This paper presents an implementation of a low-power and pure-hardware advanced-audio-coding (AAC) audio decoder system. Based on the characteristics of each decoding block, the AAC system is partitioned into four separate modules. For low-power and low-complexity considerations, architectural- and algorithmic-level approaches are adopted in both individual modules and whole system. In parallel PLA-based codeword decoder, we achieve a constant output rate of Huffman decoding in 2.5 cycles for the worst case, and memory usage is decreased compared to that in the binary-tree memory-based method. In reduced lookup table inverse quantizer, a table lookup with interpolation scheme is adopted which reduces the size of the lookup table from 8192 to 256. In hardware-shared signal processor, we use a hardware-sharing technique which integrates several similar blocks into a common hardware to reduce cost and enhance hardware utilization. In fully pipelined filterbank, a fast algorithm decreases the numbers of multiplication and addition largely to factors of 24 and 144 for the short and long blocks, respectively. A corresponding hardware for filterbank processing is proposed with fully pipelined architecture. Referring to stereo processing, a single hardware is shared for the channel pairs with low-cost consideration. The hardware operations of each module are well scheduled with high utilization of pipeline, and furthermore, the parallel processing among blocks is joined to increase efficiency. A 48% power savings can be reached by using the pipeline and parallel techniques of the channel pair. The proposed AAC decoder is realized in UMC 0.18-μ 1P6M technology and is operated at only 3 MHz in the worst case. The power dissipation is only 2.45 mW at the sampling frequency of 44.1 kHz.
|Number of pages||12|
|Journal||IEEE Transactions on Circuits and Systems I: Regular Papers|
|State||Published - 2009|
- Advanced audio coding (AAC)
- Low power