Controller power estimation using information from behavioral description

We present a method to estimate power in finite-state-machine controllers at the logic level using information from behavioral description. This provides an ample opportunity of feeding back this power estimate to higher level design steps, like scheduling, allocation and state assignment. The method makes use of information regarding the state transitions from behavioral description using Markov-chain modeling. This information is percolated to the lower level to achieve better estimation of switching activity at the inputs of the next-state logic. Expected energy, dissipated due to switching of the input-output lines of the next-state logic, is estimated by considering transitions over the expected execution duration of the algorithm. Since the probabilistic model used here captures conditional state transitions and loops, power estimation for iterative algorithms would be more accurate than traditional methods.