[M4IF Technotes] Research areas for error resilience

Yang, Wei Jian (Mac) yangwj lucent.com
Sat Sep 28 15:12:13 EDT 2002


Dear MPEG expert:
In this letter, I try to figure out my start point of research on video over
wireless.
It covers a large scope of field, even focus on error resilience. I will
discuss it in three steps:
source coding, channel coding and receiver. If you have any ideas, welcome
to discuss with
me.
1. Source Coding:  
MPEG-4 has enclosed some techniques for error resilience, i.e, Video Packet
structure, Data partitioning 
and RVLC which are arranged in a hierarchical manner. The test environment
in MPEG4 VM is 
BER=10e-3(random), burst error(10ms). Some papers are discussed on
Rate-Control and bit-allocations 
when more sync. (in Packet head and HEC) and maker (between data
partitioning) are included even 
BER is up to 10e-1.  
2. Channel Coding: 
If the source coding resilience tool is not enough when channel has severer
BER or fading, channel coding 
is necessary to have better visual quality. Some papers are discussed on UEP
using BCH, Convolutional code, Turbo 
or EP  using Reed-Solomon. UEP(UnEqual Protection) only gives those
important bits in stream such as
Sequence start, picture relative info.,etc. While EP (Equal Protection)
will protect whole stream. Scalable coding (MDC
& SPIHT) is another way to provide UEP to assign important bits to more
reliable channel. 
An interesting field is to apply Reed-Solomon to UEP, Implementation
speed-up may be an optimization.
3. Receiver:
After source and channel coding, message will be transmitted over wireless
channel to receive after live through 
fading, interference and noise. Some papers have discussed on RAKE in
Nakagami model and MUD to see the performance
improvement in real environment. An interesting field is to apply MUD to
Nakagami model.
Regards
Yang, Wei Jian(Mac)
Lucent Technologies


More information about the Mp4-tech mailing list