1. Feedback-jamming Multicast ARQ results with capture effect
Nov 09
doc.: IEEE /1150r0
Nov 09
Feedback-jamming Multicast ARQ results with capture effect
Date:
Authors:
Jochen Miroll, Saarland University
Jochen Miroll, Saarland University

2. Nov 09
doc.: IEEE /1150r0
Nov 09
Abstract
This presentation is an update on the NACK jamming LBP proposal previously made to TGaa by the authors and aims at disclosing some synergies between D0.02 MRG and LBP
The capture effect with reasonable ACK/NACK frame structures is tested and evaluated to give an idea about the feasibility of NACK jamming in 11aa.
Jochen Miroll, Saarland University
Jochen Miroll, Saarland University

3. Glossary ACK-polling As MRG BA does in P802.11aa D0.01
Nov 09
Glossary
ACK-polling As MRG BA does in P802.11aa D0.01
LBP Leader-Based Protocol
NACK-jamming Enforced collision of leader-ACK with non-leader NACK
‘‘ probability Probability that ACK is lost due to simultaneous NACK
HLBP Hybrid (w.r.t. coding) LBP
Jochen Miroll, Saarland University

4. NACK jamming Leader Based Protocols
Nov 09
NACK jamming Leader Based Protocols
We propose(d) a leader based feedback jamming Multicast retransmission protocol for 11aa
Neither compulsory nor contrary but complementary to the already specified MRG-BA in D0.02
Previously we have shown an increased scalability of such a scheme w.r.t group size
Refer to document 09/0290r1
Recap on next two slides
Jochen Miroll, Saarland University

5. Recap LBP Mar 09 Nov 09 doc.: IEEE 802.11-09/1150r0 optional
SEQ# indicator and NAV updater to synchronize jamming
Jochen Miroll et al., Saarland University
Jochen Miroll, Saarland University

6. Nov 09
Recap HLBP*
Phase I Transmit a block of frames, similar to block-ACK
Phase II Replace actual block-ACK phase with parity-NACK phase
* A rateless FEC code is assumed in the figure, for e.g. for RS, parity would be parity OR data, receiver counts FEC packets
Jochen Miroll, Saarland University

7. NACK jamming leader-based protocols
Nov 09
NACK jamming leader-based protocols
Scalability
Independent of number of number of client STAs per group
Leader selection and rate/link adaptation become related
For the AP and leader the behaviour is similar to unicast (use ARF?)
Error correction delay
Independent of number of client STAs per group
HLBP
Enable cross-layer error correction, e.g. RS on application layer
Then the remaining MAC mechanism is quite simple
Parity / NACK phase could be moved to concealed retransmission multicast group similar to MRG unsolicited retries
Jochen Miroll, Saarland University

8. NACK jamming leader-based protocols
Nov 09
NACK jamming leader-based protocols
From March 09 meetings two fundamental questions arose:
Does the „capture effect“ impede the ACK/NACK jamming?
How is the jamming probability related to leader selection?
Thus we have conducted ns-2 and Matlab simulations for deriving NACK jamming probabilities
Jochen Miroll, Saarland University

9. Nov 09
The „Capture Effect“
Capture Effect is described in the literature as the effect that a signal can be correctly received in the presence of another transmission.
Naively thinking: „how could this affect wireless networks?“
If an OFDM data symbol is being received and some other OFDM data symbol is „on the air“ (same channel), we assume increased (white) noise at the receiver
=> correct reception depends on the power difference
If two frames start at approximately the same time with different content
=> Correct reception depends on phase and power difference
Jochen Miroll, Saarland University

10. The „Capture Effect“ Power level difference Figure: Nov 09
Assume interfering signal has flat spectrum within 20MHz channel
„Capture effect“ depends on the resulting SNR
Figure:
Packet error rate (1500B) in Rician fading channel with Rician factor 1 (analytical results from text books)
Jochen Miroll, Saarland University

11. The „Capture Effect“ Transmitting at the same time
Nov 09
The „Capture Effect“
Transmitting at the same time
Receiver is not synchronised
PLCP of TX1, TX2 and TX3 overlap
Question is: Can the receiver still obtain
Coarse timing from short preambles
AGC
Clock sync
Fine timing from long preambles
Channel equalization
Rcv
Sender
Interferer 1
Interferer 2
Frames transmitted
at the same time
TX 1
TX 3
SIFS
TX 2
Jochen Miroll, Saarland University

12. Nov 09
The „Capture Effect“
Transmitting (nearly) the same content at the same time
Assume ACK and NACK differ in only one bit
Conv. coded, scrambled signal „on the air“ will differ starting from the differing bit position
Receiver will see an echo
Could be beneficial, sync will probably be obtained
But starting from the differing bit position, the „echo“ ceases to be an echo and becomes interference
SIFS
Rcv
ACK
Sender
similar
Interferer 1
similar
Interferer 2
Very simliar frames
Transmitted at the
same time
Jochen Miroll, Saarland University

13. NACK frame Why „very similar“ ACK/NACK frames?
Nov 09
NACK frame
Why „very similar“ ACK/NACK frames?
We assume it to be unlikely that 11aa will introduce a „jamming signal“
Instead, a new frame type is more likely to be specified
In the worst case (w.r.t. jamming probability) a new frame (NACK) may only differ from a standard ACK in one bit, i.e. the frame type.
Jochen Miroll, Saarland University

14. Nov 09
Matlab Simulation
To simulate „capturing“ in matlab we implemented a complex baseband from scratch:
PLCP and OFDM signal generator
10x upsampling to simulate „analog“
Interferer with 1 bit difference in the MAC frame (type bit NACK/ACK)
Interferer jitters by +- GI (5% SIFS)*
AWGN / multipath channels
Viterbi decoding 10,000 simultaneous ACK/NACK transmissions
AP1
STA 1 (leader)
STA r-1
STA r
NACK jitter
ACK
NACK
SIFS
Jamming
* a GI: 16 samples, 0.8 us, SIFS: 16 us
Jochen Miroll, Saarland University

15. ACK packet error rate over power difference of ACK and NACK
Nov 09
Matlab results (AWGN)
ACK packet error rate over power difference of ACK and NACK
Jochen Miroll, Saarland University

16. ACK packet error rate over power difference of ACK and NACK
Nov 09
Matlab results (AWGN)
ACK packet error rate over power difference of ACK and NACK
Jochen Miroll, Saarland University

17. Ns-2 simulation environment
Nov 09
Ns-2 simulation environment
ns-allinone-2.34
802.11Ext model which supports capture effect (see ref)
Jochen Miroll, Saarland University

18. Ns-2 results Ns-2 parameters
Nov 09
Ns-2 results
Ns-2 parameters
PLE: 3
Rayleigh fading (param. Nakagami)
Tx power: 20dBm
1 Leader and 1 non-leader, same distance to AP
Periodic leader selection (mean SNR decision)
Jochen Miroll, Saarland University

19. Nov 09
Questions?
Jochen Miroll, Saarland University

20. Nov 09
Straw Poll (1)
Is TGaa interested in further results on leader based NACK jamming for more reliable video groupcast?
Yes:
No:
Abstain:
Jochen Miroll, Saarland University

21. Nov 09
References
Q. Chen, F. Schmidt-Eisenlohr, D. Jiang, M. Torrent-Moreno, L. Delgrossi, and H. Hartenstein, "Overhaul of IEEE modeling and simulation in NS-2," in Proc. Of the 10th ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM), 2007
Multicast MAC Extensions for high rate real-time traffic in Wireless LANs
Jochen Miroll, Saarland University