Scribd
Upload
157 views38 pages

Signal Processing and Communications

application for matlab and simulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulink

Uploaded by

oveiskntu
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
157 views38 pages

Signal Processing and Communications

application for matlab and simulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulinksimulink

Uploaded by

oveiskntu
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 38

Signal Processing and Communications with MATLAB and Simulink

Giorgia Zucchelli Application Engineer - MathWorks

Key Takeaways
Quickly analyze and develop new algorithms with MATLAB

Accurate system-level multi-domain analysis with Simulink

With MATLAB and Simulink you can quickly design entire systems with better performance

Motivations
Quickly analyze and develop new algorithms with MATLAB
>> Evaluating innovative ideas is time-consuming

Accurate system-level multi-domain analysis with Simulink


>> Modeling implementation constraints requires specific knowledge

With MATLAB and Simulink you can quickly design entire systems with better performance
>> Optimizing the tradeoff between reuse and innovation is challenging

MATLAB for Signal Processing


Digital Filter Design Fixed-point in MATLAB and more

MATLAB for Signal Processing: Whats New


Multirate Digital Filter Design Fixed-point in MATLAB for streaming applications and more

Challenges: Digital Filter Design


During design:
Is it meeting the specs?

During implementation:
How can I minimize cost? Which filter structure will be optimal? Will it work correctly on the target hardware? What can I trade off?

Digital Filter Design in MATLAB


Design FIR and IIR filters
Many frequency responses Optimized design methods

Implement filters using various filter architectures Visualize filter response Export MATLAB filters into Simulink for system-level simulation

Digital Filter Design Tradeoffs


Example: high speed, low pass decimation filter
Specifications:
Input sampling freq = 88.4kHz Output sampling freq = 11.05kHz Decimation factor = 8 Passband ripple = 0.1dB Stopband attenuation = 90dB Passband = 3000Hz Stopband = 3200Hz
8

Takeaways: Digital Filter Design in MATLAB


Different filter responses Many optimized design methods Control of the filter architecture Evaluation of tradeoffs between performances, costs and specs Automation of the design process Rapid design iterations Visualization of filter characteristics

Challenges: Fixed-Point Design


Dynamic analysis of data range Finite word length = quantization error
Overflow (overload distortion) Underflow (granular noise)

Algorithms supporting fixed-point data type


7+1=8 bit: 7 word length & 1 fractional bit Range =[-64 63.5) Step =1/2
s 32 16 8 4 2 1 1/2 1/4 1/8 1/16 1/32

7+1=8: 7 bit word length & 5 fractional bits Range =[-4 3.9688) Step = 1/32

10

MATLAB for Fixed-Point


Represent fixed-point data type Analyze quantization effects
Built-in logging and visualizations

Accelerate execution of fixed-point code System objects for more fixed-point functions

11

New in R2010a: System Objects


MATLAB objects that represent time-based and data-driven algorithms, sources, and sinks System objects enable streaming in MATLAB Support of fixed-point data type and automatic C code generation Made available by:
Signal Processing Blockset Video and Image Processing Blockset Communication Blockset
12

MATLAB is Best at Batch Processing


All the data Work on all the data at once

Deliver all at once


13

Many Systems Demand Stream Processing


All the data Incremental delivery

All the data is not available at once


Limited memory footprint Real-time requirements

Typical applications
Communications simulation Audio / video processing Data acquisition

14

Example: Filtering of an Audio Stream


filename = 'dspafxf_8000.wav'; [audio Fs] = wavread(filename); filt = fir1(40, 0.8, 'high'); audiofilt = filter(filt,1,audio); wavplay(audiofilt,Fs);

15

Filtering in MATLAB
filename = music.wav'; [audio Fs] = wavread(filename); filt = fir1(40, 0.8, 'high'); audiofilt = filter(filt,1,audio); wavplay(audiofilt,Fs);
Must wait for all data to be processed before listening to results Loads entire dataset into workspace

audio data uses more space than needed (double vs. uint16) Overall, code uses several copies of the audio dataset in memory

16

Stream Processing in MATLAB Today


%% Streaming the MATLAB way % set up initializations filename = music.wav'; Fs = 8000; info = mmfileinfo(filename); num_samples = info.Duration*Fs; frame_size = 40; bLP = fir1(12, 0.8, low'); zLP = zeros(1,numel(bLP)-1); output = zeros(1,num_samples);

Explicit state management requires programmer to manage details that should be implicit

Indexing is tedious and %% Processing in the loop error prone index= 1; while index < (num_samples-frame_size+1) data = wavread(filename,[index index+frame_size-1]); [datafilt, zLP] = filter(bLP,1,data,zLP); output(index:index+frame_size-1) = datafilt; index = index + frame_size; end Need to maintain output buffer wavplay(output,Fs); because wavplay requires all data before being called
17

Stream Filtering with System Objects


%% Streaming with System Objects before use % set up initializations filename = music.wav'; hFilter = dsp.DigitalFilter; hFilter.TransferFunction = 'FIR (all zeros)'; hFilter.Numerator = fir1(12, 0.8, low'); hAudioSource = dsp.AudioFileReader(filename, ... 'SamplesPerFrame',40, 'OutputDataType','double'); hAudioOut = dsp.AudioPlayer('SampleRate', 11050); %% Processing in the loop while ~isDone(hAudioSource) data = step(hAudioSource); datafilt = step(hFilter, data); step(hAudioOut, datafilt); end

Initialize objects

Many ways to set object properties Source and FIR filter states are implicit Audio player runs in-the-loop with the current frame, avoiding lengthy buffer
18

In-the-loop code is much simpler

No management of indexing

Batch Processing
Load the entire video file and process it all at once

MATLAB

Ease of Use Memory


Batch Video Algorithm

Source

19 19

Traditional Stream Processing in MATLAB


Load a video frame and process it before moving on to the next frame Manually maintain indexing, buffering, states
MATLAB Ease of Use Memory

Stream Source

Stream Processing

20

Stream Processing with System Objects


Load a video frame and process it before moving on to the next frame Implicit indexing, buffering, handling of states
MATLAB Ease of Use Memory

Stream Source

Stream Processing

21

Fixed-Point Algorithms in MATLAB


Represent fixed-point data types in MATLAB as fi objects Run simulation in floating-point or fixed-point modes using data type override Log min, max and overflow

22

Takeaways: Fixed-Point System Objects


Stream processing in MATLAB
Easier to write and be correct the first time Improves handling of large data sets

Fixed-point modeling
All relevant objects support fixed-point data types Compatible with Fixed-Point Toolbox

C-code generation
Most objects support code generation using EMLC Compatible with Embedded MATLAB

23

Simulink for Signal Processing


Systems with complex timing System-level simulation and more

24

Simulink for Signal Processing: Whats New


Mixed-signal systems with complex timing System-level simulation including RF and more

25

Challenges: Mixed-Signal Systems


Anticipate physical constraints
Analog and digital electronics

Complex timing:
Continuous and discrete timing Feedback loops Threshold crossing Asynchronous behavior Concurrent paths

26

Mixed-Signal Modeling with Simulink


Design embedded systems:
Use the most suitable modeling approach Anticipate physical impairments (mixed-signal) Define the system architecture

Verify embedded systems:


Analyze close-loop behavior

27

Takeaways: Mixed-Signal Simulation


Simulation of continuous and discrete signals Multi-rate digital signals with arbitrary sample rates Complex timing
Built-in notion of concurrency Detect zero-crossings and discontinuities Enable feedback loops Asynchronously triggered blocks

Share the MATLAB workspace


28

Challenges: RF System-Level Simulation


Model RF front-ends:
Without being an expert With acceptable simulation speed

Integrate baseband and RF simulation


Develop a system-level view

29

New in R2010b: SimRF


Simulation Speed Fidelity

Circuit envelope analysis


Multi-carrier systems and arbitrary architectures

Equivalent baseband models


Single carrier super-heterodyne cascaded systems (former RF Blockset technology)

Complex baseband models


Mathematical analytical models

30

Former RF Blockset Technology


Single carrier simulation of cascaded RF systems

Linear elements are modeled with baseband-complex equivalent descriptions Nonlinear elements are described by means of (static) AM/AM AM/PM characteristics Elements are cascaded for fast single carrier simulations in the time domain

31

Modeling Paradigm: RF Blockset


From passband-real signal to baseband-complex equivalent signal
Pass-band transfer function

MHz GHz 0

fc Bandwidth = 1/Ts

Baseband-complex equivalent transfer function


-0.5/Ts

frequency
+0.5/Ts

RF Blockset Baseband equivalent time-domain impulse response


Number of sub bands (freq. resolution) equals length of impulse response
0

time
32

New SimRF Technology


Multi-carrier simulation for arbitrary topology of RF systems

Use circuit-envelope analysis for multi-carrier linear and non-linear elements


Extend in band analysis to multiple bands

Based on Simscape to model networks of physical components


Possibility to build arbitrary topologies Possibility to probe within the network

Enables extended interferers and spurs analysis at system-level

33

Simulation Paradigm: SimRF


Envelope analysis of the modulations centered on multiple carriers
Complex envelope of modulated input signals

MHz GHz 0

fc1

fc2 frequency

Circuit-envelope analysis
MHz GHz 0

fc2-fc1

fc1

fc2

fc2+fc1 frequency

SimRF Complex envelope response around selected carrier


MHz GHz 0

fc2+fc1 frequency
34

RF Modeling with Simulink


Design embedded systems:
Use the best modeling approach Anticipate physical impairments (RF) Define the system hierarchy

Verify embedded systems:


Analyze results in streaming conditions Evaluate system-level performances

35

Takeaways: RF Modeling with Simulink


Complex hierarchical model description Streaming capabilities with frame-based processing
Make use of MATLAB for matrix processing

Use Embedded MATLAB


Increase expressivity Use of legacy-code

36

Design and Implement Signal Processing Systems with MATLAB and Simulink
Algorithm design Fast simulation Architecture exploration Targeting implementation Verification and testing Rapid prototyping

37

Conclusions
Quickly analyze and develop new algorithms with MATLAB

Accurate system-level multi-domain analysis with Simulink

With MATLAB and Simulink you can quickly design entire systems with better performance

38

You might also like