January 2015
doc.: IEEE 802.19-15/0007r0
Comments on LAA EVM
Date: 2015-01-14
Authors:
Name
Company Address
Alireza Babaei CableLabs
858 Coal Creek Cir
Louisville, CO 80027
USA
Phone
email
+1-3036613405
a.babaei@cablelabs.com
Notice: This document has been prepared to assist IEEE 802.19. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in
this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.
Submission
Slide 1
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Abstract
• This document provides recommendations related to
Annex A (Evaluation methodology) of the 3GPP draft
study item document on Licensed-Assisted Access to
Unlicensed Spectrum: 3GPP TR 36.889 v0.1.0 (2014-11).
• The purpose of this contribution is to include these
recommendations in an upcoming liaison from 802 to
3GPP on its ongoing LAA study item
Submission
Slide 2
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Summary of Recommendations
• Incorporate Truncated Exponential Backoff in LBT
requirements and simulate its effect.
• For a complete understanding of LAA impact on Wi-Fi,
a range of load densities should be simulated.
• Include VoIP and other traffic types as mandatory
traffic models and evaluate corresponding performance
metrics.
• Consider 256 QAM, LDPC and RTS/CTS mandatory
for simulation.
Submission
Slide 3
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
LBT
• Listen-Before-Talk (LBT) based on European regulations (ETSI
301 893 v1.7.1) is mainly considered as a coexistence mechanism
in 3GPP contributions.
• LBT, as defined in the above ETSI document, has two forms :
– Load Based and Frame Based Equipment (LBE and FBE)
– 3GPP contributions indicate majority opinion is tending towards LBE
• Concerns:
– Fixed linear backoff window for extended CCA (q is fixed for a given device)
– No increase with collision (q is fixed during operation)
CC
A
CC
A
Extended CCA
20 µs
20 µs
Channel Busy
Extended CCA = N * 20 µs
N ∈ {0, 1, … , q} and q ∈ {4, 5, … , 32}
4
Submission
Slide 4
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
LBT Simulation Results
•
Assume Load Based Equipment
LBT, q = 32
•
Tx Success Rate: probability of
any client sending successful
burst
•
Steps to Tx Success:
Transmission Success Rate Gain/Loss
1.68X
1.53X
1X
0.23X
– Client must pass CCA
– Client must Tx without collision
•
Results relative to the Tx Success
Rate when both operators use
Wi-Fi
– 1X applies to any client count
Wi-Fi
(both Operators)
0.12X
Operator A: Wi-Fi
Operator B: LAA-LBT
15 devices/operator
Operator A: Wi-Fi
Operator B: LAA-LBT
20 devices/operator
Simulation Assumptions: Full buffer traffic, all nodes within CCA range of each
other, single 20 MHz channel, co-channel operation, clients and APs/eNBs are
stationary, Wi-Fi STAs use EDCA, BE AC (AIFS = 3, CWmin = 15)
Conclusion: Version 1.7.1 of ESTI LBT rules are not sufficient for fair
Coexistence of LTE and Wi-Fi
5
Submission
Slide 5
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
What version of LBT to use?
• 802.11 has proven success in enabling sharing between independent
systems in a wide variety of use cases an load scenarios.
• The basis of this success is EDCA which is based on CSMA/CA;
CSMA/CA is an LBT mechanism using truncated exponential backoff.
• The use of a similar mechanism by LAA should provide the basis for
fair sharing between LAA and Wi-Fi (and also between LAA and
LAA)
– The Wi-Fi Alliance made a proposal to ETSI BRAN that defines LBT with exponential
backoff for LAA (See [3])
Recommendation: Incorporate Truncated Exponential Backoff in LBT
requirements and simulate its effect
Submission
Slide 6
John Doe, Some Company
January 2015
doc.: IEEE 802.19-15/0007r0
System Load
• Section A.1.1 lists the parameters for indoor LAA
coexistence evaluation.
• Only 10 LAA UEs or Wi-Fi clients is assumed per
unlicensed band carrier for DL-only LAA coexistence
evaluation.
• The impact of LAA (using EU LBT) on Wi-Fi clients is
more evident at high system load particularly when
number of nodes is large
– See the backup slides.
Recommendation: For a complete understanding of LAA impact on Wi-Fi,
a range of load densities should be simulated
Submission
Slide 7
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Traffic model
• Wi-Fi and LAA have to operate in unlicensed spectrum
carrying a variety of traffic types including voice,
video, FTP, etc.
• However, the simulations evaluating the fairness of
LAA with Wi-Fi is currently limited to FTP.
• Once initial simulations showing basic fairness between
Wi-Fi and LAA are complete, further simulations will
be required to ensure fairness and functionality are
maintained using a variety of traffic types in a diversity
of use and load scenarios.
Recommendation: Include VoIP and other traffic types as a mandatory
traffic models and evaluate corresponding performance metrics.
Submission
Slide 8
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Assumptions on Wi-Fi Parameters
• Section A.2.1 lists Wi-Fi system evaluation
assumptions.
• 256 QAM and LDPC are considered optional.
– Use of lower order modulation (when SINR is sufficient for 256
QAM) means unnecessarily longer frame duration. Longer frame
duration increases the backoff period and decreases the channel
utilization for other Wi-Fi clients.
• RTS/CTS is considered optional.
– The hidden node behavior of two wireless systems is key to
coexistence. RTS/CTS is optional but commonly used in congested
environments.
Recommendation: Consider 256 QAM, LDPC and RTS/CTS mandatory
for simulation
Submission
Slide 9
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
References
[1] Study on Licensed-Assisted Access to Unlicensed
Spectrum, 3GPP TR 36.889 v0.1.0 (2014-11)
[2] 802.19-14/0082r0: Overview of EU LBT and its
Effectiveness for Coexistence of LAA LTE and Wi-Fi
[3] ETSI BRAN(14)000102
Submission
Slide 10
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Backup
Submission
Slide 11
Alireza Babaei, CableLabs
January 2015
doc.: IEEE 802.19-15/0007r0
Simulation Parameters
•
•
Overall Simulation Assumptions:
– All nodes are within CCA range of each other
– Single channel access scenario
– Full buffer traffic
– 802.11 EDCA channel access modeled
– Best Effort access category (AIFSN = 3, CWmin = 15)
– Stationary nodes
– Monte Carlo simulations with 1,000,000 TXOPs
evaluated per data point
ETSI LBT Assumptions:
– Always use extended CCA
– LBE rules Implemented by LAA nodes
Simple Co-Channel Two Network Topology
STA/UE
Operator A
AP/eNBs
Operator A&B
STA/UE
Operator A
12
Submission
Slide 12
STA/UE
Operator B
STA/UE
Operator B
Alireza Babaei, CableLabs
CCA
Range
Boundary
January 2015
doc.: IEEE 802.19-15/0007r0
Successful Packet Transmission
• ETSI LBT performs poorly
compared to Wi-Fi channel
access
–
Lower q value = worse performance
• Poor coexistence behavior
when sharing and >5
clients/operator
–
–
LAA clients get significantly higher
P(success) compared to Wi-Fi clients i.e.
black lines are all above the red lines.
At a certain client count, LAA clients nearly
starve Wi-Fi clients i.e. P(success)
approaches zero
• This value varies with q, best case
shown
Probability of Successful Channel Access
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
2
13
Submission
Wi−Fi Only
LAA + Wi−Fi, Wi−Fi Nodes, q=32
LAA + Wi−Fi, LAA Nodes, q=32
Slide 13
4
6
8
10 12 14 16 18
Per Operator Client Count (Total is 2x)
Alireza Babaei, CableLabs
20
January 2015
doc.: IEEE 802.19-15/0007r0
Collisions in the Shared Channel
1
• ETSI LBT performs poorly
compared to Wi-Fi channel
access
• Inefficient channel usage for low
q and/or scenarios with ~25 or
more transmitting nodes.
• Attributed to non-increasing
backoff range
14
Submission
0.8
0.7
Collision Probability
– Collision rate increases rapidly to
1 for low q values
– Even for q = 32, collision rate
dominates after 12 transmitting
clients per operator
0.9
0.6
0.5
0.4
0.3
0.2
0.1
0
0
Slide 14
Wi−Fi Only
LAA + Wi−Fi, q=8
LAA + Wi−Fi, q=32
5
10
15
Per Operator Client Count (Total is 2x)
Alireza Babaei, CableLabs
20