I
have to say that the simplicity and power of wireline communication
encoders have not been replicated in wireless communications systems.
When examining systems that use LSFR scramblers , RS encoding,
Constitutional Encoding, Turbo Product Coders, LDPC encoders, nothing
really matches the features and simplicity of your garden variety 8B10B
encoder and decoder.
Sadly
the one needed feature of channel encoders is the ability to have a
data space and control space. K characters allow the link to separate
the messages passed into unique spaces such as a data space and control
space. This feature allows the in-band control to be sent co-mingled
with the data payloads. The current systems force an upper level
protocol to make a header that may not be unique then encapsulate
control data. Byte or word synchronization is always an issue with the
wireless channel encoders. On scrambled systems there always exists a
time shifted version of the synchronizing sequence. In an effort to solve the problem longer sync sequences are selected. Possibly with CRCs to validate the sync header.
Using
8b10b carries a hefty penalty though, for 10 available bit times only 8
are transmitted unless you count the "isK" to be one of the bits
received then 9 bits are received in the span of a 10 bit time.
Therefore "bits received" / "bit times available" = 8/10 or 80% of the
bandwidth is utilized for actual data. 8B/10B codes are derived from
5B/6B and 3B/4B codes. The filing date for the 5B/6B codes is 1991. So
nothing new for 23 years now.
64/66
bit codes is a sad attempt to do the same thing but suffers from DC
bias wondering based data pattern as a result of the scrambling. DC bias
is controlled on a word by word basis on a 8B/10B system by calculating
the running disparity. Also 64/66 bit encoding is not generally used on
wireless communications. One can argue that DC bias is not needed in
wireless system since in today's modern communications systems are phase and amplitude modulated and many sub carriers are used (QAM, OFDM)
Leonard Dieguez
No comments:
Post a Comment