constraint length (K) in convolution encoding

convolution encoing

Hello,

After consulting the following 2 ref. for convolution encoding, i got confused about constraint length as they are defined differently in both ref. My questions are as follow...

From the source below:
**broken link removed**

convolution codes are define by three parameter: (n,k,m)
n= # of output bits
k= # of input bits
m= # of memory registers

Constraint length = k(m-1)

for e.g. (2,1,4) Memory registers = 4 , constraint length = 1(4-1) = 3 . that means # of state required for output is 2^3 = 8. For the all other examples listed in this file, constraint length works out as described by k(m-1) formula.

Now what's confusing is the Constraint length described in the following slides.

**broken link removed**

Example with code rate=1/2 and K=3. It would be defined as (2,1,4) if the constraint length is 3. But then # of memory registers would be 4 but its' not. Number of registers is 3 here. If we define it as (2,1,3) then everything works out accordingly except one thing-> constraint length.According to the formula k(m-1), constraint length is 1(3-1)=2.

Why is it different? Is it user defined?

constraint length tells us the # of states required for output. but what does it represent in terms of encoding?

Thanks a lot!!

Shows how carefully you have studied the topic!

There is some "convention confusion " on the definition of constraint legth, that's all. Some guys have constraint length = number of mem bits, some have +1 of that. If you are writing an exam, just follow your teacher's defintion . If not, keep in mind both are used in literature, find out what the author is follwong and stick to that.
-b

Very good answer by Bulx I really liked it,