TY - JOUR
T1 - Optimal erasure protection for scalably compressed video streams with limited retransmission on channels with IID and bursty loss characteristics
AU - Thie, Johnson
AU - Taubman, David
PY - 2005/9
Y1 - 2005/9
N2 - In this paper we combine priority encoding transmission (PET) with a limited retransmission (LR) capacity. We propose the resulting LR-PET scheme as a framework for efficient RD optimized delivery of streaming media. Previous work on scalable media protection with PET has largely ignored the possibility of retransmission. In the proposed LR-PET framework, an optimization algorithm determines the level of protection for each element in each transmission slot, subject to transmission bandwidth constraints. To balance the protection assigned to elements being transmitted for the first time with those being retransmitted, the proposed algorithm formulates a collection of hypotheses concerning its own behavior in future transmission slots. We show that this formulation of hypotheses is central to the success of the proposed LR-PET algorithm. Indeed, without this element, a greedy version of LR-PET performs only slightly better than PET without retransmission. Experimental results are reported using both IID and GE channel models, with a Motion JPEG2000 video source, demonstrating substantial performance benefits from the proposed framework.
AB - In this paper we combine priority encoding transmission (PET) with a limited retransmission (LR) capacity. We propose the resulting LR-PET scheme as a framework for efficient RD optimized delivery of streaming media. Previous work on scalable media protection with PET has largely ignored the possibility of retransmission. In the proposed LR-PET framework, an optimization algorithm determines the level of protection for each element in each transmission slot, subject to transmission bandwidth constraints. To balance the protection assigned to elements being transmitted for the first time with those being retransmitted, the proposed algorithm formulates a collection of hypotheses concerning its own behavior in future transmission slots. We show that this formulation of hypotheses is central to the success of the proposed LR-PET algorithm. Indeed, without this element, a greedy version of LR-PET performs only slightly better than PET without retransmission. Experimental results are reported using both IID and GE channel models, with a Motion JPEG2000 video source, demonstrating substantial performance benefits from the proposed framework.
KW - Bursty loss
KW - Channel coding
KW - Erasure channels
KW - Forward error correction
KW - Priority encoding transmission
KW - Retransmission
KW - Scalable video
UR - http://www.scopus.com/inward/record.url?scp=24344488380&partnerID=8YFLogxK
U2 - 10.1016/j.image.2005.05.004
DO - 10.1016/j.image.2005.05.004
M3 - Article
AN - SCOPUS:24344488380
SN - 0923-5965
VL - 20
SP - 697
EP - 709
JO - Signal Processing: Image Communication
JF - Signal Processing: Image Communication
IS - 8
ER -