Information-theoretic bounds for switching activity analysis in finite-state machines under temporally correlated inputs

This paper presents information-theoretic bounds for finite-state machine (FSM) switching activity under high-order temporal correlations on the primary inputs. As the main theoretical contributions: (1) we extend the previous work done on switching activity bounds calculation to account for complex spatiotemporal effects which are present when the target machine models real hardware and receives data from real applications; (2) we show that, for the special case of unifilar FSMs, the order and probabilities characterizing the Markov chain on the state lines of the FSM can be easily deduced and the case of general FSMs, can be reduced to this simple scenario. As the experimental results show, the lower and upper bounds are very sensitive to the input environment and if appropriately derived, provide a reliable measure of the switching activity on FSM state lines.