A Data-Driven IDCT Architecture for Low Power Video Applications

Analysis of transform coded (MPEG) video data streams reveals a large percentage of zero-valued Discrete Cosine Transform (DCT) coefficients. A Data-Driven 2D IDCT architecture (DDIDCT) is proposed which exploits this observation for energy efficiency. The DDIDCT architecture exploits variability in the computational workload caused by the presence of zero-valued DCT coefficients by adaptively changing the power supply and the clock frequency of the main computation units. Adaptive minimization of switching events and power supply voltage make the DDIDCT approach more energy efficient than a conventional fast row-column Distributed Arithmetic IDCT implementation by more than an order of magnitude for the same sample rate.