Download Aug 2009 - IEEE Mentor

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

IEEE 802.11 wikipedia , lookup

Transcript
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Stray and Overlapping STAs
Date: 2009-08-17
Authors:
Name
Affiliations
Address
Phone
email
Graham Smith
DSP Group
2491 Sunrise Blvd,
#100,
Rancho Cordova,
CA 95742
916 851 9191
X209
[email protected]
John Janecek
DSP Group
2491 Sunrise Blvd,
#100,
Rancho Cordova,
CA 95742
916 851 9191
X208
[email protected]
Submission
Slide 1
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Abstract
The objective is to first estimate the practical problem of
STAs that overlap with another network when the APs
do not overlap.
Then to look at the effects – what happens
Then look to see if any solutions are required, and if so,
what are they.
Make proposals, if any
Submission
Slide 2
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
The Perceived Problem
Case1
QAP
A
QAP
B
STA within range of both
APs but APs are not
overlapping
Case2
QAP
A
QAP
B
STAs within range of both
networks but APs are not
overlapping
Looks obvious in these diagrams, but what about in practice?
Submission
Slide 3
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Propagation Formula
Indoor propagation loss formula (11n) *,
F in MHz, d in feet
For d<16.5ft
Lp = – 38 + 20 log F + 20 log d + Wall/Floor loss(Free Space formula)
For d>16.5ft
Lp = – 38 + 20 log F + 20 log 16.5 + 35 log (d/16.5) + Wall/Floor Loss
Std. Dev 3-4dB (Shadow Loss)
*Erceg et al (2004) as per 11n, Channel Model B – Residential
• 10dB Outer Wall loss has been used in calculations
• No internal wall or floor losses used in calculations
•AP Antenna Gain 2dB
•STA Antenna Gain 0dB
Submission
Slide 4
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Terraced/Town Houses
Terraced Houses @20ft
-45dBm
X
-52dBm
-72dBm
150ft
-88dBm
-102dBm
X
3 Neighbors
Worse Case position of STAs
60 ft, 4 walls = -100dBm HIDDEN
X
-92dBm
150 ft, 2 walls = -96dBm HIDDEN
Neighbors 3 houses down, and opposite houses within 150 feet have potential to overlap
Note: No internal wall losses, external wall loss only.
STAs within range of each other, or another STA, when APs are not? – NO
Submission
Slide 5
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Detached Houses
15ft
X
-53dBm
-70dBm
X
-100dBm
-125dBm
20ft
Worse Case position of STAs
50 ft, 4 walls = -100dBm (no internal walls)
HIDDEN
5ft
X
-53dBm
20ft
X
-65dBm -95dBm
-121dBm
Worse Case position of STAs
30 ft, 4 walls = -92dBm
Add loss with one internal wall in middle house
= -95dBm HIDDEN
House opposite scenario as per previous slide – STAs are HIDDEN
STAs within range of each other, or another STA when APs are not? – NO
NOTE: 09/0474r0 by Alex Ashley (NDS) also predicts similar results
(Simulation with 5 STAs per house)
Submission
Slide 6
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Apartment Block
-45dBm
-52dBm
-72dBm
-88dBm
-44dBm
-63dBm
-84dBm
-99dBm
-61dBm
-77dBm
-97dBm
-77dBm
-91dBm
-105dBm
-92dBm
-103dBm
-102dBm
Compared to the AP signal strengths, no
internal walls, STA signal strength is:
• 4dB less due to antennas (-4)
• 20ft less due to position in x axis (+2)
• 10ft less due to position in z axis (+4dB)
• i.e. either same as for AP or 2dB less
6
Each Apartment
20 x 35 feet
about 700 square feet
RESULT, -2dB cf AP in x-axis
0dB cf AP in z-axis
STAs within range of each other, when APs are not? – NO
Submission
Slide 7
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Summary for domestic scenarios
• In practice for residential scenarios, the overlapping
STA is very rare if at all
– The outer wall attenuation dominates
• Now let us consider when the STA is outside of the
building:
– What is the effect on the two (non-overlapping) networks?
– What is the effect on the STA?
– Will this be a permanent situation?
– How common?
Submission
Slide 8
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Case 1 – Stray STA
x
A
x
x
Channel selection ensures that B is the only
other AP on same channel as A
X
x
A and B are on same channel
Stray STA (X) can see and be seen by B
STA to B ~-81dBm (see below)
STA to A ~-64dBm
x
B
x
Submission
Slide 9
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
What Happens – Case 1: Stray STA
Case 1 – STA within range of both APs, (in the garden)
– Possibly both ‘hidden’ and ‘seen’ nodes in network B
• ‘Hidden’ nodes in network B have no effect
• Worse case is STA ‘sees’ all other STAs and AP in B
• TX
STA sees communication (maybe just the preambles) in networks B
plus its own network A (“Captive Effect”), resulting in reduced ability
to gain the air (depends upon traffic), so STA can only use the “left
over bandwidth”
– Result is that only the STA is affected, not network B or the rest of A
• RX
At best “Step-up Re-start” will result in reception for the STA from its
own AP not being affected. (Network A signal strength is ~17dB
higher than any transmission from network B)
At worse, reception is noisy, resulting in dropped packets
– SBA limits retries
• Bottom line – STA is affected, mostly for TX, but not the networks
Submission
Slide 10
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Extreme Example of Stray STA
Apartment Block around swimming pool
Outside STA ‘sees’ multiple networks
What Happens?
• STA is effectively silenced or reduced
TX due to ‘captive’ effect
• RX also impeded due to numerous
‘hidden’ nodes
X
• Overall effect is poor performance for
an outside STA. No worse, no better
due to OBSS .
• Only “solution” is STA’s network needs
to be on own, unique channel
or else every other Network reduces it
traffic – I don’t think so.
Submission
Slide 11
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Case 2 – Two Overlapping STAs
x
A
x
X
x
Networks A and B are hidden from each other
STAs see each other at ~-78dBm (up to -92dBm)
STAs see own QAPs at ~-64dBm
X
Note: Distance from STA to QAP B is >200ft
for B to be hidden from STA
x
x
B
x
What Happens?
• TX - STAs slightly reduced in ability to gain air,
only by the traffic on the other STA
• RX – Step-up re-Start should enable good reception from AP
Reception, at worse, only prohibited if two STAs transmitting
at same time and stepping on each other. Effect depends on the
relative traffic.
• Basically Networks not affected , just the STAs (mostly TX). Not as much as
‘problem’ as Stray STAs.
Submission
Slide 12
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Stray STAs – Solutions
Notes:
1.
Presence of Stray STAs is a dynamic, relatively short-lived problem
2.
Stray STA could use CTS to self to improve its TX performance – this impinges on Network B
3.
AP decision to re-scan (as below) should also depend upon the QLoad of the Stray STA
Possible “solutions”
1. 08/1260r1 outlined method for a QAP to determine if there was an
overlapping STA - Uses Beacon Report and continues channel search is
Stray STA present
•
Requires the STA to be 11k compliant
2. AP B recognizes a ‘stray STA’, initiates a new Channel Search
•
•
Switch over after ~200ms, switch <5ms (assuming network is 11h)
Not a great incentive for B to do this, but the stray does present a possible
hindrance.
3. Stray STA informs own AP that it is a stray
•
•
Use an “unsolicited” 11h Measurement Report to inform AP. AP can then
initiate new channel search. Three spare bits are available.
This also works for the overlapping STA situation
Note: These could result in several tries for a new channel… but so what? If no
better channel found, OK, not made worse. If changing several times, no
real new problem, channel change is not that a big deal.
Submission
Slide 13
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Stray and Overlapping STAs
Do we need to adjust OBSS solution to cater for them?
– OBSS does not make situation worse, in fact correct Channel
Selection does lessen chance of Stray STAs
– Stray STAs occurrence is pretty unlikely
– Enlarging QLoad to include QAPs at Distance 1 does not add any
information that can be used for a Stray STA.
• Only in case of a 1:2:1 Overlap is there a way to know about a hidden
QAP, and then is the chance of a Stray STA enough to warrant Sharing
based upon Distance 2 Networks? Basically, only the STA is affected,
not the network
– Solution #2 does not require any additions to the proposal, but
could be added in informative text
– Solution #3 does require addition to the proposal and has merit –
Shall we add this?
Submission
Slide 14
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Conclusions
• Stray and Overlapping STAs will not occur if the STAs are within
the confines of the house/apartment
• Stray Overlapping STAs are corner cases
– Temporary, dynamic
– Fences etc. Make the occurrence pretty unlikely for houses
• Stray STA does not adversely affect the performances of the two
hidden networks, but its own performance is impaired
– RX performance can be mitigated through “Step-up Re-Start”
• APs can carry out new Channel Search at any time
– If it sees a Stray STA for example
• OBSS Channel Search lessens chance of Stray STAs
• Could consider adding “unsolicited 11h Measurement Report”,
sent by STA to inform AP of situation.
Submission
Slide 15
Graham Smith, DSP Group
Aug 2009
doc.: IEEE 802.11-09/0931-01-00aa
Proposal
• Do not amend QLoad Element beyond reporting on self
QLoad only
• Add informative text covering possible actions when a
Stray STA is present – AP can search for new channel
• Should we consider adding a ‘Report’ for a STA to
report that it is in a Stray STA situation?
– If so, use 11h Measurement Report basic format?
• Channel Number, Measurement Start time, Stop time, and Map (use
bit 5 of Map for example)
– Could do it for completeness, but is the ‘problem’ big enough to
warrant it? – I personally do not think so
Submission
Slide 16
Graham Smith, DSP Group