Regression with Time Series data

I have count data, some of which forms an exponential decay type distribution and other data sets which are log normal. They are time series and I’m looking to create autoregressive models using first or second order lags.
I have a couple of questions which are follows
  • Is normalising the exponential decay data the best way to proceed and if so, what is the simplest method of doing so ?
  • I have read that Poisson Autoregression for count data may be a good way of predicting the next value in the time series. Is this similar in nature to linear autoregressive models ?