Main Content

Modeling and Testing an LTE RF Transmitter

This example shows how to characterize the impact of radio frequency (RF) impairments in an LTE transmitter. The example generates the baseband LTE waveform, which consists of an E-UTRA test model (E-TM), by using LTE Toolbox™ and models the RF transmitter by using RF Blockset™.

Introduction

This example characterizes the impact of RF impairments such as in-phase and quadrature (IQ) imbalance, phase noise, and power amplifier (PA) nonlinearities in the transmission of an LTE waveform. To evaluate the impact of these impairments, the example performs these measurements:

  • Error vector magnitude (EVM): vector difference at a given time between the ideal (transmitted) signal and the measured (received) signal. The example performs EVM measurements according to the specifications in TS 36.104, Annex E [ 2 ].

  • Occupied bandwidth: bandwidth that contains 99% of the total integrated power of the signal, centered on the assigned channel frequency

  • Channel power: filtered mean power centered on the assigned channel frequency

  • Complementary cumulative distribution function (CCDF): probability that the signal's instantaneous power is at a specified level above its average power

  • Peak-to-average power ratio (PAPR): relation between the peak power of the signal and its average power

The example works on a subframe-by-subframe basis. For each subframe, the workflow consists of these steps:

  1. Generate the baseband E-TM waveform using LTE Toolbox.

  2. Import the baseband waveform as an RF signal into the RF Transmitter block implemented using RF Blockset. The model uses an RF intermediate frequency to carry the baseband information in RF Blockset.

  3. Upconvert the waveform to the carrier frequency by using an RF transmitter. You can accurately model the impairments introduced by an actual RF transmitter by using the RF components available in RF Blockset.

  4. Calculate the occupied bandwidth, channel power, CCDF and PAPR by using the Spectrum Analyzer block.

  5. Measure the EVM by demodulating the baseband waveform.

This example performs these operations by using a Simulink® model in which MATLAB® Function blocks carry out the baseband signal processing (steps 1 and 5) and RF Blockset performs the RF transmitter modeling (steps 2 and 3). This model supports Normal and Accelerator simulation modes.

Simulink Model Structure

The model contains three main parts:

  • Baseband Waveform Generation: generates the baseband E-TM waveforms

  • RF Transmission: upconverts the waveform to the carrier frequency

  • Baseband Reception and Measurements: performs the RF measurements and calculates EVM by demodulating the baseband waveform

modelName = 'RFLTETransmitterModel';
open_system(modelName);

Baseband Generation

The E-TM Waveform block generates standards-compliant E-TM waveforms [ 1 ]. For the waveform generation, you can specify the TM name, channel bandwidth, duplexing mode, and cell identity in the E-TM Waveform block.

For more information on how to generate E-TMs, see LTE Downlink Test Model (E-TM) Waveform Generation.

As the example works on a subframe-by-subframe basis, the E-TM Waveform block generates one subframe at a time. Transmitting ten subframes, corresponding to one frame in the case of FDD duplexing mode, lasts 10 ms. If the simulation time is longer than 10 ms, this block transmits the same frame cyclically. The Subframe Counter block stores the number of the currently transmitted subframes. If the simulation time is longer than a frame period, the Subframe Counter block resets to 0 for every new frame simulated.

RF Transmission

The RF Transmitter block is based on a superheterodyne transmitter architecture. This architecture upconverts the received waveform to the carrier frequency and applies passband filtering and amplification. The RF components of this superheterodyne transmitter are:

  • IQ modulator consisting of mixers, a phase shifter, and a local oscillator

  • Bandpass filter

  • Power amplifier

In addition to these components, this RF Transmitter block also includes a variable gain amplifier (VGA) to control the input back-off (IBO) level of the high power amplifier (HPA).

set_param(modelName,'Open','off');
set_param([modelName '/RF Transmitter'],'Open','on');

The Inport block inside the RF Transmitter converts the complex baseband waveform into an RF signal. You can vary the center frequency of this RF signal by modifying the Intermediate frequency (MHz): parameter of the RF Transmitter block mask (in this example, the default frequency is 70 MHz). The Outport block converts the RF signal back into complex baseband.

You can use an Input Buffer before the RF Transmitter block to reduce the number of samples sent to the RF Transmitter. For simplicity, the Input Buffer in the current configuration sends one sample at a time, resulting in the RF Transmitter being sample-based, i.e. it consumes one sample at a time.

Before sending the samples onto the Decode Subframe block, the Output Buffer (after the RF Transmitter) collects all samples within a subframe.

The Delay block, which is connected to the Subframe Counter block, accounts for buffer-induced delays. As the duration of the delay is equivalent to the transmission of a subframe, the Decode Subframe block does not demodulate the samples received during the first subframe period.

You can configure the RF Transmitter components by using the RF Transmitter block mask.

The RF Transmitter block exhibits typical impairments, including:

  • I/Q imbalance as a result of gain or phase mismatches between the parallel sections of the transmitter chain dealing with the IQ signal paths

  • Phase noise as an effect directly related to the thermal noise within the active devices of the oscillator

  • PA nonlinearities due to DC power limitation when the amplifier works in the saturation region

Baseband Reception and Measurements

The Decode Subframe block recovers and plots the PDSCH symbols in the Constellation Diagram by performing OFDM demodulation of the received subframe, channel estimation, and equalization. This block also performs EVM measurements according to the specifications in TS 36.104, Annex E [ 2 ]. These specifications suggest measuring the EVM at two locations in time (low and high), where the low and high locations correspond to the alignment of the FFT window within the start and end of the cyclic prefix, respectively. For more information regarding how to measure EVM, see PDSCH Error Vector Magnitude (EVM) Measurement.

This example carries out EVM measurements averaged over the allocated PDSCH symbols within a subcarrier, resource block, OFDM symbol, subframe, frame, and overall grid. The EVM per subframe is measured for the high and low locations in time while the other results only depict the highest EVM of both locations.

The Spectrum Analyzer block provides additional measurements such as occupied bandwidth, channel power, CCDF and PAPR.

Effect of Power Amplifier

To characterize the impact of the HPA in the EVM evaluation, you can measure the amplitude-to-amplitude modulation (AM/AM) of the HPA. The AM/AM refers to the output power levels in terms of the input power levels. The helper function hPlotHPACurveLTE displays the AM/AM characteristic of the HPA selected for this model.

hPlotHPACurveLTE();
figHPA = gcf;

P1dB is the power at 1 dB compression point and is normally used as a reference when selecting the IBO level of the HPA. You can see the HPA impact on the RF transmitter by analyzing the EVM results for different operating points of the HPA. For example, compare the case when IBO = 12 dB, corresponding to HPA operating in the linear region, with the case when IBO = 6 dB, corresponding to HPA starting to operate in the nonlinear region. The gain of the VGA controls the IBO level. To keep a VGA linear behavior, select gain values lower than 20 dB.

  • Linear HPA (IBO = 12 dB). To operate at an IBO level of 12 dB, set the Available power gain parameter of the VGA block to 0 dB. To simulate a whole frame, run the simulation long enough to capture 10 subframes (Stop Time equal to 10 ms). During simulation, the model displays the spectrum and the constellation diagram.

set_param([modelName '/RF Transmitter'],'vgaGain','0');
sim(modelName);
Low edge EVM, subframe 0: 1.433%
High edge EVM, subframe 0: 1.472%
Low edge EVM, subframe 1: 1.317%
High edge EVM, subframe 1: 1.356%
Low edge EVM, subframe 2: 1.514%
High edge EVM, subframe 2: 1.541%
Low edge EVM, subframe 3: 1.523%
High edge EVM, subframe 3: 1.554%
Low edge EVM, subframe 4: 1.297%
High edge EVM, subframe 4: 1.313%
Low edge EVM, subframe 5: 1.105%
High edge EVM, subframe 5: 1.146%
Low edge EVM, subframe 6: 1.295%
High edge EVM, subframe 6: 1.332%
Low edge EVM, subframe 7: 1.868%
High edge EVM, subframe 7: 1.875%
Low edge EVM, subframe 8: 1.902%
High edge EVM, subframe 8: 1.912%
Low edge EVM, subframe 9: 1.586%
High edge EVM, subframe 9: 1.600%
Averaged low edge EVM, frame 0: 1.507%
Averaged high edge EVM, frame 0: 1.531%
Averaged EVM frame 0: 1.531%
Averaged overall EVM: 1.531%

According to TS 36.104 [ 2 ], the maximum EVM when the constellation is 64-QAM is 8%. As the overall EVM is around 1.5%, this architecture falls within the requirements of TS 36.104 [ 2 ].

  • Nonlinear HPA (IBO = 6 dB). To operate at an IBO level of 6 dB, set the Available power gain parameter of the VGA block to 6 dB.

set_param([modelName '/RF Transmitter'],'vgaGain','6');
sim(modelName);
slmsgviewer.DeleteInstance();

% Restore to default parameters
set_param([modelName '/RF Transmitter'],'vgaGain','0');
Low edge EVM, subframe 0: 2.751%
High edge EVM, subframe 0: 2.776%
Low edge EVM, subframe 1: 2.562%
High edge EVM, subframe 1: 2.579%
Low edge EVM, subframe 2: 2.621%
High edge EVM, subframe 2: 2.641%
Low edge EVM, subframe 3: 2.621%
High edge EVM, subframe 3: 2.640%
Low edge EVM, subframe 4: 2.542%
High edge EVM, subframe 4: 2.547%
Low edge EVM, subframe 5: 2.307%
High edge EVM, subframe 5: 2.325%
Low edge EVM, subframe 6: 2.515%
High edge EVM, subframe 6: 2.535%
Low edge EVM, subframe 7: 3.465%
High edge EVM, subframe 7: 3.466%
Low edge EVM, subframe 8: 3.385%
High edge EVM, subframe 8: 3.383%
Low edge EVM, subframe 9: 2.913%
High edge EVM, subframe 9: 2.916%
Averaged low edge EVM, frame 0: 2.795%
Averaged high edge EVM, frame 0: 2.806%
Averaged EVM frame 0: 2.806%
Averaged overall EVM: 2.806%

Compared to the previous case, the constellation diagram is more distorted. In terms of measurements, the overall EVM, around 2.8%, still falls within the requirements of TS 36.104 [ 2 ].

If you want to push the HPA to operate further in the nonlinear region, you will need to oversample the signal (around 5 times its baseband bandwidth) so that the simulation bandwidth is large enough to capture in-band spectral regrowth.

Summary and Further Exploration

This example demonstrates how to model and test the transmission of an LTE waveform. The RF Transmitter consists of a bandpass filter, amplifiers and an IQ modulator. The example highlights the effect of HPA nonlinearities on the performance of the RF Transmitter. You can explore the impact of altering other impairments as well. For example:

  • Increase I/Q imbalance by using the I/Q gain mismatch (dB) and I/Q phase mismatch (Deg) parameters on the IQ Modulator tab of the RF Transmitter block.

  • Increase the phase noise by using Phase noise offset (Hz) and Phase noise level (dBc/Hz) parameters on the IQ Modulator tab of the RF Transmitter block.

Additionally, you can check the CCDF and PAPR measurements by using the Spectrum Analyzer window: select the CCDF Measurements button on the toolbar.

The RF Transmitter is configured to work with the default values of the E-TM Waveform block and with the LTE carrier centered at 2140 MHz. This carrier is within the E-UTRA operating band 1 [ 3 ]. If you change the carrier frequency or the values in the E-TM Waveform block, you may need to update the parameters of the RF Transmitter components, as these parameters have been selected to work for the default configuration of the example. For instance, a change in the carrier frequency requires revising the bandwidth of the filters. Modifying the waveform bandwidth may require updating the Impulse response duration and Phase noise frequency offset (Hz) parameters of the IQ Modulator block. The phase noise offset determines the lower limit of the impulse response duration. If the phase noise frequency offset resolution is too high for a given impulse response duration, a warning message appears, specifying the minimum duration suitable for the required resolution. For more information, see IQ Modulator (RF Blockset).

This example could be the basis for testing E-TM waveforms for different RF configurations. You can replace the RF Transmitter block by another RF subsystem of your choice and configuring the model accordingly.

References

  1. 3GPP TS 36.141 "E-UTRA; Base Station (BS) conformance testing" 3rd Generation Partnership Project; Technical Specification Group Radio Access Network.

  2. 3GPP TS 36.104 "E-UTRA; Base Station (BS) radio transmission and reception" 3rd Generation Partnership Project; Technical Specification Group Radio Access Network.

  3. 3GPP TS 36.101. "E-UTRA; User Equipment (UE) radio transmission and reception." 3rd Generation Partnership Project; Technical Specification Group Radio Access Network.