I think there is no need to post the full frame, because it will look the same as in my first example. Just couple more bits on the right side.
So it must be a proprietary coding then. It's by 1/3 less effective than Manchester coding, but using this coding it is very easy to clock data, because each transition from high to low represents an end of a bit (except for the start pulse). I was thinking that maybe it is some encoding standard which I don't know.