The best known general strategy over multiple iterations of Prisoner's dilemma (and I assume for this game as well) is called tit-for-tat. In the first round you play 'nice' i.e. cooperatively, and in every subsequent round, you do whatever the opponent did to you the last round.
How they came across this strategy is a cool story. There was a prisoner's dilemma contest where contestants had to write bots that played prisoner's dilemma against other contestants' bots. Many contestants submitted highly intricate and complex artificial intelligence programs that configured their playing parameters based on the opponent's playing style etc. After pitting all the bots together, the winner was the tit-for-tat bot, which comprised of 4 lines of code.