Building a university test laboratory with instruments employing a software-defined-radio (SDR) architecture helps educate students and promotes research efforts.
Engineers have already played major roles in the development of wireless technologies and they will continue to do so. But engineers are made, and not born, requiring new integrative curricula at the undergraduate and graduate levels to foster the development of the next generation of wireless engineers. Part of this effort is already taking place at the University of South Florida, where a state-of-theart wireless-communications laboratory has been assembled to enable students to understand basic theory, software simulation, hardware test and modeling, system and component testing, and software and hardware interactions and co-simulations. The laboratory has been built in such a way that it can be used in support of wireless courses at all levels and can serve as a resource for research for both undergraduate and graduate students.
Figure 1 presents a generic wireless/microwave curriculum and the position of the proposed laboratory. The lab builds a bridge between the courses on the wireless system, wireless networks, wireless circuits/devices, and digital signal processing. The lab course is the outcome of some of the research activities developed at the electrical engineering department of University of South Florida (USF).1-4 A variety of test beds have been developed in recent years, such as for orthogonal-frequency-division-multiplexing (OFDM) wireless-local-area-network (WLAN) and ultrawideband (UWB) measurements, with the support of industry partners including Honeywell, Conexant, Agilent, Logus broadband solutions, Anritsu Co., and Custom Manufacturing and Engineering.
The test beds integrate vector signal generators (VSGs), vector signal analyzers (VSAs), and RF hardware with computer-aided-engineering (CAE) tools such as the Advanced Design System suite of tools from Agilent and the Matlab software from The Math- Works. The flexible test beds allow generation of a wide range of waveforms, measurement and modeling of the RF and baseband circuitry under different stimulus conditions, modeling of wireless radio-channel effects and RF impairments, and optimization of the transceiver structures and baseband algorithms.
The model for using these test beds for research is integrated into the USF educational curriculum in order to help advance students in the wireless-communications area. The model is used to expose the students to real-world wireless- communication problems and prepare them for the competitive job market. It also enables students to combine theoretical knowledge with practical, hands-on experiments.
Figure 2 shows the laboratory setup integrated with various components and instruments. The three key elements, computer software and simulations, test instruments, and hardware, are shown in different rows. The top row shows simulation capabilities, to help students gain an intuitive feeling of how theoretical knowledge is related to the real world. These capabilities allow students to model and implement real-world wireless- communication systems and help them learn how different parameters impact system performance.
The second row shows the test equipment used to connect the world of simulation with hardware. Instruments include VSGs, VSAs, and spectrum analyzers. Newer spectrum analyzers, such as the Agilent PSA Series (model E4440A) instruments that can provide in-phase (I) and quadrature (Q) signal samples by means of an integral broadband digitizer, can also be used as signal analyzers for studying the broadband modulated signals found in modern wireless-communications systems.
The VSG serves as a form of "waveform playback system." It can produce custom waveforms as well as standardsbased waveforms for communicationssystems testing. The waveforms can be created and stored within the VSG's memory or generated through software (such as Matlab) on an external computer and downloaded to the VSG's memory for signal generation. Given the VSG's capabilities of generating signals from software, it can integrate seamlessly with the modeling tools in the first row of the measurement laboratory. For example, a baseband signal can be developed using Matlab (or Agilent's Signal Studio software or ADS simulator), and then downloaded to the VSG to create the physical signal. The versatile waveform- generation instrument can also create signals with noise and other impairments to evaluate a receiver's ability to demodulate desired signals in the presence of noise, interference, and other signal impairments. Similarly, signals with fading and interference can be generated to check receiver performance under different communications channel conditions. Figure 3 shows a simplified block diagram of a VSG.
A baseband, intermediate-frequency (IF), or RF signal generated by the VSG is passed through a device under test (DUT) to study the behavior of different communications components, such as RF upconverters, filters, amplifiers, and antennas. Such test signals can also be passed through real radio channels to emulate a wireless channel. Signals can also be passed through a multipath channel emulator; the emulator provides adjustable multipath channel models.
The VSG's broad frequency range and wide modulation bandwidth allows it to cover the main frequency bands used in wireless communications and generate waveforms used for high-datarate wireless communications. Important parameters for a test-signal source include amplitude accuracy, level (amplitude) repeatability, phase noise, broadband noise, output power, and frequency accuracy.
Test signals from the VSG along with interference sources from another signal generator can be received by a DUT via antenna or through direct connection by cable. Interference models can also be generated as part of baseband signals modeled for the VSG. Received signals are passed through the receiver hardware and digitized by the VSA, which can demodulate a wide range of standard signal formats. The VSA can also capture arbitrary digital I/Q samples and process these with the software components shown in the first row of the test laboratory. Using baseband receiver algorithms, the Matlab, ADS, or other simulation software tools can process the received data. This interaction between the VSA and simulation tools provides an excellent mechanism to study and analyze present and emerging wireless systems for the purposes of research and education.
Figure 4 presents a simplified diagram of a VSA. Input signals to the instrument can be RF, IF, or baseband signals. In general, important parameters for evaluating the performance of a signal analyzer include analysis or demodulation bandwidth, which is the maximum instantaneous bandwidth the instrument can analyze; the dynamic range; the I/Q memory, which determines how many signal samples can be stored and is critical for wideband system measurements; residual error vector magnitude (EVM), and measurement speed (for increased test throughput).
Based on the model presented for the communications measurement laboratory, transceiver components can be studied in a step-by-step approach throughout a semester. The bench setup allows students to study transmitted and received signals at different levels of transceiver circuitry. Students can also change noise, interference, and other impairment sources to see their effects on the component, subsystem, and overall system performance.
Continued on page 2
In addition to these experiments, students or groups of students can be assigned projects that help them discover new ideas with the test bed. Experiments include modeling and testing power-amplifier nonlinearities; performing channelsounding measurements to understand the effects of multipath and fading; modeling I/Q impairments (such as I/Q imbalance and I/Q offset) and studying the effects of these impairments on system performance; applying various digital modulation/demodulation formats to examine performance trade-offs such as power versus spectrum efficiency; performing spectral, temporal, and spatial signal measurements to understand the multidimensional characteristics of wireless-communications signals; understanding the different types of measurements of presentation formats (such as eye and constellation diagrams, spectral masks, and ACPR measurements) used for wireless-communications testing; studying different types of filters and their effects on system-level performance; and simply understanding the basic operation of essential measurement instruments and relating these instruments with SDRs and cognitive radio concepts.
As an example of an experiment that would be performed by students in the USF wireless-communications measurement laboratory, tests were run to help students better understand nonlinearities in power amplifiers and how this nonlinear behavior can affect the performance of a wireless system. Theoretical details of power-amplifier nonlinearities abound in the literature.6-8 Specifically, this experiment was based on the use of a broadband orthogonal-frequencydivision- multiplexing (OFDM) waveform developed with Matlab software. The waveform generation code has several input variables, such as type of filter, modulation, and Fast Fourier Transform (FFT) size, which students can modify. For this experiment, the baseband software along with the VSG will be used to generate the transmitted signal that will be passed through the DUT (the power amplifier).
In the experiment, the students connect the power amplifier's output port to the VSA input by means of coaxial cable. The VSA downconverts the amplifier's output signals by means of its tunable RF front end and digitizes the signals to provide I/Q samples. The students are provided with a custom baseband digital- signal-processing (DSP) algorithm to process the I/Q signals and show results in various formats. The students can also use the Agilent VSA software to analyze the captured signal.
Students will have the opportunity to vary the transmitted waveform parameters (including the input power) and see how the amplifier behaves for different inputs. They will be able to see the timedomain signal and the corresponding power spectral density on the screen and evaluate the spectral growth of the signals etc. Also, students will be able to examine the peak-to-average-power ratio (crest factor) of the transmitted signal and some other statistics of the transmitted and received signals, such as the cumulative complementary distribution function (CCDF).
Figure 5 shows one of the plots that students are expected to generate at the end of the experiment, allowing them to observe spectral regrowth due to the power amplifier's nonlinearities. As the input power increases, the peak-clipping increases thereby causing more spectral regrowth. Figure 6 shows another plot students should generate as part of this experiment, to show the amplifier's saturation: how amplitude gain decreases as the input power increases beyond a certain level. At the system level, students can examine constellation diagrams for the effects of amplifier nonlinearities on the system's bit-error-rate (BER) performance. Figure 7 shows a constellation diagram that describes the effects of power-amplifier nonlinearities at the system level.
In summary, the wireless-communications laboratory assembled at USF provides the flexibility to meet the demands for teaching a wide range of wireless component and system engineering concepts. The laboratory complements traditional courses and serves as a resource for research.
The author would like to thank Dr. Dunleavy and Dr. Weller of the University of South Florida for their support and encouragement for the author to develop the lab course. Also, the author would like to thank to Kevin Bertlin and Pablo Estrada of Agilent Technologies (www.agilent.com) for reviewing the article.
Note: This lab course is supported in part by National Science Foundation (NSF) under CCLI program and by Agilent Technologies.
- Jiang Liu, Lawrence P. Dunleavy, and Huseyin Arslan, "Large Signal Behavioral Modeling of Nonlinear Amplifiers based on Load Pull AM-AM and AM-PM Measurements," IEEE Transactions on Microwave Theory and Techniques, Special Issue on Measurements for Large-Signal Characterization and Modeling of Nonlinear Analog Devices, Circuits, and Systems, Vol. 54, No. 8, August 2006, pp. 3191-3196.
- H. Arslan and D. Singh, "WiMAX Transceiver Testing: PART-1 - Establish Test Procedures For WiMAX Transceivers," Microwaves & RF, July 2006, pp. 63-96 (also available online http://www.mwrf.com/Articles/ArticleID/13004/13004.html).
- J. Liu, L.P. Dunleavy, and H. Arslan, "Exploration of power amplifier performance using a digital demodulation loadpull measurement procedure," in Proceedings of the 65th ARFTG Conference on Measurements for Millimeter- Wave Applications, Long Beach, CA, June 2005.
- A. Webster, J. Liu, H. Arslan, L. Dunleavy, and J. Paviol, "Measurement-based Modeling of a 5 GHz WLAN Transmitter," in Proceedings of the IEEE RAWCON, September 2004.
- H. Celebi, Y. Zhang, R. Sankar, and H. Arslan, "DSP/FPGA Laboratory for Software Defined Radio," Symposium on 21st Century Teaching Technologies and Vendor Expositions, Tampa, FL, March 2005.
- J. Li and M. Kavehrad, "OFDM-CDMA systems with nonlinear power amplifier," in Proceedings of the IEEE WCNC '99, Vol. 13, New Orleans, LA, September 1999, pp. 1167- 1171.
- A. Katz, "Linearization: Reducing distortion in power amplifiers," IEEE Microwave Magazine, Vol. 2, December 2001, pp. 37-49.
- J.S. Park, S.R. Park, H.J. Roh, and K.H. Koo, "Power amplifier back-off analysis with AM-to-PM for millimeter-wave OFDM wireless LAN," in Proceedings of the IEEE RAWCON, Waltham, MA, August 2001, pp. 189-192.