Design Techniques and Implementation of Low Power High-Throughput Discrete Wavelet Transform Tilters for JPEG 2000 Standard

In this paper, the implementation of low power, high throughput convolutional filters for the one dimensional Discrete Wavelet Transform and its inverse are presented. The analysis filters have already been used for the implementation of a high performance DWT encoder [15] with minimum memory requirements for the JPEG 2000 standard. This paper presents the design techniques and the implementation of the convolutional filters included in the JPEG2000 standard for the forward and inverse DWT for achieving low-power operation, high performance and reduced memory accesses. Moreover, they have the ability of performing progressive computations so as to minimize the buffering between the decomposition and reconstruction phases. The experimental results illustrate the filters- low power high throughput characteristics as well as their memory efficient operation.

Performance Improvements of DSP Applications on a Generic Reconfigurable Platform

Speedups from mapping four real-life DSP applications on an embedded system-on-chip that couples coarsegrained reconfigurable logic with an instruction-set processor are presented. The reconfigurable logic is realized by a 2-Dimensional Array of Processing Elements. A design flow for improving application-s performance is proposed. Critical software parts, called kernels, are accelerated on the Coarse-Grained Reconfigurable Array. The kernels are detected by profiling the source code. For mapping the detected kernels on the reconfigurable logic a prioritybased mapping algorithm has been developed. Two 4x4 array architectures, which differ in their interconnection structure among the Processing Elements, are considered. The experiments for eight different instances of a generic system show that important overall application speedups have been reported for the four applications. The performance improvements range from 1.86 to 3.67, with an average value of 2.53, compared with an all-software execution. These speedups are quite close to the maximum theoretical speedups imposed by Amdahl-s law.