Interferences arising between wireless sensor-actuator systems communicating over shared wireless channels impact closed loop control performance. We design interference-aware channel access policies where the total transmit power of the sensors is minimized while desired control performance is guaranteed for each involved control loop. Control performance is abstracted as an expected decrease rate of a Lyapunov function for each loop. We prove that the optimal channel access policies are decoupled so that, intuitively, each sensor balances the gains from transmitting to its actuator with the negative interference effect on all other control loops. Moreover the optimal policies are of a threshold nature with respect to channel conditions, that is, a sensor transmits only under favorable local fading conditions. Finally, the optimal policies can be computed by a distributed iterative procedure which does not require coordination between the sensors. (C) 2018 Elsevier Ltd. All rights reserved.