I have time-dependent data which show an inreasing variance with a certain covariate X (which can be modelled using a fixed cariance structure, I guess, i.e. considering sigma*X). Furthermore independence is violated in two respects: 1.) I have multiple measurements in time for different individuals (thus I want to use ID as a random coefficient) and 2.) The data show an autoregressive structure, especially an AR(3) dependence.

Now I look for a function in R which combines fixed variance, random coefficient for ID and the AR(3) dependence. Any suggestions?