• Summary
• Related Documents
  • Active Bibliography
  • Co-citation
• Version History

Abstract:

Hierarchical coding techniques split signals into components of varying importance. The aggregation of these components reconstructs the original data, but subsets of them can also provide various degrees of approximation to the original signal. Hierarchical coding is very important for the efficient use of high-speed packet-switching networks. The main issue for such network architectures is the significant congestion control problems that can arise due to the statistical multiplexing of many very high burstiness signals. A key property of any congestion control approach, which is not based on resource reservation at the peak rate of the sources, is the ability to shed load quickly without causing an avalanche of retransmissions of dropped traffic. With hierarchical coding of continuous media, when network congestion arises it is possible to drop the less important signal components without causing service interruption, and without the need for retransmissions. Since it is expected that continuous media will constitute the bulk of the network traffic, this technique can be very effective as a last resort for congestion control. Hierarchical coding can also play an important role at the receiver because it provides the system software with the capability of allocating resources based on local specifications and priorities. This might, for example, entail deciding to gracefully and dynamically degrade the quality of the received and played-back signal when resources are limited. Video conferencing for scientific collaboration and animation of scientific visualization sequences, are Sequoia 2000 applications that might benefit from the use of hierarchical coding for communication over packet-switching networks. Image database browsing can also take advantage of the reduced perceived latency of either progressive transmission or transmission of only the most important signal components. 1.

Citations