This paper addresses the optimization problem of minimizing the number of memory access subject to a rate constraint for any Huffman decoding of various standard codecs. We propose a Lagrangian multiplier based penalty-resource metric to be the targeting cost function. To the best of our knowledge, there is few related discussion, in the literature, on providing a criterion to judge the approaches of entropy decoding under resource constraint. The existing approaches which dealt with the decoding of the single-side growing Huffman tree may not be memory-efficient for arbitrary-side growing Huffman trees adopted in current codecs. By grouping the common prefix part of a Huffman tree, in stead of the commonly used single-side growing Huffman tree, we provide a memory efficient hierarchical lookup table to speed up the Huffman decoding. Simulation results show that the proposed hierarchical table outperforms previous methods. A Viterbi-like algorithm is also proposed to efficiently find the optimal hierarchical table. More importantly, the Viterbi-like algorithm obtains the same results as that of the brute-force search algorithm.
