Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

help needed in viterbi decoding

Status
Not open for further replies.

shameem

Member level 5
Joined
Oct 27, 2005
Messages
90
Helped
5
Reputation
10
Reaction score
3
Trophy points
1,288
Activity points
2,031
hi,
i have implemented a viterbi decoder with constraint length 7. My input buffer to the vitebi decoder is of length 768 bits. When i traceback after processing all the 768 bits, the output of viterbi is correct, but when i traceback after processing 64 bits at a time, a few bits in the output of viterbi are corrupted. I think in both the cases, the result should be same untill my traceback lenth is 5 times the constraint length.
If anybody have any idea, please help me.

thanks

shameem
 

salam

How many bits are corrupted?
traceback length of 5 times the constraint length makes the error rate low enough, but the more traceback length , the less error bit rate

bye
navid
 

hi,
two or three bits at the boundry of the trace back lengths (i.e 64 at the discontinuity of the traceback and decoding) are corrupted. if i process only 64 bits at a time and traceback and decode upto 73 no of bits(i.e i overwrite some of the output bits) the output is correct. But i do not want to do unnecessary calculation to overwrite the output buffer.
 

Hi

if the signal have memory of L, you can make a decision on a symbol after 5L, it means if L=7 and your buffer length is 64, you can only peak 29=64-5L symbols after doing the viterbi procedure on 64 symbols, then you shift them left and get in 29 new symbols from right.
 

The concept of the tracebakc is similare to "sliding window".
That's to say,
Assume constraint length=7,buffer for 64 bits.
Some textbooks said that 5 times constraint length is good enough for traceback to reach some reasonable BER.
So u first traceback 35 stages and then find a merge point.
From the merge point to the final stage, u can get 29 decoded bit.
After this, u store another 29 bits and do the same procedure as before.
 

thanks jimmy
you explain better than me, what i've in mind.
 

hi,
guys thank u very much for the discussion. But i think u people have taken me wrong. My input buffer to the viterbi decoder is of length 768 bits. To avoid maintain the history for 768 bits, i want to process the input buffer in the bunches of 64 bits. In this way i have to keep the history for only 64 bits.

thanks

shameem
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top