APPLIED SIGNAL PROCESSING AND IMPLEMENTATION Introduction Spring 2005 Embedded Systems group: pk, oo, ylm,.... Dicom group: kjh, pr, uh,..

2 ASPI Introduction Outline ASPI Paradigm (Strategy) ASPI Course Structure (Tactics) ASPI Project Proposals (Operations) 8 Semester Courses Group Rooms & Embedded Systems Lab Others

3 ASPI Introduction A 3 Paradigm (Strategy) Application: Non-Linear Signal Processing/Mobile Communication 1.Algorithm selection 2.Simulation 3.Architecture selection and modelling 4.Design Space Exploration 5.HW/SW Co-Design Applications Algorithms Architectures

4 ASPI Introduction ASPI Course Structure (Tactics) SW Platform analysisHW Platform analysis SW compilersHW compilers Design Methodology Algorithm analysis Design Space Expoloration 8.sem9.Sem

5 ASPI Introduction 8th Applied Signal Processing and Implementation THEME: Adaptive DSP Algorithms and Advanced Signal Analysis PERIOD: 1 February – 30 June PURPOSE: To provide the students with skills in comprehending and employing basic and advanced theories and methods for analysis, design and implementation of adaptive algorithms (single or multiple input) which are required for signal processing, -analysis, and –transmission in various application domains. To enable the students to apply theories for (numerical analysis), optimization and simulation of adaptive algorithms. To give a solid comprehension of the mechanisms related to optimizing the interaction between algorithms and architectures.

6 ASPI Introduction Project Proposals Overview Application & Algorithms Application & Algorithms Simulation & Implementation Environments MATLAB C ASM TMS320-C6713 MATLAB C ASM TMS320-C Sem. Proposals MATLAB C HANDEL-C  P + FPGA MATLAB C HANDEL-C  P + FPGA 9.Sem. Proposals

7 ASPI Introduction Project Proposal1 Overview

8 ASPI Introduction Project Proposals Overview

9 ASPI Introduction Project Proposals Overview

10 ASPI Introduction 8th Sem. Proposal1 The project will cover activities from analysis to implementation of a signal processing application, where the following activities should be included in a time schedule: 1.Analysis of the system 1.Specification of I/O transfer function and acceptance criteria (a GSM voice modeling principle) 2.Design and test of a test bed for all stages in the project 1.Specification of testbd (of yet unknown parts) 3.Implementation and test of primary modules 1.LPC-model with noise free speech input 2.Plot and test modules (since correct results are known) 3.Compare results with noisefree and noise infected input 4.Selection of project options. Investigation of noise properties of Residual part or LPC/model part 5.Analysis, design and test of noise reduction algorithms in chosen part. End of Matlab activity ~ 21/4 (Gate passage)

11 ASPI Introduction 8th Sem. Proposal 1 End of Matlab activity ~21/4 (Gate passage) 6.Mapping of (chosen) function onto a DSP processor 1.Specifications for conformance test criteria 2.Translation from Matlab to C 3.Test of mapping 4.Analysis of optimization options 5.Evaluation of code efficiency in time and space End of Project activity ~ 4/5. The main benefit of the project is to highlight (part of) the traditional design trajectory from Application to Implementation using Matlab and a high level language C/C++ incorporating a selected design method such as SPU or OOA&D. The project will emphasize signal processing as well as optimization techniques for code size and/or execution time.

12 ASPI Introduction 8th Sem. Proposal 2 A la proposal 1

13 ASPI Introduction 9th Applied Signal Processing and Implementation THEME: Non-linear DSP Methods and Real-Time Architectures PERIOD: 1 September – 31 January PURPOSE: To enable the students to understand, analyze, and employ state-of-the-art DSP methods and algorithms, particular in the domain of non-linear techniques. For these methods, the students will get an in-depth knowledge of the theoretical background, the applicability in various domains, the numerical and computational characteristics, as well as the possibilities and limitations as compared to other (traditional) methods.

14 ASPI Introduction 9th Applied Signal Processing and Implementation PURPOSE: To enable the students to apply theories and methods for analyzing programmable DSP (mono- and) multiprocessor architectures in order to qualified evaluate whether an architecture suits a given DSP functionality under the constraint of some cost function. Beside, to give insight into the various methods required in order to map an algorithm onto a DSP multiprocessor architecture and to exploit significant parts of the implementational solution space subject to optimizing one or more objective functions related to the algorithmic/architectural interaction. Change: Multiprocessor => heterogeneous architecture

15 ASPI Introduction 8th Semester Courses F8-1Engineering Responsibilities1 ECTSSE FP8-16Adaptive Systems2 ECTSPE FP8-19Inverse Filtering and Deconvolution1 ECTSPE FP8-12Higher Order Statistical Analysis1 ECTSSE FP8-18Multidimensional Signal Processing1 ECTSPE FP8-9Joint Time Frequency Analysis1 ECTSSE ASPI8-4 DSP Design Methodology 1 ECTSPE FP8-13DSP Algorithms and Architectures1 ECTSSE FP8-17 Software Programmable Platform Analysis 1 ECTSPE

16 ASPI Introduction ASPI Web Page

17 ASPI Introduction Group Rooms and EmbSys Lab ASPI8 14 studerende i 3 grupperum (1x14 + 2x18 = 50 m2) RUM:NJ a b Embedded Systems Lab A6-013

18 ASPI Introduction Any Other Business Keys Access Cards Computer Accounts Library Canteen ETC