For c = a + b, UNSIGNED
The overflow condition for unsigned 8 bit integer would be c < a or c < b
For c = a + b, SIGNED
We can't say anything like that. We need to say only with respect to the sign of the input numbers. If the sign of a and b are same but the sign of c is different, there's overflow. If a and b are of different sign, overflow cannot happen.
(When a = -3 and b = -5, c = -8 but there is no overflow and c < a and c < b)