Exponentiation is widely used in practical applications related with cryptography, and as the discrete log is easily solved in case of a low exponent n, a large exponent n is needed for a more secure system. However. since the time complexity for exponentiation algorithm increases in proportion to the n figure, the development of an exponentiation algorithm that can quickly process the results is becoming a crucial problem. In this paper, we propose a parallel exponentiation algorithm which can reduce the number of rounds with a fixed number of processors, where the field elements are in GF($2^m$), and also analyzed the round bound of the proposed algorithm. The proposed method uses window method which divides the exponent in a particular bit length and make idle processors in window value computation phase to multiply some terms of windows where the values are already computed. By this way. the proposed method has improved round bound.