Multiple codes state assignment and code length reduction for power minimization of finite state machines

Abstract

The method of a minimization of the power consumed by the finite state machine (FSM) is discussed in the presented paper. The proposed algorithm uses two performed sequentially methods of equivalent transformations of the FSM, which do not change the function, but the structure of an FSM. One method assigns multiple codes to the internal states of an FSM. Using more than one code gives more opportunities to assign to the neighbor states the codes with lower Hamming distance, which in result lead to decrease a power consumption. The other method reduces the length of the internal state’s code using a special model of an FSM, in which the orthogonal state codes are obtained from a concatenation of the parts of the input and output vectors, and the subcode stored in a memory. This approach gives the possibility to reduce the size of the memory used for storing the internal state’s code, which leads to reducing the power consumption. Proposed algorithm executes methods starting from the state splitting, followed by the implementation the common architectural model. The experimental results show that the proposed method reduces the power consumption compared to NOVA, JEDI, column based and sequential algorithms.