November 2008
doc.: IEEE 802.11-08/1306r0
Designing for Coexistence
Date: 2008-11-11
Authors:
Name
Affiliations
Address
Phone
email
John A. Stine
Self
9322 Eagle Court
Manassas Park, VA
703-983-6281 [email protected]
John Stine is employed by The MITRE Corporation but represents himself in this presentation.
The MITRE Corporation is a not for profit company and has no economic interest in the
outcome of the 802 standards process. The author's affiliation with The MITRE Corporation is
provided for identification purposes only, and is not intended to convey or imply MITRE's
concurrence with, or support for, the positions, opinions or viewpoints expressed by the author.
Submission
Slide 1
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Patent Statement
• Methods described in this presentation are covered in
claims in patents and patents pending.
• The MITRE Corporation is a not for profit company
that does not own the patents and has no economic
stake in the outcome of the 802 standards activity
Submission
Slide 2
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Abstract
• The VHT 60 PAR scope includes the intent of
providing mechanisms for coexistent use of spectrum
• This presentation provides a taxonomy of spectrum
sharing techniques based on design intent
• Provides an overview of a contention-based technique
to arbitrate use of spectrum shared by multiple
technologies
• This same technique provides solutions for many hard
problems we want to solve in the VHT 60 standard
Submission
Slide 3
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Spectrum Compatibility - 1
• Many terms cover part of the problem space
– Electromagnetic compatibility (EMC) – “The condition that prevails when
telecommunications equipment is performing its individually designed
function in a common electromagnetic environment without causing or
suffering unacceptable degradation due to unintentional electromagnetic
interference (EMI) to or from other equipment in the same environment.”
(NTIA Red Book)
– Coexistence – “The ability of one system to perform a task in a given shared
environment where other systems have an ability to perform their tasks and
may or may not be using the same set of rules.” (IEEE 802.15.2)
– Cooperation – Differentiated from coexistence by whether a protocol is used
to arbitrate sharing of spectrum (Peha 2009)
– Dynamic Spectrum Access (DSA) – A variety of technologies that allow
different systems to dynamically access spectrum based on its availability.
One of those technologies is cognitive radio (CR).
Submission
Slide 4
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Spectrum Compatibility - 2
• To bring order to the design space I propose a taxonomy
based on design intent
– Reactive design – design efforts in reaction to the unintentional
interference that occurs when systems operate in proximity to each
other (The traditional EMC efforts)
– Non-cooperative design – design without overt consideration of what
other systems the DSA system must be compatible
– Cooperative design – design with a deliberate effort to make systems
compatible with known peer or incumbent systems
Submission
Slide 5
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Taxonomy-1
Non-cooperative design
Technologies
Signal
space
Cooperative design
Strategies
Power Antennas
Temporal
Targeted cooperation
Spatial
Policy driven
approaches
Design
Redesign
Dynamic Static Collaborative Independent
Design to rules
Compliant
Managed
Collaborative
• In non-cooperative design
– The technologies branch are static techniques to at least mitigate and
hopefully prevent interference
– The strategies branch are dynamic techniques that usually depend on
some form of sensing to inform use decision
• Where you would find most cognitive radio and policy driven technologies
Submission
Slide 6
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Taxonomy-2
Non-cooperative design
Technologies
Signal
space
Power Antennas
Cooperative design
Strategies
Temporal
Spatial
Policy driven
approaches
Targeted cooperation
Design
Redesign
Design to rules
Compliant
Dynamic Static Collaborative Independent
Managed
Collaborative
• Targeted cooperation is where most coexistence work lies
– In the design branch new systems are designed to coexist with an incumbent
• The dynamic branch encompasses CR where design is specified by policy
– In the redesign branch existing systems are modified to coexist
• Progress is constrained by “tyranny of the incumbent”
Submission
Slide 7
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Taxonomy-3
Non-cooperative design
Technologies
Signal
space
Power Antennas
Cooperative design
Strategies
Temporal
Spatial
Policy driven
approaches
Targeted cooperation
Design
Redesign
Dynamic Static Collaborative Independent
Design to rules
Compliant
Managed
Collaborative
• In design to rules, the rules of sharing are established first
– The collaboration branch encompasses systems designed together
– The managed branch encompasses technical methods that support a
third party manager that manages which systems may use spectrum
– The compliant branch encompasses approaches where rules are written
and then any technologies that can follow the rules are allowed to use
the spectrum
Submission
Slide 8
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Challenges in Creating Rules for Compliant Design-1
• Likely to be a contention-based design
– Non-exclusive licensing rules of the 3650-3700 band are an example
• Prefer techniques that allow different technologies to play together
– “restricted” protocols which are only capable of avoiding interference with other
co-frequency devices using the same protocol, can only use the lower 25 MHz
– “unrestricted” protocols that prevent interference among different contention
technologies may use the whole band
• What are the alternative “unrestricted” protocols
– Listen-before-talk
– ?
Submission
Slide 9
John Doe, Some Company
November 2008
doc.: IEEE 802.11-08/1306r0
Challenges in Creating Rules for Compliant Design-2
• What’s wrong with listen-before-talk
– Suffers from all the same problems as carrier sensing protocols
•
•
•
•
Hidden terminals
Exposed terminals
Deafness
Muteness
– Still suffers “tragedy of the commons”
• Can I beat the other guy’s backoff scheme
– Returns the sharing problem to “tyranny of the incumbent”
• You have to be able to sense my signal and I don’t have to sense yours
• You better not send any signals I interpret incorrectly
Returns the compatibility problem to targeted cooperation
Submission
Slide 10
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Creating Rules
• The focus of creating rules should be the design of a
generic mechanism that
– Arbitrates the fair use of time, space, and frequency
– Uses the simplest possible signals for detection and arbitration
– Allows open ended design of all other aspects of the system in the
arbitrated periods of spectrum use
• New technologies should only have to play by the rules
to participate
We propose that synchronous collision resolution is such a mechanism
Submission
Slide 11
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Rule Design
• Recommended starting point
The
Original
Sins
– Defining channels (mechanisms should allow the
arbitration and use of contiguous channels)
– Division of time into periods (aka slots) common for
all channels
• Recommended approach
– Use Synchronous Collision Resolution
• Given the starting point, can simultaneously arbitrate
the use of time, space, and channel
• Absolutely fair unless you design it to be unfair for
purposes of service differentiation
Submission
Slide 12
Slide 12
John Doe, Some Company
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Characteristics of SCR used for
Spectrum Arbitration
CR
Signaling
Transmission Slot
…
•
•
•
•
Time slotted channels with common time boundaries
Nodes with packets to send contend in every slot
Signaling is used to arbitrate contention
A paradigm not
Unique signals assigned to each channel
a specific
(e.g. tones)
design
Submission
Slide 13
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Purpose of Collision Resolution Signaling
Assertion signals
Signaling slots
Signaling phases
...
1 2 3 4 5 6 7 8 9
CR
Signaling
...
Transmission Slot
Signaling Process
Red
Rednodes
nodesare
arecontenders
contenders
•
Red
Rednodes
nodesare
arewinners
winners
Prune the set of contenders to a subset which can transmit without colliding
Submission
Slide 14
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Collision Resolution Signaling Example - 1
Red = contender
Gray = non-contender
All contending nodes do a random number draw
and those beneath a specified threshold transmit a
signal. Signalers and those that do not hear the
signal survive this phase of the signaling
In this example all nodes start off as contenders
Submission
Slide 15
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Collision Resolution Signaling Example - 2
Signaling and attrition proceeds for several iterations
with the threshold for signaling changing for each phase
Submission
Slide 16
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Collision Resolution Signaling Example - 3
Submission
Slide 17
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Collision Resolution Signaling Example - 4
Submission
Slide 18
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Collision Resolution Signaling Example - 5
• The end result of collision
resolution signaling
– When all nodes are in range of
each other – one surviving
node
– In a multihop environment as
shown – a set of surviving
nodes separated by the range
of their signals
• The range of signaling’s effect
can be extended by using
echoing (See subsequent
slides)
Demonstration
Submission
Slide 19
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
How effective is CRS in resolving contention ?
7 slots
8 slots
9 slots
1
1
0.95
kt = 500
0.995
P5 k 1  0
P(One Survivor)
kt = 50
kt = 200
6 slots
P(One Survivor)
P4 k 1  0
5 slots
kt = 1000
P k2 1  0
P6 k 1  0 0.9
Q k2 1  0
P7 k 1  0
0.99
U k2 1  0
0.85
4 slots
P8 k 1  0
Sk2 1  0
0.985
P9 k 1  0 0.8
0.75
0
10
20
Number
30
40
50
0.98
0
200
of kContenders
4, 5, 6 , 7, 8, and 9 single-slot phase designs
optimized for a 50 contender density
400
600
800
1000
Number ofk2Contenders
Comparison of 9 single-slot phase designs optimized
for various target densities of contenders
• It is a function of design, # of signaling phases, threshold
probabilities for signaling
• We have a simple design methodology that yields the performance
illustrated
> 99% of the transmissions slots can be resolved to one transmitter
for all practical densities of contenders!
Submission
Slide 20
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Signal Echoing
• One hop signaling may result in deadlock
– Two survivors repeatedly win the contention but attempt to send to
the same destination thus blocking each other
– Most likely to occur in lightly loaded and less dense networks
• Signal echoing will break this deadlock
The contenders will not
interact with each other
since they are out of
range of each other
Submission
Slide 21
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Signal Echoing
phase
C E
With echoing a
contender signals as
before but we add a slot
to a phase for an echo
phase
C E
Nodes that hear the
contender’s signal echo
it allowing a two hop
effect
The blocking terminal
defers from gaining
access
Submission
Slide 22
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Echoing Example
Red = contender
Gray = non-contender
Blue square = echoer
19 contenders after echoing
75 contenders after contention
Demonstration
Submission
Slide 23
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Spatial Reuse-1
0.25
Fraction of Survivors
Survivor Density, SA
1.6
1.4
1.2
1.0
0.8
0.6
0.4
4 Slots
5 Slots
6 Slots
7 Slots
8 Slots
9 Slots
0.2
0.15
0.1
0.05
0.2
Contender Density,  A
Simulated survivor densities using a 9-phase CRS design,
kt = 50
2.00
1.75
25
1.50
20
1.25
15
1.00
10
0.75
8
0.50
5
0.25
2
0.00
0
0.0
Fraction of Range
Density of range to the nearest surviving neighbor
when the average contending neighbor density is10
• Simulations of signaling without echoes reveal
– The density of survivors levels off at about 1.4 survivors per signaling
area (the area covered by the range of a signal)
– Depending on signaling effectiveness, survivors are separated by at
least the range of their signals
Submission
Slide 24
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Spatial Reuse-2
0.9
0.12
0.7
Fraction of Survivors
0.6
0.5
0.4
0.3
0.2
0.1
0
2
5
8
10
15
20
25
25
0.1
2
0.08
20
10
15
5
0.06
8
0.04
0.02
Contender Density,
4.00
A
3.50
25
3.00
20
2.50
15
2.00
10
1.50
8
1.00
5
0.50
0
2
0.00
Survivor Density, SA
0.8
Fraction of Range
Simulated survivor densities using SUMA version of
signaling
Density of range to the nearest surviving neighbor
using SUMA version of signaling
• Simulations of signaling with echoes reveal
– The density of survivors decreases with contender density
– Average separation range increases with the density of the contenders
Submission
Slide 25
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Arbitrating Channels among Technologies
Channels systems
want
Signaling Schedule
Channels systems
receive
Station of system A
Station of system B
Station of system C
The signals in this
scenario
• Stations may contend for multiple channels
– Signals contain the tones of the channels a station wants to use
– A station may win the right to use a subset of the channels it
initially contends to use
Submission
Slide 26
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Features Useful to the VHT 60 Goals
• Synchronous Collision Resolution has additional features
– Creates the conditions for effective CDMA use and other
channelization schemes
– Enables multiple antenna adaptation schemes
– Differentiates prioritization of access among nodes
– Supports resource reservations without scheduling
– Provides multiple mechanisms to enhance energy conservation
SCR creates the access
conditions that allow most
PHY technologies to perform
at their best and will enable the
very high throughput sought
Submission
Slide 27
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
The Hurdles in Designing the Rules
• Methods to synchronize different technologies
• Agreement on
–
–
–
–
–
Submission
The features to include in the signaling
Precedence in arbitration
Boundaries on slots and channels
The common signals
The synchronization bounds
Slide 28
The
Original
Sins
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
Conclusion
• Listen before talk methods of sharing suffer
shortcomings and do not solve the contentious issues in
spectrum sharing across technologies
• We have proposed a method of sharing that
– Does not suffer listen-before-talk’s failure modes
– Enables sharing across all RF spectrum’s dimensions
– Avoids the “tyranny of the incumbent” problem
• This technique can also serve as a very effective
contention-based access mechanism for high
throughput applications
Submission
Slide 29
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
References
•
•
•
•
•
•
•
•
•
J. M. Peha, “Sharing Spectrum through Spectrum Policy Reform and Cognitive Radio,”
TBP Proc. of the IEEE, 2009.
J. A. Stine, “Enabling secondary spectrum markets using ad hoc and mesh networking
protocols,” Academy Publisher J. of Commun., Vol. 1, No. 1, April 2006, pp. 26 - 37.
J. Stine, G. de Veciana, K. Grace, and R. Durst, “Orchestrating spatial reuse in wireless
ad hoc networks using Synchronous Collision Resolution,” J. of Interconnection
Networks, Vol. 3 No. 3 & 4, Sep. and Dec. 2002, pp. 167 – 195.
J.A. Stine and G. de Veciana, “A paradigm for quality of service in wireless ad hoc
networks using synchronous signaling and node states,” IEEE J. Selected Areas of
Communications, Sep 2004.
J. A. Stine and G. de Veciana, “A comprehensive energy conservation solution for
mobile ad hoc networks,” IEEE Int. Communication Conf., 2002, pp. 3341 - 3345.
K. Grace, “”SUMA – The synchronous unscheduled multiple access protocol for
mobile ad hoc networks,” IEEE ICCCN, 2002.
J. A. Stine, “Exploiting processing gain in wireless ad hoc networks using synchronous
collision resolution medium access control schemes,” Proc. IEEE WCNC, Mar 2005.
J.A. Stine, “Cooperative contention-based MAC protocols and smart antennas in
Mobile Ad Hoc Networks,” Chapter 8 in Distributed Antenna Systems: Open
Architecture for Future Wireless Communications, Auerbach Publications, Editors H.
Hu, Y. Zhang, and J. Luo. 2007.
J. A. Stine, “Exploiting smart antennas in wireless mesh networks,” IEEE Wireless
Comm Mag. Apr 2006.
Submission
Slide 30
John A. Stine, Self
March 2008
doc.: IEEE 802.11-08/1306r0
Backup
Submission
Slide 31
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
How well does signaling isolate just one survivor?
Signaling slots
Signaling phases
...
1 2 3 4 5 6 7 8 9
...
• Consider a signaling design where all phases have one slot
• Let px be the probability that a contending node will signal in phase x
• A transition matrix may be populated where the element k,s corresponds to the
probability that s of k contending nodes survive the signaling phase
k s
 k  x s
x
p
1

p

    
s
 

k
k
x
Pk,s   p x    1  p x 


0

Submission
Slide 32
0sk
0sk
otherwise .
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
How well does signaling isolate just one survivor? (2)
• The transition matrix of the signaling process with n phases may be calculated
Qn   x1 P x
n
• The probability that just 1 of k contending nodes survives signaling is
Q nk,1
• It is easy to optimally select a set of probabilities that maximizes the probability that
there will be 1 survivor when there are some k = k1 contenders at the beginning but
this problem formulation may result in a lower probability that one survivor remains
when there are k < k1 contenders.
P(one survivor)
k
k1
Improvement at k1 may results in
decreased performance at k < k1
Submission
Slide 33
John A. Stine, Self
November 2008
doc.: IEEE 802.11-08/1306r0
How well does signaling isolate just one survivor? (3)
• A redefined optimization problem
– Let qn be the set of px for an n phase CRS design
– Let kt be a target density of contending nodes
– Let m be the total number of signaling slots allowed (in this case n = m)
– Let S(qn,kt,m) be the probability that there will be only one surviving contender
max
S  q n ,kt ,m 
n
q
7 slots
8 slots
s.t. S  q n ,k ,m   S  q n ,kt ,m 
9 slots
1
1
0.95
kt = 500
0.995
P5 k 1  0
P(One Survivor)
kt = 50
kt = 200
6 slots
P(One Survivor)
P4 k 1  0
k ,0  k  kt .
5 slots
kt = 1000
P k2 1  0
P6 k 1  0 0.9
Q k2 1  0
P7 k 1  0
0.99
U k2 1  0
0.85
4 slots
P8 k 1  0
Sk2 1  0
0.985
P9 k 1  0 0.8
0.75
0
10
20
30
40
50
Number of kContenders
4, 5, 6 , 7, 8, and 9 single-slot phase designs
optimized for a 50 contender density
Submission
0.98
0
200
400
600
800
1000
Number ofk2Contenders
Comparison of 9 single-slot phase designs optimized
for various target densities of contenders
Slide 34
John A. Stine, Self