Generally speaking, linking people (even son/daughter-parent pairs) across US Census samples is made impossible by the fact that each census sample only collects detailed information on a relatively small share of the entire US population. Therefore, the chances that the same people or son/daughter-parent pairs appear in more than one census sample is quite low. Additionally, even if the same people or son/daughter-parent pairs appear in more than one census sample it would be impossible to appropriately link these people.
That said, this sort of task is possible in several historical samples where “full count” information is available. I am not sure what years you are interested in analyzing, so this option may not be helpful if you are interested in more recent sample years. The IPUMS linked representative samples, 1950-1930 make this sort of linking possible. These linked samples include 7 pairs of years: 1850-1880, 1860-1880, 1870-1880, 1880-1900, 1880-1910, 1880-1920, and 1880-1930. Each of these contains three independent linked samples: one of men, one of women, and one of married couples. Additional information and documentation is available here.