| The increased concern for energy aware system design has kept pace with current practices among circuit and system engineers like never before. The big thrusts for this concern are the excessive heat generation and the corresponding expensive cooling techniques, the devices reliability problems due to increased temperatures, and most importantly, the incapability of battery technology to supply systems with the anticipated amounts of energy, especially for portable devices. Multimedia applications, which are data-intensive applications, play a chief role in the wireless revolution, and the key requirements for the success of systems supporting multimedia applications will be suitable transmission/storage bandwidths and system power consumption. In this work, these two issues are addressed. A new DCT-based Algorithm, Coefficient Elimination, is proposed and implemented with three DCT architectures: NEDA, Direct-DCT, and LP-DCT. Simulation results show that this algorithm can reduce the DCT energy dissipation by 26–60%. In addition, new algorithms, Dynamic Profiling, are proposed for Low Bit-Rate video coding. Unlike many of the available LBR techniques, this technique is suitable for mobile devices with reduced sources such as computing power and energy resources. These algorithms not only outperform coarse quantization technique, but also result in a low energy implementation of the decoder where IQ-IDCT operations are replaced by simple buffer accessing operation. |
