SIRAM Handout

1
Scalable Internet Real-time Media
Wavelet-based multiresolution video codec In
order to efficiently support layered multicast
transmission of video a hierarchical (layered)
video encoding is needed. The layering should
ideally be applied both temporally, spatially and
in terms of quantization to provide a possibility
to tradeoff between frame-rate and resolution for
the receivers of the video. Furthermore, the
encoding need to be scalable both in terms of
bandwidth and complexity to address both the
bandwidth heterogeneity problem as well as the
end user equipment heterogeneity problem.

In the SIRAM project a prototype Wavelet video
codec is being developed and evaluated. Prototype
implementations on heterogeneous hardware ranging
from computationally constrained handheld devices
to powerful workstations will demonstrate the
scalability of the codec.
Flow control and redundancy coding In order for
real-time multimedia flows to be adaptive to the
bandwidth available on an internetwork a dynamic
flow control algorithm is needed. The algorithm
needs to provide fairness between users of
multimedia conferencing sessions and need to
behave in a well-defined manner compared to TCPs
flow control. By using a delay-based flow control
algorithm the overall packet loss rate can be
reduced, which is crucial for loss-sensitive
real-time multimedia applications. Closely
related to the congestion avoidance algorithm are
algorithms that provide redundancy to cope with
packet loss. Receiver-driven redundancy control
can be exploited to adapt the amount of redundant
information for each receiver to the experienced
packet loss-rate.