A brief discussion on the design and testing of software radio

btty038

A brief discussion on the design and testing of software radio [Copy link]

Radio is no stranger to everyone. It is actually radio waves, that is, electromagnetic waves in the radio frequency band that propagate in free space (including air and vacuum). The emergence of new wireless technologies has forced people to use multi-standard and multi-band radios, so software defined radio (SDR) will play a key role in future radio architectures. SDR uses only one hardware front-end, but can change its operating frequency, occupied bandwidth, and different wireless standards by calling different software algorithms. This solution can achieve inexpensive and efficient interoperability between existing standards and frequency bands.
Today's introduction includes the main parts of SDR, highlighting several possible implementation methods for receivers and transmitters. Many of these structures are actually quite old technologies, which have become feasible due to the huge increase in the capacity of digital signal processors. We also introduce the measurement and characterization methods of such devices. SDR usually works in both analog and digital domains, so it is necessary to use mixed-domain equipment for measurement.
The concept of SDR was first reflected in the research done by Mitola in 1995. In this research work, he proposed to create a radio that is completely software-controlled, so that the radio can automatically adjust to several communication scenarios. This concept is shown in Figure 1.

Figure 1. Common implementation of software radio. A signal incident on the antenna port is routed through a circulator to a low noise amplifier (LNA) and then digitized. Demodulation and coding of several modulation formats and access modes can be performed using a digital signal processor (DSP). The transmit chain uses the opposite process: the baseband signal is generated and up-converted in the DSP module, converted to an analog waveform, amplified and bandpass filtered before passing through the circulator and antenna. The
SDR front end consists of standard subsystems used in most receivers and transmitters: modulator and demodulator, frequency converter, power amplifier (PA), and low noise amplifier (LNA). However, modulation and coding as well as operating frequency are controlled by software. Such radios generally rely on digital signal processors (DSP) to achieve their flexibility. SDRs can adjust themselves according to the transmission conditions to minimize interference caused by other signals present in the air interface. The implementation of such a system requires the ability to scan the spectrum from low to high frequencies using software. This concept has motivated many researchers to work on the concept of cognitive radio (CR) proposed by Mitola, in which the radio self-adjusts to the air interface conditions by optimizing the carrier frequency, modulation scheme and radio standard, so as to minimize interference and maintain smooth communication under given conditions. One of the
most promising applications of CR technology is to improve spectrum occupancy by using opportunistic radio, where the radio will use the spectrum that is not occupied by other radio systems at a certain time. In order to implement this ideal solution, the radio should see and understand the complete spectrum or communication status at a specific time. The
motivation behind the SDR concept is not only the high flexibility of adapting the front end to work with any modulation mode, channel bandwidth or carrier frequency simultaneously, but also the potential cost savings by using a fully digital system.
Software Radio Receiver Architectures
We now give a review of several front-end architectures that may be used in SDR receivers. The first structure [Figure 2(a)] is the well-known superheterodyne receiver, in which the signal received by the antenna is converted to baseband by two down-conversion mixers, bandpass filtered and amplified. The baseband signal is converted to the digital domain for processing. Since the first mixing process is from RF to IF, image rejection filters must be used before the mixers. Currently, this structure is mostly used in designs with higher RF bands and millimeter wave bands, such as point-to-point wireless links. In these applications, the solution we will discuss next is not practical. In fact, superheterodyne receivers have many substantial problems when used for SDR. Generally, there are many manufacturing technologies involved, which makes it difficult to integrate all components on chip. Also, they are usually designed for a specific channel (in a specific wireless standard). This prevents the reception band from being extended to use signals with different modulation formats and bandwidth occupation. Therefore, the superheterodyne structure is not interesting for use in SDR receivers because of the complexity of extension when receiving in multiple bands.
Another approach is the zero-IF receiver shown in Figure 2(b), which is a simplified superheterodyne structure. As in the previous structure, the RF band of the entire receiver is selected by a bandpass filter and amplified by a low noise amplifier. It is then directly down-converted to DC with a mixer and converted to the digital domain by an analog-to-digital converter (ADC). Compared with the heterodyne structure, this approach significantly reduces the number of analog components and allows the use of filters that are not as stringent as image rejection filters. Therefore, this structure can have a high degree of integration. However, due to the performance requirements of the components, some components are difficult to design. Also, converting the signal directly to DC will cause some problems, such as DC offset. There are other problems related to second-order intermodulation products near DC, and because the output of the mixer is a baseband signal, it is easily corrupted by the large flicker noise of the mixer. Its advantages make it the most commonly used structure in recent radio receivers. Figure 2. (a) A superheterodyne receiver architecture, where the RF signal is received, filtered, amplified, downconverted to an IF frequency, and then filtered and amplified again. The signal is then converted to baseband by a quadrature demodulator, filtered in each path (I and Q), amplified, and subsequently converted to the digital domain. (b) A zero-IF architecture, where the RF signal is filtered, amplified, and directly converted to baseband by a quadrature demodulator. The signal is then filtered, amplified, and digitized. (c) A bandpass sampling receiver, where the signal is filtered, amplified, and sampled by a sample-and-hold circuit, which is usually part of an analog-to-digital converter. The signal is mixed down to the first Nyquist zone, digitized by an analog-to-digital converter, and processed in the digital domain. ADC: analog-to-digital converter, BPF: bandpass filter, FIR: finite impulse response filter, I: in-phase component, LNA: low noise amplifier, LO: local oscillator, LPF: low pass filter, Q: quadrature component; VGA: variable gain amplifier.
Similar to the zero-IF architecture is the low-IF receiver [14], in which the RF signal is down-converted to a non-zero lower or medium IF signal instead of directly converting to DC. In this case, an RF bandpass filter is applied to the incident signal, which is then amplified. The signal is converted to the digital domain by a relatively robust ADC, so that digital filtering can be performed using a DSP to select channels and eliminate in-phase and quadrature (I/Q) imbalance problems in the orthogonal demodulator. This architecture still allows for a high level of integration without the problems of the zero-IF architecture because the desired signals are not near DC. However, in this architecture, the image frequency problem is introduced again, and the power consumption of the ADC is increased due to the higher conversion rate required.
Finally, an alternative to the previously described approach is the bandpass sampling receiver, see Figure 2(c). In this configuration, the received signal is filtered by an RF bandpass filter, which can be a tuned filter or a filter bank. The signal is amplified by a wideband low noise amplifier. The signal is sampled by a high sampling rate analog-to-digital converter and converted to the digital domain for digital processing. This configuration is based on the fact that the energy between DC and the input analog signal bandwidth can be folded into the first Nyquist zone [0, fs/2] by the sampling and holding circuits in the analog-to-digital converter without any down-conversion. This configuration exploits some of the advantages of the sample and hold circuit.
In this case, the RF bandpass signal filter plays an important role because it must reduce all signal energy (essentially noise) outside the Nyquist zone of the desired frequency band, otherwise it will alias with the signal. If no filtering is performed, the signal energy (noise) outside the required Nyquist zone will be folded back into the first Nyquist zone along with the desired signal, resulting in a degradation of the signal-to-noise ratio. This is given by

Where S represents the power of the desired signal, Ni and N0 are the noise in and out of band, respectively, and n is the number of aliased Nyquist zones. The
benefit of this approach is that the required sampling frequency and subsequent processing speed is proportional to the signal bandwidth rather than the carrier frequency. This reduces the number of components.
However, there are some critical requirements. For example, the bandwidth of the analog input signal to the sample and hold circuit (usually within the ADC) must include the RF carrier frequency, which can be a serious problem given the sampling rates of modern ADCs. Clock jitter is also a problem. Also, RF bandpass filtering is required to avoid signal overlap.
Other proposed architectures for SDR receivers include receiving the signal using direct sampling techniques of the RF signal based on discrete-time analog signal processing. These approaches are still in a very immature stage, but they should be studied further due to their potential efficiency in implementing reconfigurable receivers.
Testing of Software Radio Implementations
Experimentation and testing of SDR systems. The key to this discussion is the concept of mixed-domain test techniques, since SDR systems always have one input in the analog domain and another in the digital logic domain. In the SDR concept, the main idea is to push the analog-to-digital/digital-to-analog converter as close to the antenna as possible, as shown in Figure 1. Therefore, fewer signals exist in the analog domain, and the importance of digital signal testing is not reflected in traditional RF system characterization.
Hardware
The instrumentation industry has developed various instruments suitable for SDR characterization, such as mixed-signal oscilloscopes that can operate in both the analog and digital domains. This allows analog and digital signals to be time-synchronized on the same instrument. However, mixed-signal oscilloscopes only provide asynchronous sampling capabilities. This means that, like traditional sampling oscilloscopes, mixed-signal oscilloscopes use their internal clock to sample data. As discussed in [38] and [39], when testing SDR devices (including analog-to-digital converters), accurate estimation of the phase and amplitude of the transfer function requires correlated sampling between the input, output, and clock signals. If these signals are sampled asynchronously, spectral leakage can occur that is sufficient to completely degrade any amplitude and phase information from the SDR. Spectral leakage occurs because the two signals do not share the same time domain grid when performing the necessary Fourier transforms (DFT or FFT) and are therefore uncorrelated with each other.
Other issues that may exist with mixed signal oscilloscopes include, for example, the memory space required to capture the behavioral models. Because these instruments typically use very high sampling rates, a large number of points are required to capture the low/medium symbol rate modulated signals that are commonly used. Therefore, this type of instrumentation cannot fully characterize a complete SDR front end.
Other approaches proposed by the instrumentation industry combine several instruments, including logic analyzers, oscilloscopes, vector signal analyzers, or real-time signal analyzers [40]-[42]. To test an SDR transmitter architecture, these instruments can be built in a configuration similar to that shown in Figure 3. By using reference signals, trigger signals, and markers, one can synchronize measurements between the digital and analog domains and between the time and frequency domains. Typical tests performed using these systems to evaluate the transmit and receive links in an SDR include error vector magnitude (EVM) and adjacent channel power ratio (ACPR) in the signal chain. Figure 3. Equipment used to test a software radio transmitter, where several instruments are used together. A logic analyzer samples the digital logic bits at the output of the digital signal processor (DSP), an oscilloscope analyzes the analog signal after signal reconstruction by the digital-to-analog converter (DAC) and low-pass filter (LPF), and a spectrum analyzer or vector signal analyzer acquires the analog RF signal after the quadrature modulator or after signal amplification.
In [39], the authors discuss the requirements for signal timing and synchronization and propose some solutions, such as embedding a trigger signal in the experimental stimulus. Some important issues remain to be resolved, such as the calibration process of mixed-signal instruments. The analog channels in mixed-signal instruments should ideally be able to measure the reflection coefficient at the input port. Directional couplers should be used to provide a wave-based impedance mismatch calibration representation of the RF signal incident on the device under test. With this information, it is possible to relate the analog input and digital output to find the transfer function of the SDR system, or even a complete behavioral model of the system. It is possible to build such an instrument using off-the-shelf components and algorithms. However, a complete test setup does not exist on the market today.
With such mixed-signal test equipment, it is possible to measure quality factors originally used for analog front ends, as well as quality factors originally used for digital communication signals.
Probability density function
In probability theory, the probability density function (PDF) is a function that represents the probability that a random variable X has a value less than x. Usually, the PDF is determined based on a large number of measurements and determines the probability of all possible values of x. It is a non-negative function with unit area where a and b represent the probability intervals for X to be determined.
Complementary Cumulative Distribution Function
The complementary cumulative distribution function (CCDF) curve is closely related to the PDF because it is obtained by CCDF=1-PDF. CDF is a cumulative distribution function that can be directly obtained from PDF statistics
A CCDF curve shows the time a signal is above a certain power level. It is usually expressed in decibels above the average power.
Peak to Average Power
Ratio The peak to average power ratio (PAPR) is the ratio of the maximum peak power to the average power of a given signal, and is the most interesting measurement indicator in wireless communications. The evaluation of the impact of PAPR on the communication system is mainly obtained by analyzing the CCDF curve. We can define a specific percentage in the CCDF curve to obtain the PAPR value
Where NT is the total number of samples (time interval), which is used to determine the PAPR value.
Adjacent Channel Power Ratio The
adjacent channel power ratio (ACPR) is a measure of the amount of distortion a wireless system produces in adjacent channels relative to the main channel. It is usually defined as the ratio of the average power of the adjacent frequency channel (offset channel) to the average power of the transmit frequency channel
where F1 and F2 represent spectral intervals, S(W) is the baseband signal, and U1 and U2 are the spectral intervals of the upper adjacent channels.
As defined in the wireless standards, there are two ways to measure ACPR. One is to consider the ratio of the entire baseband signal to the entire adjacent channel. The second method (which is more widely used because it is easier to measure) is to find the ratio of the power in the entire main frequency band or in a smaller bandwidth near the carrier center frequency to the power in the adjacent channel of the same smaller bandwidth. Bit
Error Rate
The bit error ratio (BER) is the ratio of the number of erroneous bits in the received information to the total number of data bits transmitted. The BER is usually expressed as a percentage, where 0% represents no bit errors detected at the receiver.
This measurement can be performed in the digital domain by a software function implemented by the test engineer, but also requires the use of a well-known BER tester, which inputs a known data string to the transmitter and compares it with the data from the receiver output.
Error Vector Magnitude
The error vector magnitude (EVM) is a parameter used to test the accuracy of modulation and demodulation, as well as the degree of channel impairments. It can be used to quantify the performance of a digital radio transmitter or receiver. The signal transmitted by the transmitter or received by the receiver is affected by all the different imperfections in the hardware and software implementations that can cause the K modulated signal constellation points Zc(k) to deviate from their ideal positions, S(k). In daily use, EVM is a measure of how far these points deviate from their ideal positions, where, for N transmitted symbols, we can get
Test Example
To illustrate the test of an SDR receiver, we use a mixed domain measurement setup (similar to the structure shown in Figure 3), as shown in Figure 4. An arbitrary waveform generator used to simulate the transmitted digitally modulated RF signal and a receiver are simulated using the components in the block diagram.
Figure 4. The test setup of the SDR front end implemented by the instrumentation in the experiment. The device under test (DUT) is stimulated by an arbitrary waveform generator, and an oscilloscope is used to sample the analog input signal of the DUT. A logic analyzer is used to sample the digital output of the DUT. Reference signals and trigger signals are used to synchronize the input and output measurements. These devices are controlled by a computer connected using a general purpose interface bus (GPIB).
The DUT was stimulated with a single-user WiMAX signal in frequency division duplex mode with a bandwidth of 3 MHz and 64QAM (3/4) modulation.
Figure 5 shows the results measured at the output port of the SDR receiver using a logic analyzer. This figure shows the total power averaged over the stimulus band and the power in the upper adjacent channel due to nonlinear distortion. This figure shows the essence of mixed-mode testing of the SDR: the ACPR of the analog output has been reconstructed using the digital output signal and the analog input signal.
Figure 5. Measurement results at the output port of the SDR front-end under WiMAX signal stimulation. The performance of the DUT has also been evaluated using EVM at a given input power. The received digitized WiMAX signal was demodulated and error corrected based on gain and phase delay, resulting in the constellation diagram shown in Figure 6. In this particular test, the EVM achieved was approximately 5.05%.
Figure 6. Constellation diagram comparing input and output results for a WiMAX signal with 64-QAM modulation.
The characterization of SDR components is possible thanks to the use of a mixed-mode instrument that can characterize both analog and digital waveforms.
Summary and Conclusions
In this article, we have reviewed the receivers and transmitters available for SDR front-ends. We discussed the advantages and disadvantages of each. As we have seen, a well-designed architecture for a multi-band multi-mode receiver should optimally share the available hardware resources and use tunable and software programmable devices. Not every receiver architecture has this property. In this sense, in our opinion, when SDR receiver front-ends become more mature, they will be based on zero/low-IF architectures or bandpass sampling designs.
For transmitters, EER techniques and their modifications are promising options for SDR applications because their efficiency is largely independent of PAPR. Therefore, they can be easily applied to multi-standard and multi-band operations. Such SDR and CR transmitter architectures require not only efficient amplifiers, but also broadband amplifiers. The SDR world is moving from analog to digital in terms of signal transmission, so the requirements for faster RF amplifier switching speeds are becoming more pronounced and more stringent, leading to Class S transmitters in the future.
Regarding test equipment used to characterize SDR systems, we explained why mixed-domain equipment is essential for SDR characterization. We also described why some improvements are needed to develop synchronized instruments that can quickly and automatically characterize the front end and perform mismatch correction. Such equipment should ideally provide information such as EVM of different modulation types and adjacent channel power ratio of different technologies, and be able to test multi-standard and multi-band radio architectures. As SDR technology matures, we expect to see these types of instruments on the market.