This version is outdated by a newer approved version.
This version (17 Dec 2013 17:07) was approved by Robin Getz.The Previously approved version (18 Oct 2013 23:16) is available.
This version (17 Dec 2013 17:07) was approved by Robin Getz.The Previously approved version (18 Oct 2013 23:16) is available.This is an old revision of the document!
Table of Contents
Introduction
This is work in progress, and is not yet 100% complete.
In this section, we provide the reference data files of several frequently used digital baseband modulation schemes and their corresponding Simulink models. With these data files, you are ready to transmit them using the AD9361 found on the FMComms2 / FMComms3 and test your board. By changing the given Simulink models, you can easily generate your own data files.
In order to run the models on this page, your MATLAB license needs to include the following components:
- MATLAB Version 8.2
- Simulink Version 8.2
- Communications System Toolbox Version 5.5
- DSP System Toolbox Version 8.5
QPSK
PSK is one type of the phase modulation (PM) schemes. In this section, two flavors of QPSK data are generated, one with raised cosine pulse shaping filters, and the other one without.
Since we would like to transmit the QPSK at a sample rate of 20M Symbols/s with AD9361, the sample time specified in “Random-Integer Generator” block is 1/(20e6). The users can specify their own sample rate.
QPSK without Pulse Shaping
The figure below shows a QPSK transmission model without any pulse shaping filter. Since there are two independent channels on the Tx paths, we model both of them. However, for simplicity, we make them identical. The transmitted signals are saved to workspace.
Model
You can download the Simulink model from below:
In order to observe the transmitted spectrum, we use a “Spectrum Analyzer” block. Since there is no pulse shaping on the Tx path, observe how rectangular-pulse QPSK symbols occupy the entire signal bandwidth in the frequency domain, potentially disturbing adjacent channels in the RF spectrum. Clearly this is not an efficient use of signal bandwidth.
Data
You can download the generated data from below:
- For each line, the data is in the order of [I1, Q1, I2, Q2].
- The first line must be the word “TEXT”.
- In Windows, you are suggested to open the txt file using WordPad to show the line breaks properly.
QPSK with Pulse Shaping
The QPSK symbols observed in the previous model cannot be transmitted over the air as perfect rectangular pulses, which would require infinite signal bandwidth. They must first be smoothed in the time domain, in order to limit the signal bandwidth in the frequency domain. Consequently, the first stage of the Tx signal chain is often a pulse-shaping filter, followed by subsequent stages of interpolation filters to increase the sampling rate to the DAC.
In this section, we will employ a root-raised cosine pulse-shaping filter to smooth the QPSK pulses in the time domain, limiting the signal bandwidth to 10 MHz in the frequency domain.
Model
The figure below shows a QPSK transmission model with the pulse shaping filters. Since pulse shaping filters are often distributed as a matched pair between transmitter and receiver, we use the filter shape of ‘Square root’1).
You can download the Simulink model from below:
The data rate is defined in the parameter Fs.
- When the ‘square root raised cosine’ response of the transmit filter is convolved in the time domain with that of the receive filter, the aggregate response is a ‘raised cosine’.
- It is the ‘raised cosine’ that has the desirable quality of mitigating inter-symbol interference in the time domain.
In order to observe the effect of the pulse shaping filters, we use a “Spectrum Analyzer” block, as well as several “Constellation” blocks.
We have the following observations:
- The transmitted signal bandwidth is constrained to 10 MHz in the frequency domain.
- Given the transmit filter, if there is no matched receive filter, the received symbols are completely noise (refer to “Rx Without Receive Filter”).
- Given the transmit filter, if there exists matched receive filter, the received symbols are perfect (refer to “Rx With Receive Filter”).
Data
You can download the generated data from below:
The data rate is 30.72 MSPS.
It is in the same format as the data in previous section.
Data Verification
Since there is no match filter on AD9361 receive path, the data obtained from AD9361 receiver side does not show the constellation of QPSK clearly. Therefore, by looking at the ADI IIO Oscilloscope2)., it is difficult to see whether the received data is valid or not, as shown in the figure below.
However, with the save data function of the IIO application, we can now save the received data in MATLAB compatible format (.mat file), let the data pass through a match filter in Simulink, and then verify whether the received data is a QPSK or not. The save data function can be accessed via “File”-“Save As”, as shown in the figure below. There are several data formats available. Since we want the data to be used in MATLAB, we will pick up the .mat format.
Before you save the data, make sure the sampling rate of ADC and DAC is set at 30.72 MSPS, as shown in the figure below. Otherwise, it may incur some problems. In this panel, there are quite a few other parameters you can tune, in order to get the optimal transmission and reception performance for your system.
Given the received data, we can now proceed to the Simulink receiver model to verify the data. After you launch MATLAB and the receiver model qpsk_receiver.slx, the next step is to load the .mat file in workspace, because the Signal From Workspace blocks use the data as input.
In this model, the data rate is the same as the generated data rate (30.72 MSPS), and the receive filter is the match of the transmit filter. When looking at the IIO scope, we found the received signal from the AD9361 shows a 45 degree rotation compared to the transmitted signal, so a Phase/Frequency Offset block is employed here to compensate for the phase offset. According to the constellation plot, it is very clear that the received data is a QPSK, so the AD9361 Tx/Rx chain gets verified.
You can download the saved mat data file from below:
You can download the receiver model from below:
The data rate is defined in the parameter Fs and the phase offset is defined in the parameter PhaseOffset.
A few tips for verification, which can be applied to the other modulation cases:
- Make the data rate consistent in data generation and receiver.
- Set the IIO parameters properly, such as ADC and DAC sample rate.
- Compensate for the phase offset occurred in AD9361.
16 - QAM
QAM is one type of the amplitude modulation (AM) schemes. In this section, we use “Rectangular QAM Modulator Baseband” block to modulate the input random integer. If you would like a lower or higher order of QAM modulation, you only need to change the “M-ary number” of this block.
Model
The figure below shows a 16-QAM transmission model with the square root raised cosine pulse shaping filters. In order to observe the effect of the pulse shaping filters, we use a “Spectrum Analyzer” block, as well as several “Constellation” blocks on one of the channels.
You can download the Simulink model from below:
The data rate is defined in the parameter Fs.
Data
You can download the generated 16-QAM data from below:
It is in the same format as the data in previous sections and the data rate is 20M.
Data Verification
Since there is no match filter on AD9361 receive path, the data obtained from AD9361 receiver side does not show a clear constellation of 16-QAM. Actually, since the order of 16-QAM is higher than that of QPSK in the previous section, the constellation here is even more noisy compared to QPSK, as shown in the figure below.
However, with the save data function of the IIO application, we can now save the received data in MATLAB compatible format (.mat file), let the data pass through a match filter in Simulink, and then verify whether the received data is a 16-QAM or not. Before you save the data, make sure the sampling rate of ADC and DAC is set properly. In this example, we set the sample rate of ADC and DAC 5 times of the generated data rate (20 MSPS), which is 100 MSPS.
You can download the saved mat data file from below:
Given the received data, we can now proceed to the Simulink receiver model to verify the data. After you launch MATLAB and the receiver model qam16_receiver.slx, the next step is to load the .mat file in workspace, because the Signal From Workspace blocks use the data as input. In this model, the data rate is the same as the generated data rate (20 MSPS), and the receive filter is the match of the transmit filter. When looking at the IIO scope, we found the received signal from AD9361 shows a 45 degree rotation compared to the transmitted signal, so a Phase/Frequency Offset block is employed here to compensate for the phase offset. According to the constellation plot, it is very clear that the received data is a 16-QAM, so the AD9361 Tx/Rx chain gets verified.
You can download the receiver model from below:
The data rate is defined in the parameter Fs and the phase offset is defined in the parameter PhaseOffset.
MSK
MSK stands for minimum shift keying. It is one type of the continuous phase modulation (CPM) schemes 3). In this section, we use “MSK Modulator Baseband” block to modulate the input random binary bits. In other words, the input is either 0 or 1.
Model
The figure below shows a MSK transmission model with the square root raised cosine pulse shaping filters. In order to observe the effect of the pulse shaping filters, we use a “Spectrum Analyzer” block, as well as several “Constellation” blocks on one of the channels.
You can download the Simulink model from below:
The data rate is defined in the parameter Fs.
Data
You can download the generated MSK data from below:
It is in the same format as the data in previous sections and the data rate is 20M.
Data Verification
Since there is no match filter on AD9361 receive path, the data obtained from AD9361 receiver side does not show a clear constellation of MSK signals, as shown in the figure below. However, it matches the “Rx Without Receive Filter” plot from Simulink.
With the save data function of the IIO application, we can now save the received data in MATLAB compatible format (.mat file), let the data pass through a match filter in Simulink, and then verify whether the received data is a MSK or not. Before you save the data, make sure the sampling rate of ADC and DAC is set properly. In this example, we set the sample rate of ADC and DAC 5 times of the generated data rate (20 MSPS), which is 100 MSPS.
You can download the saved mat data file from below:
Given the received data, we can now proceed to the Simulink receiver model to verify the data. After you launch MATLAB and the receiver model msk_receiver.slx, the next step is to load the .mat file in workspace, because the Signal From Workspace blocks use the data as input. In this model, the data rate is the same as the generated data rate (20 MSPS), and the receive filter is the match of the transmit filter. According to the constellation plot, it is very clear that the received data is a MSK, so the AD9361 Tx/Rx chain gets verified.
You can download the receiver model from below:
The data rate is defined in the parameter Fs and the phase offset is defined in the parameter PhaseOffset. In this example, there is no rotation shown by the received signal, so PhaseOffset=0.
LTE
Besides the data generated by the basic modulation schemes, it is also possible to transmit the data from some more advanced wireless communication standards. In this section, we will generate some LTE (Long Term Evolution) data from an existing Simulink example.
- You can open the example by typing “LTEPDSCHExample” in the command window.
- You must have Communications System Toolbox license in order to run this example.
Model
The figure below shows a LTE example according to the specifications developed by the Third Generation Partnership Project (3GPP). It highlights only the downlink physical channel (PDSCH) processing. In order to obtain the transmitted data, we add a “Signal to Workspace” block on the transmitter side (circled in red). By double clicking the “Model Parameters” block, we can change the model settings such as channel bandwidth, antenna configuration and etc.
Since LTE is a sophisticated standard, you are encouraged to read the Help Document of this example and its related references to get more information 4).
After you run the model, the transmitted data named “Tx” will be saved in workspace. Since it is a Nx2 complex vector, we need to do some processing on it to make the data format compatible with AD9361 transmission.
You can download the Simulink model and the post-processing m-file from below:
When the data is transmitted and received by real-world hardware, like AD9361 , it is difficult to observe the clear constellation without proper synchronization techniques. However, even with the basic settings, you are still expected to observe the transmit and receive spectrum, which is similar to the following:
Note that the bandwidth shown in the spectrum is approximately 10 MHz, which is consistent with the “Channel bandwidth” setting of the model parameters.
Data
You can download the generated LTE data from below:
It is in the same format as the data in previous sections.
resources/eval/user-guides/ad-fmcomms2-ebz/software/datafiles.1387296423.txt.gz · Last modified: 17 Dec 2013 17:07 by Robin Getz