ELM 460 Software Based Communication Spring 2024

ELM 460 Lab 3: Pulse Shaping and Matched Filtering

1. Introduction

The objective of this lab session is to introduce the basic concepts of pulse shaping, matched filter and
pulse alignment. In digital communication, the message in digital must be converted into an analog signal
to be transmitted. This conversion is done by the pulse shaping filter, which changes each symbol into a
suitable analog pulse. Designing the pulse shape is important because the spectrum of the pulse dictates
the spectrum of the whole transmission. However, to confine the spectrum, we have to smooth the pulse
with slow transition. This requires the pulse extending beyond a symbol time, which introduces inter-
symbol interference (ISI). Thus, there is a trade-off between the bandwidth and the ISI of the pulse shape.

After transmission, the matched filter helps to recapture the symbols from the received pulses. The
matched filter is aimed at reducing the sensitivity of noise by maximizing the signal-to-noise ratio (SNR)
and minimizing the ISI at the receiver.

In a more realistic channel model, the receiver never knows the precise arrival time of the pulse. Therefore,
the receiver must know the exact symbol timing in order to correctly demodulate the transmitted symbols
from the transmitter.

1.1. Contents

1. Introduction
2. Pulse Shaping
3. Matched Filtering
4. Pulse Alignment (Symbol Timing Recovery)

1.2. Report

Submit the answers, figures and the discussions on all the questions. The report is due as a hard copy at the
beginning of the next lab.

*Since the simulation will be performed in Lab 3, do not create separate VI files but create both
Tx.vi and Rx.vi in the same VI file.

*Sentences with red subtitles will not be done, but they are necessary for the real time system and
should be learned.
1

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

2. Key Components

The following components will be used in this lab session.

• MT Generate Bits (Fibonacci, PN Order) (RF

Communications  Modulation  Digital): Generates
Fibonacci pseudonoise (PN) bit sequences. Use this block to
generate the message bits to transmit.

• Upsample (Signal Processing  Signal Operation): Inserts

zeros in a sequence according to a specific upsampling factor.

• MT Generate Filter Coefficients (RF

Communications  Modulation  Digital
 Utilities): Calculates filter coefficients for
pulse-shaping and matched filters applied by
the digital modulation nodes and demodulation
nodes.

• Convolution (Signal Processing  Signal Operation): Computes the

convolution of two sequences.

• FFT Power Spectrum and PSD (Waveform 

Analog Waveform  Waveform Measurement): Computes
the averaged auto power spectrum of a time-domain signal.

• MT Format Constellation (RF Communications 

Modulation  Digital  Visualization): Prepares a signal
for presentation on a graph that shows the detected symbol
locations and the transitions between those symbols. The VI
specifies a complex-valued waveform and displays a
constellation plot of the waveform contents.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

3. Pulse Shaping

Start by loading “Tx_template.vi” which you saved from lab 2. In this section, we will build a transmitter that
performs pulse shaping and upsampling to create the transmit waveform. Binary phase-shift-keying (BPSK) is used
for modulation scheme. BPSK is the simplest form of phase-shift-keying (PSK). The data is conveyed in the phase
of the carrier signal (i.e. “1” for 0 degree and “0” for 180 degree in BPSK).

1. Set the default parameter values in the panel:

o Carrier frequency: 1G
o IQ rate: 400k
o Gain: 0
o Active antenna: TX1

2. Create two Integer-type controls: Message length, and Symbol rate. Set the message length to 1000 and
the symbol rate to 10k. The IQ rate and the symbol rate will determine the upsampling factor. We will
upsample the symbols before performing pulse shaping.
o Message length: 1000
o Symbol rate: 10k

3. (Create data) Add MT Generate Bits to generate a Fibonacci pseudonoise (PN) bit sequence. The result
bit sequence is a sequence of binary digits 0 and 1. Notice that the default value of “PN sequence order”
is 7. Use the default value. Connect the “Message length” control to “total bits” port.

4. (Bits to Symbols: BPSK) Use Multiply and Decrement or Case to map the binary 0 and 1 into the symbols
-1 and 1.

5. Create a DBL-type indicator and name it as Upsampling Factor. Use the appropriate math blocks to
calculate the upsampling factor (Upsampling factor is defined by sampling rate/symbol rate).

6. (Upsampling) Add Upsample. Use the symbols and the upsampling factor in the previous steps, and
generate the upsampled symbols.

7. (Pulse Shaping) Add MT Generate Filter Coefficients. Create “pulse shaping filter” terminal by right
clicking the “pulse shaping filter” port on the component choosing Create Control. Similarly, create
“modulation type” terminal by right clicking and choosing Create Constant. Set it to “PSK”. Wire the
upsampling factor to “pulse shaping samples per symbol” port. Use “pulse shaping filter coefficients” for
the output. Notice that the “filter length” is 8 and the filter parameter (roll-off factor for raised cosine or
root raised cosine) is 0.5 by default. The filter length means the desired length, in symbols, of the pulse
shaping filter. Use the default values for now.

8. (Convolution) Use Convolution to convolve two sequences: the upsampled symbols and the pulse shaping
filter coefficient. The below figure summarizes the process you’ve done so far.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

9. (subVI: norm1D) We will create a subVI that normalizes the filtered symbols by the maximum value of
its absolute value. This will ensure the filtered symbols are in between -1 and 1. Create a VI in the project
and name it as “norm1D.vi”. Use Absolute Value and Array Max and Min. Click “Edit Icon” and add
the input/output ports. Keep this subVI and we will use it for the future labs.

10. Go back to your transmitter diagram and drag-drop “norm1D” subVI. Use this subVI to normalize the
output of the convolution block. Then, wire the normalized sampled to input of convolution at RX. Then,
wire the normalized samples to niUSRP Write Tx Data.

11. (Power Spectrum and IQ Data Constellation) We will measure the power spectrum and the IQ
constellation of the outgoing transmitting waveform. Use Build Waveform and construct the time-
domain waveform from the Tx data. Calculate the sampling period (dt) from the IQ rate (use Reciprocal
block) and wire to Build Waveform. Add FFT Power Spectrum and PSD and set the function
configuration to “Power Spectrum” and “Continuous”. Check “dB On” box in the terminals. Create an
indicator by right-clicking the “power spectrum” port of the block. Add MT Format Constellation. Wire
the upsampling factor to “samples per symbol” port. Select “bundle by name” from cluster for adding to
input of waveform. Wire the normalized samples to input of “bundle by name” with “IQ Waveform
Cluster”. After wire “bundle by name” and MT Format Constellation. Create an indicator by right-
clicking the “IQ Data Constellation” port of the block.

*IQ Waveform Cluster = This subVI will be given you during lesson.

12. Save the VI as “BPSKTx.vi”.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

3.1. Questions

3.1.1. The PN sequence order for PN bit sequence was 7. What is the period of the PN bit sequence?

3.1.2. Run the code with different pulse shaping filters: None (rectangular) and Root Raised Cosine. Save
the baseband waveform, the spectrum and the IQ data constellation of each filter. Compare the
bandwidth, the sidelobes for each of the filters. (Use the cursors to report quantitative values.)

3.1.3. Set the pulse shape to Root Raised Cosine and change the filter length to 2. Compare with filter
length 8 and discuss their baseband waveform, spectrum and IQ data constellation.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

4. Matched Filtering

Start by loading “Rx_template.vi”. In this section, we will build a receiver that performs matched filtering. The
receiver built in this lab does not perform the complete receiver capability such as frame synchronization, symbol
mapping, and etc. We will resume the rest in the next lab session.

1. Set the default parameter values in the panel:

o Carrier frequency: 1G
o IQ rate: 1M
o Gain: 0
o Active antenna: RX2

2. Create two Integer-type controls: Message length, and Symbol rate. Set the message length to 1000 and
the symbol rate to 10k. The IQ rate and the symbol rate will determine the downsampling factor. We will
upsample the symbols before performing pulse shaping.
o Message length: 1000 (one frame)
o Symbol rate: 10k

3. (Decimating Factor) The receiver will decimate the samples to obtain the symbols. Calculate the
decimating factor similar to the upsampling factor. Note that the two factors do not have to be the same.

4. We will capture 1 frame of data. Calculate the number of samples to be captured in the receiver (Hint: Use
the decimating factor and the message length) and wire to “number of samples” of niUSRP Fetch Rx
Data. Remove the while loop around the block for a single-shot acquisition. The receiver will stop once it
acquires a frame of data. Make sure the niUSRP Fetch Rx Data block uses CDB WDT type in the function
configuration.

5. (Matched Filtering) Find MT Generate Filter Coefficients. Create “pulse shaping filter” terminal by
clicking Create Constant in the Terminals section of the component. Change the constant to Control. Create
“modulation type” terminal by clocking Create Constant. Set to “PSK”. Wire the decimating factor found
in the previous step to “matched samples per symbol”. Use “matched filter coefficients” for the output.

6. (Convolution) Use Convolution to convolve two sequences: the Rx samples and the pulse shaping filter
coefficient. Remember the niUSRP Fetch Rx Data block outputs CDB WDT type and the convolution
block takes CDB array type, so we need to extract “Y” (CDB Array) from the Rx data. Use Waveform
Properties and set it to “Set All to Read”. Wire “Y” to one input port of the convolution and the matched
filter coefficient to the other port.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

7. (Downsampling & BPSK demodulation) Find Decimate block and wire the matched filtered samples to
the input array port. Connect the decimating factor. Use Complex to Real and Imaginary block to take
the real value of the symbols. Recover the bits by returning “0” if the symbol is less than 0 and “1” if the
symbol is greater or equal to 0 (use Greater or Equal? and Boolean to Integer blocks). Display the
recovered bits using an array indicator (name it as “unaligned bits”).

8. (Power Spectrum and IQ Data Constellation) Take the real values of the matched filtered samples.
Similar to Tx, show the Baseband Waveform, use FFT Power Spectrum and PSD and MT Format
Constellation to display the power spectrum and IQ constellation (name it as “unaligned”).

9. Save the VI as “BPSKRx.vi”.

4.1. Questions

4.1.1. Change the filter parameter of a Root Raised Cosine (both Tx and Rx) to 0.2 and 1. Does the
amount of ISI increase or decrease with increasing the filter parameter? Does the bandwidth of
received signal increase or decrease with increasing the filter parameter? (Connect a power
spectrum block to the Rx samples to measure the spectrum)

4.1.2. Compare IQ data constellation, Power Spectrum and Waveform between Tx and Rx (for pulse
shaping filters: None (rectangular), Raised Cosine and Root Raised Cosine).

4.1.3. Review IQ data constellation. What problem do you see if the pulse is not aligned?

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

5. Pulse Alignment

1. Duplicate “BPSKTx.vi” and rename it as “BPSKTx_shortPN.vi”.

2. Modify the MT Generate Bits and change its function configuration to “User Defined”. Create an array
constant for “user base bit pattern”. Input the bit sequence “1011100”. Create an indicator to see the output
bit stream.

3. Implement the following subVI and rename as “PulseAlign.vi”. Click “Edit icon” and create all the inputs
and the outputs.

4. (Power Spectrum and IQ Data Constellation) By use “PulseAlign.vi” to align the matched filtered
samples, take the real values of the matched filtered samples. Similar to Tx, show the Baseband
Waveform, use FFT Power Spectrum and PSD and MT Format Constellation to display the power
spectrum and IQ constellation (name it as “aligned”).

5. Demodulate the aligned samples to recover the bits (name it as “aligned bits”).

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.
ELM 460 Software Based Communication Spring 2024

5.1. Questions

5.1.1. Explain how “PulseAlign.vi” works.

5.1.2. Compare “unaligned IQ constellation” and “aligned IQ constellation”.

5.1.3. Compare “unaligned bits” and “aligned bits”. Can you detect the transmitted bits “1011100”?

5.1.4. Give a slight frequency offset in Tx, i.e. 1kHz offset by setting carrier frequency as 1.000001G.
What can you see in the eye-diagram and the recovered bits? Does the alignment still work?

5.1.5. Save all the VIs and they Project since they will be used in subsequent labs.

In this laboratory study, the document of the ECE463: IoT and Software Defined Radio Lab course at the University of
Illinois Urbana-Champaign was used.