model baseddesign ofdigital communication systems

21
Model Based Design of Digital Communication Systems Intro, Organization, Ressources Prof. Dr.‐Ing. Marcus Purat Beuth Hochschule für Technik Berlin 2019

Upload: others

Post on 01-Jan-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Model BasedDesign ofDigital Communication Systems

Model Based Design of Digital Communication Systems

‐Intro, Organization, Ressources

Prof. Dr.‐Ing. Marcus PuratBeuth Hochschule für Technik Berlin

2019

Page 2: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 2

• Module Content and Course Objective• Introduction to HW Platform components

– Overview Zync Board ZyBo– Overview Zync 7000– Overview Audio Codec

• Introduction to Workflow (first steps in lab)– Building a Reference Design and Integration into a Simulink‐Model– Generate an I²C Control IP core for audio codec chip configuration– Generate an I²S IP core for audio codec chip data transfer– Implementing a configurable audio filter IP core for signal processing

• Project Description• Course Organization and Assessment

Structure

Page 3: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 3

Implementation and deployment of complex embeddedsignal processing systems with HW/SW Co‐Design  

Module Content

Compiler

Algorithm / Idea All DigitalPlatformwith I/O

SW/HW‐Split and Design

SW: µC, DSP

HW: ASIC, FPGA

Coding /Implementation

C, C++,Java, Python

VHDL,Verilog

Synthesis

Binaries

Bitstream

Programming

Page 4: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 4

DSP vs. FPGA / SW vs. HWDSP / SW:• 1 or 2 fast array multipliers and accumulator• MAC intensive algorithms (Filter, Correlation, Fourier Transform, …)• Speed only limited by speed of multiplier (with pipelining)• Fixed number representation (fixed MAC core, Floating / Fixed Point)• Typically lower cost than FPGA• Efficient programming (Sequential progr. like sequential processing)• Useful for complex, highly structured (if-then-else) algorithms

FPGA / HW:• Many IO pins for IO connections (e.g. ADC/DAC)• Multiple parallel MAC cells possbile / parallel processing possible• Number representation adaptable to algorithm needs• High-bandwidth algorithms (wireless, video)• Bit-oriented algorithms (error control, cryptography, scrambling, …)• Useful for front-end algorithms (Filtering, CORDIC, DFT), co-processor

Page 5: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 5

Efficient implementation and deployment of complex systemsusing model‐based design and SW/HW code generators

Module Content

Algorithm / Idea All DigitalPlatformwith I/O

SW/HW‐Split and Design

System Model

• Code Generation• Programming

Deployment

HW in the loop (HIL)

Optimization/Test

Integrated Toolchain (here: Mathworks Simulink)

SW

HW

HW Reference Design incl. I/O

Page 6: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 6

Model‐Based Design• Common framework for the design process supporting the 

complete development cycle– Mathematically modeling a system– Simulating, analyzing and optimizing the system performance– Testing and verifying the system and its interfaces using HW in the 

loop– Deployment of the system on a SW/HW‐Platform

• Advantages– Common design environment for communication, data analysis, and 

system verification between various (development) groups.– Early location and correction of errors in system design– Design reuse, for upgrades and for derivative systems– Less time‐consuming for implementation and debugging– Less prone to coding errors, very good documentation– Reduced complexity by visual hierarchies of individual design blocks

Page 7: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 7

Model‐Based Design• Disadvantages

– Porting model between different platforms may be more time consuming (reference designs, coverall approach), and sometimes not possible at all due to missing tool chain support

– Hidden compilation tool chain is causes non‐transparency and difficulties to combat post errors (such as compilation errors)

– Care needs to be taken in real world production environments because of reliance on a given toolchain can with significant rework and possibly compromising engineering approaches.

Page 8: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 8

Platform

ZyBo (Zync Board) with the Xilinx Zynq 7000 all programmable System‐on‐Chip (AP SoC)

Implementation and deployment of an audio processingsystem on a all programmable digital platform

Course Objective

[Mathworks]

[Digilent]

Page 9: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 9

Overview Zync 7000 AP SoC

[Xilinx]

References• zync/userguide.pdf

Page 10: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 10

[Xilinx]

Overview Zync 7000 AP SoC

ARM NEON technology is an advanced SIMD (single instruction multiple data) architecture extension for the ARM Cortex A series. NEON technology accelerates digital signal processing algorithms and functions to speed up audio and video processing applications.

Page 11: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 11

[Xilinx]

Overview Zync 7000 AP SoC

Page 12: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 12

• High quality stereo audio codec for portable applications• Stereo 24‐bit analog‐to‐digital and digital‐to‐analog converters• SNR 90‐100 dB• Various configurable sampling rates from 8 kHz to 96 kHz• Programmable gain amplifier PGA

References• codec/datasheet.pdf

Overview Audio Codec SSM2603

[Analog Dev.]

I²CI²S

Analog IN Analog OUT

Page 13: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 13

Overview Zync Board ZyBo

[Digilent]

ETH

Audio Connectors(Codec on the back)

Logic LEDs and Switches Zync AP SoC

µSD card slotPower Switch

UART/JTAG

References• zybo/ref_manual.pdf• Zybo/zync_book_tutorial.pdf

Page 14: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 14

Introduction to Workflow

1.  Building a Reference Design and Integration into a Simulink‐Model

References• mathworks/Ref_design.html  Follow and carry out in practice the

steps to learn how to integrate the SoC in MATLAB• mathworks/AXI.html Read and understand• mathworks/HWSW_codesign.html Follow and carry out in 

practice the steps to learn how to integrate the SoC in MATLAB

[Mathworks]

Page 15: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 15

Introduction to Workflow

2.  Generate an I²C Control IP core for audio codec chip configuration

References• mathworks/audio_i2c.html Read and understand

[Mathworks]

Page 16: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 16

Introduction to Workflow

3.  Generate an I²S IP core for audio codec chipdata transfer

References• mathworks/audio_i2s.html Read and understand

[Mathworks]

Page 17: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 17

Introduction to Workflow

4.  Implementing a configurable audio filter IP corefor signal processing

References• mathworks/audio_filter.html Follow and carry out in practice the

steps to learn how to integrate the SoC in MATLAB

[Mathworks]

Page 18: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 18

Project Description

Switchable 3‐Band parametric equalizer

LP

Bypass

BP

Bypass

HP

Bypass

IN OUT

Shelving Peak Shelving

H [dB]

f

On‐board switch

GLPGBP GHP

fC,LP fC,HPfC,BP

BWLBP

G: Filter Gain, fC,LP/HP: cut‐off frequency, fC,BP: center frequency, BWBP: Bandwidth

LED LED LED

Page 19: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 19

Dashboard in Simulink

Project Description

Dashboard control via Simulink and Processing System

References• filter/extract_dafx.pdf

G, fC,LP/HP, fC,BP, BWBP

Parametric EQin PL

IN OUT

Filter CoefficientCalculation in PS

Digital Filter Coeff.

Gain Frequency

PC

Zybo / Zync

Page 20: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 20

• Build project groups with maximum three students per workstation

• Reading and working through workflow introduction in groups, appr. 4‐5 weeks

• Organize project work within groups– Who does what and when?– Split work (theory, model filter, dashboard, …)– Set milestones

• In case of any questions – ask supervisor!

Course Organization

Page 21: Model BasedDesign ofDigital Communication Systems

© Prof. Dr.‐Ing. Marcus Purat – BHT Berlin, 2019 / Model‐Based Design of Digital Communication Systems 21

• Attendance is mandatory every week• Active contribution and engagement in course• Functionality of the System• Documented Model and Sources for proper re‐use• Short report (5‐10 pages) per group about the project including

– describing the theory to convert parameters into filter coefficients– verifying the functionality of both the model and the system deployment

using appropriate measurements and diagrams

• Project deadline: last week of lecture period (CW 5, 2019)• Report to be provided

– by 26 January 2020 for 1st assessment period, or alt.– by 15 March 2020 for 2nd assessment period

Course Assessment