Adaptive FEC for Congestion Control Var arun Sing un Singh, Marcin Nagy, Jörg Ott, Lars Eggert IETF 90, Toronto, 24. July 2014 tools.ietf.org/html/draft-singh-rmcat-adaptive-fec-00
IPR • Lars made a 3 rd party disclosure: – https://datatracker.ietf.org/ipr/2394/ – Also sent to the mailing list
Error Resilience End-to-End Delay UEP SSA RPS NACK (RTX) 3G Link Packet Loss
CONCEPT FEC FEC media ra media rate te Available Capacity Throughput (B) Swapping (B) g (C) Par (C) artiall tially y swapping swapping FEC with media FEC with media FEC with media FEC with media STAY state � UP state � PROBE state � UP state � STAY state � PROBE state � DOWN state � STAY state � (A) Adding FEC (A) dding FEC time RTC TCP P Inter In terval val
CC Framework Media Encoder Rate RTP Packets Controller Source RTP RTCP FB Pacing Buffer Transport Layer (UDP) IP
RFC 6363: FEC Framework Media Encoder Rate RTP Packets Controller Source RTP FEC FEC Module Code Repair RTP RTCP FB RTP (De)mux Transport Layer (UDP) IP
State Machine Good conditions, probe network STEADY PROBE PROBE STAY No more bandwidth available Unstable conditions U n s t a b l e c o n d i t i o Stable conditions n DOWN s DECREASE S t a b l e c o n d Unstable i t i o conditions n s UP INCREASE
FEC Scheme • Open Issue – Leave it open i.e., generic – Or, RF5109 (ULP), RFC6015 (1-d interleaved), parity, 1- or 2-d interleaved XOR, Reed-Solomon, … – There was discussion in RTCWEB on FEC Schemes. S, Holmer, et al., " Handling ¡Packet ¡Loss ¡in ¡WebRTC ", Proc. of IEEE International Conference on Image Processing (ICIP 2013) , 9 2013.
RTCP Feedback • RLE of Post-repair (RFC5725) • RLE of loss packets (RFC3611) • RLE of discarded packets (RFC7097) • Packet count of lost and repaired packets – (draft-ietf-xrblock-post-loss-repair)
Applicability • Implemented over a delay-based congestion control – See paper for details • However would like to generalize it. – Apply to SCReAM, GCC, … Nagy M., Singh V., Ott J., Eggert L., Congestion ¡Control ¡using ¡FEC ¡for ¡Conversational ¡Multimedia ¡ Communication , Proc. of ACM Multimedia Systems, Singapore, SG, Mar, 2014,
Next steps • Code: (coming soon) – https://github.com/protocols-comnet/rmcat-adaptive-fec-code • Evaluation Paper: – Nagy M., Singh V., Ott J., Eggert L., Congestion Control using FEC for Conversational Multimedia Communication, Proc. of ACM Multimedia Systems, Singapore, SG, Mar, 2014, • Initial comments received from Ingemar Johansson – More feedback is appreciated J
Evaluation (1/3) 400 OWD: 50ms 350 300 Throughput [kbps] 250 200 150 100 50 0 0 100 200 300 400 500 600 700 800 900 Time [s] • Ns-2 simulation, Variable link capacity, Single flow on the link
Evaluation (2/3) OWD: 100ms 0 100 200 300 400 500 600 700 800 900 Time [s]
Evaluation (3/3) Link capacity OWD: 240ms FBRA FEC probe rate 0 100 200 300 400 500 600 700 800 900 Time [s]
Compete with short TCPs 0 100 200 300 400 500 600 700 800 900 Time [s]
TESTBED Evaluation (1/2) Two RTP flows on the link 0 0 FBRA Call1 FBRA Call2 PSNR Call1 PSNR Call2 40 200 Throughput [kbps] PSNR [dB] 400 400 200 40 0 0 0 50 100 150 200 250 300 350 400 Time [s] • 1Mbit/s link capacity, 50ms one-way-delay,
TESTBED Evaluation (2/2) 0 0 FBRA Call1 FBRA Call2 PSNR Call1 PSNR Call2 40 200 Throughput [kbps] PSNR [dB] 400 400 200 40 0 0 0 50 100 150 200 250 300 350 400 Time [s] • 1Mbit/s link capacity, 100ms one-way-delay,
Recommend
More recommend