System Level Simulation Helps Validate Wireless
By Joel Kirshman, AWR
The safest, most cost-effective route to the design of a wireless transceiver is to consider all variables that will affect its performance as early as possible in the design process. This includes exposing it to “system level” measurements and evaluating the effect on communications link performance of impairments arising in both the baseband and RF sections of the design. Once this has been done, the design can be optimized, dramatically reducing the possibility for later rework. AWR’s Visual System Simulator™ (VSS) software, an integral part of AWR’s design environment, provides a way to conduct this analysis, as examples of a WiMAX transceiver will clearly show.
From Spreadsheet to VSS
The high-frequency EDA industry, which consistently enhances the capabilities of its products year after year, has lagged in the area of system-level measurements. Tools to evaluate the performance of a design using “real-world” metrics have either been expensive, poorly-integrated with the “core” design tool, or simply absent for much of the history of the wireless industry. For the lack of an alternative, designers have until recently been left to fend for themselves, developing spreadsheets chock full of macros to evaluate the effect of these parameters on their design.
This approach has obvious shortcomings, the most glaring of which are a spreadsheet’s inability to correctly calculate the true effect of noise on link quality, its empirical nature (since each one reflects the vagaries of the person who created it), and the enormous amount of time spreadsheet development requires that could be better spent doing actual design work. From a more global perspective, the traditional microwave design methodology does not include the baseband section, so standards-based modulated waveforms cannot be employed as the stimulus for the transmission chain.
It also effectively eliminates the ability to make measurements such as bit error rate (BER) that require a modulated signal source. So, rather than exposing the design to conditions it will see in actual service and optimizing its performance based on “realistic” measurements, the designer adds margin for safety purposes. This often results in over-designed (i.e., expensive) products and significant design rework as the ramifications of system-level requirements rear their heads late in the development cycle. This scenario has become all but intolerable in today’s competitive environment.
The Unified Approach
A unified approach in which baseband and RF portions of the design are evaluated together is rapidly becoming the only acceptable choice, and it is the basis of VSS and its integration within Microwave Office® design environment. VSS can model and simulate very complex systems, such as transmitters and receivers, for nearly all types of wireless systems. It can be extremely helpful when creating signal processing algorithms as well as in verifying proof-of-concept designs (including baseband and RF) through simulation. This unified approach allows co-simulation to be performed with AWR’s harmonic balance, linear, and electromagnetic simulators. A harmonic balance or linear simulation, for example, will automatically be initiated when VSS is working with a circuit designed in Microwave Office design environment.
Measurements such as Error Vector Modulation (EVM), receiver sensitivity, BER, Adjacent Channel Power Ratio (ACPR), I/Q constellations, eye diagrams, and Complex Cumulative Distribution Function (CCDF) can identify hardware-based impairments, and designs can be optimized by varying circuit and component values to achieve the best results. VSS can also be used to perform cascaded noise budget analysis and to develop specifications for filters and other RF components. Detailed frequency planning and analysis can be performed as well.
VSS allows a device under test (DUT) to be characterized using several sources of circuit information (Figure 1). For example, a DUT that contains a low-noise amplifier can be characterized using behavioral models into which characteristics of the device can be inserted (1-dB compression point, second- or third-order intercept point, noise figure, saturation, etc.) These parameters are entered through variables in the model. Another way to characterize an RF device is by using an existing circuit created with a design tool such as Microwave Office. A laboratory measurement or information from the manufacturer’s data sheet can be used as well.
A real-time simulation of an IEEE 802.11 a/g transmitter helps show the utility of VSS (Figure 2). In this case, a power amplifier was imported from a circuit designed with Microwave Office software. The simulation can investigate the amplifier’s behavior, show compliance with the spectral requirements of the WiFi standard via a spectrum mask, display the response of a VCO versus time, measure phase noise at various points in the system, or display the amplifier’s RF output power as a function of input level.
VSS also provides both fixed and mobile WiMAX design libraries that are accessible from the element tree, and the WiMAX transmit and receive modules are accessible from the top level of the element library. The library contains models utilized to build the WiMAX transmitter and receiver, such as randomizer and derandomizer, encoder and decoder, interleaver and deinterleaver, mapper and detector, and frame assembler/disassembler.
The transmit and receive blocks are built from subcircuits that are system diagrams with input and output ports that can be dragged and dropped into other system diagrams. This allows visibility into internal implementation details of these blocks, which has proven very helpful to many users. When components of the WiMAX library are included into a system diagram, all dependent subcircuits are automatically also added so that WiMAX sources or receivers can be incorporated in an existing project with minimal effort.
The VSS WiMAX transmitter and receiver blocks are programmed with standard settings defined in the specification documents and can be changed with an interface that contains drop-down menus with standard options for various parameters. For example, the Rate_ID parameter may be set using the standard options such as QPSK with rate 1/2 coding and QPSK with rate 3/4 coding. The WiMAX transmitter contains several output ports and can perform randomization, FEC encoding, interleaving, repetition encoding, mapping, frame assembling, OFDM modulation, and pilot tone generation. These components are in turn implemented as subcircuits. The WiMAX transmitter generates the modulated I/Q signal and provides auxiliary outputs such as the randomization sequence that can be used at the receiver, the transmitted data that can be used for BER calculations, and the pilot tone data that can be used at the receiver for synchronization or channel estimation.
The WiMAX receiver in the VSS library performs OFDM demodulation, frame disassembly, soft decision demodulation, deinterleaving, soft decision Viterbi decoding, and derandomization. Its output is the decoded data and the decoded pilots, the latter for use in performing channel equalization.
View from the “Test Bench”
For purposes of this discussion, the WiMAX transmitter will drive a DUT and the WiMAX receiver will be used to generate the decoded data needed for BER calculations. Using the VSS mobile WiMAX test bench, the simulation can be performed in real time. The test bench (Figure 3) includes a transmitter that outputs a 64-QAM rate-3/4 WiMAX signal with a 10 MHz channel bandwidth. The output is passed into a block that introduces amplitude and phase distortion and adds a DC offset. The next component in the system diagram adds noise defined by a phase noise mask from a data file. In this case, the DUT is a power amplifier defined using measured AM-to-AM data whose output and the signal from the WiMAX transmitter are demodulated. Their outputs are used for performing EVM measurements.
The signal level at the output of the WiMAX transmitter is swept from 0 to 10 dBm in 1 dB increments. The figure shows CCDF measurements performed using the clean, undistorted signal at the WiMAX transmitter’s output and the output of the power amplifier. As the signal level increases, the CCDF measurement at the output of the PA shows more and more saturation effects. The figure also shows EVM measurements performed for each step of transmit power plotted versus the output power of the power amplifier, which illustrates the saturation effect of the amplifier and how it causes higher EVM (greater distortion) as the input signal pushes the amplifier into saturation.
The “AM/AM and Operating Point” plot of Figure 3 shows the AM-to-AM characteristic of the power amplifier (red curve), its current operating point (the black circles), and instantaneous output power as a function of instantaneous input power of the amplifier (the blue curve). The curve slowly moves more and more into the amplifier’s saturation zone as input power increases. The I/Q constellation plots show the clean, undistorted signal and that of the signal at the output of the amplifier. The constellation becomes more and more distorted as the amplifier reaches saturation.
The spectrum mask plot shows the spectrum of the WiMAX signal and the amplifier’s output signal as the amplifier is driven into saturation. The spectral regrowth is very visible as it begins to violate the spectral mask. The ACPR measurements are shown at three frequency offsets: the left edge, center, and right edge of the adjacent channel. As expected, they reflect the spectral regrowth that results when the amplifier is driven into saturation.
Moving to the receiver “test bench” (Figure 4), the WiMAX transmitter is used to generate a standard signal that is passed through an Additive White Gaussian Noise (AWGN) propagation channel. The DUT is a typical direct conversion receiver front end that includes an RF filter, low-noise amplifier, local oscillator, mixer, and baseband filter. Its output is fed into a WiMAX receiver and the decoded data is used to evaluate receiver sensitivity. In this case, the WiMAX signal level at the input of the receiver front end is swept from -105 dBm to -98 dBm and Additive White Gaussian Noise (AWGN) of -174 dBm/Hz is added. The simulation shows the spectrum at the output of the WiMAX transmitter, the signal plus noise, and the output of the receiver front end. BER is calculated and displayed graphically.
VSS is a distinct improvement over both spreadsheet-based analysis as well as other tools that attempt to simulate and analyze a design at the system level. The software gives designers the ability to fully understand the profound effects that can be caused by impairments to RF links in systems employing complex modulated schemes. By performing the evaluation using metrics by which the completed system will be judged, it can dramatically reduce design time, rework, and cost, and ensure that the design performs well in service.
this article to a friend!