PPT Slide
Note that with “random coding” we replace Kd by
and with this r(? ), E ????> 0 for all ??< 1 provided R < C (achieves Shannon Capacity)
Is this the case for Turbo Codes?
No, even though r ???? seems to approach H(?)-(1-R) ln 2 for most values of ??
Initial ? “tail” prevents this and causes bottoming out of PE. But E ????> 0 for all