Exploiting skewed state probabilities for low power state assignment

We propose a new approach to solve the problem of state encoding for minimizing power consumption of multi-level implementations of FSMs. The algorithm consists of a partitioning step designed to reduce the switching activity, followed by a state encoding phase which minimizes the area. The cost functions used try to take all the causes of power dissipation into account. Skew in state probabilities is exploited to reduce the switching activity. The switching activity of all the nodes in the implementation is modeled using the concept of entropy. A literal-saving based measure is used for the area minimization step. Non-minimal length codes are used to achieve a further reduction in power. Experimental results show that, on an average, there is a 22% reduction in power compared to JEDI