TCP-friendly video communication

J. Viéron, C. Guillemot.

Contact: J. Viéron 

Approach

• This "demo" shows the results obtained with the new rate control algorithm we have designed that captures not only the behavior of TCP's congestion control avoidance mechanism but also the delay constraints of real-time streams. The channel state is estimated from feedback information conveyed in RTCP reports, and captured on sliding temporal windows. The channel state information is exploited for predicting the new throughput (sending rate), in accordance with TCP's congestion control avoidance. The protocol for estimating the parameters of the bandwitdh prediction model is based on the TFRC protocol described in [FHPW00], however with some additional features that allow to better take into account the multimedia flows characteristics. Given the current channel state, the encoder and decoder buffers states and the delay constraints of the real-time video are translated into encoder rate constraints. A large number of experiments has been carried out on several links, with a video source. The rate control mechanism is used in augmentation with a H.263+ loss resilient video compression algorithm.These experiments clearly demonstrate the benefits in considering some multimedia flows characteristics in the protocol used for estimating the bandwidth prediction model parameters. The design of a global rate control model that encompasses the source buffer model as well as the end-to-end delay characteristics allow to reduce significantly the source timeouts, hence to minimize the expected distortion, for a comparable usage of the TCP compatible predicted bandwidth.

Experimental Results

• Intensive experiments have been performed between INRIA Rennes (France) and R.U.S (Germany). We have used an H.263+ codec with the TMN8 [Q15A20] rate control technique (adapted to allow for varying bit budgets). Additionnaly, as described [Q15G16] in order to make the bitstream more error resilient we use the slice structured mode option and the INTRA forced updating option which is a macroblock refresh policy based on the mean data loss rate. Moreover, in order to mask the packet losses the H.263+ decoder used a algorithm of concealment. Different values have been considered for the initial buffering delay, expressed in number of frames (i.e. K=2,3,4 frames). For each case we have made more than 50 simulations, in order to have various channel realizations. The video sequence used here is the sequence News (CIF, 25 frames/s) used in loop for 30 seconds (i.e. 750 images). The reconstructed sequences shown here are representative of the results obtained. They correspond to the results and curves described in [VG00].

• STRAIGHTFORWARD APPROACH : predicted bandwidth as rate constraint
  The sequences below show the impact of the high number of timeouts (higher than 5%) which occured generally in long bursts, obtained by using this naive approach. Moreover the amount of timeouts increases with the buffer size.
  
  
  • Timeouts impact: This figure illustrates the impact of the timeouts occured between the frame number 400 and the frame number 500 on the reconstructed video for an initial bufferization of 4 images.

• • Reconstructed Sequences :

• SOURCE ADAPTIVE APPROACH : taking into account the source delay constraint and the buffer model
  With our algorithm the number of timeouts has decreased highly (lower than 0.095%) for an optimal usage of the TCP-compatible bandwidth predicted. Moreover the timeouts do not occured in bursts and thus their impact on the quality of the video less important.
  
  
  • Timeouts impact: This figure illustrates the impact of the timeouts between the frame number 400 and the frame number 450 on the reconstructed video for an initial bufferization of 4 images.

• • Reconstructed Sequences :

Publication

• This rate control algorithm is described in details in [VG00].

References

[FHPW00]

S. Floyd, M. Handley, J. Padhye and J. Widmer, "Equation-Based Congestion Control for Unicast Applications". To appear in ACM SIGCOMM 2000, May 2000.

[HOK99]

C.Y. Hsu and A. Ortega and M. Khansari, "Rate-control for robust video transmission over burst-error wireless channels". IEEE Journal On Selected Areas in Communications, VOL.17 NO.5 May 1999.

[Q15A20]

J. Ribas-Corbera and S. Lei, "Rate Control for low-delay video communications". ITU-Telecommunications Standardization, Video coding expert group, Doc Q15-A20, June 1997.

[Q15G16]

G. Côté and M. Gallant and F. Kossentini, "Test model 11". ITU-Telecommunications Standardization, Video coding expert group, Doc Q15-G16, February 1999.

[VG00]

J. Viéron and C. Guillemot, "Real-time constrained TCP-compatible rate control for video over the Internet"., December 2000. (downloadable: draft version)