability)
• Increases flexibility for manufacturers, deployers (modularity)
• Upgrade/interchange components without recertification (upgradability)
• Can certify using unit tests (testability)
43
Themes
• Data-driven policy
• Reproducible research
• Spatial variation → interesting, important challenges
• Make whitespaces as attractive as possible
• Efficient use of spectrum, enable innovation
44