Channel impairments can be disruptive on RF-based communications devices. Impairments such as fading and multipath, which are often unpredictable, can hinder wireless communications. Because channel impairments are often unpredictable and not repeatable, many engineers face the challenge of supplying a prototype receiver with a reliable model of the real-world signal environment. Fortunately, by understanding the causes of these impairments through the use of modern measurement techniques, including the use of recorded test signals, it is possible to check a receiver's capabilities of operating effectively in the presence of channel impairments.
Free-space signal propagation can be predicted with reasonable accuracy under ideal conditions. However, factors such as buildings, geographic topography, and other obstructions can significantly impact signal strength in a real world environment. Unfortunately, large-scale channel fading often produces broadcasts that are subject to variations as a result of reflection, diffraction, and scattering.
One of the most common models for predicting signal propagation in free space is the two-ray ground reflection model (Fig. 1).1 In this model, signals traveling directly from the transmitter to the receiver follow a shorter path than those reflected from other surfaces. According to the law of reflection, the initial signal phase, Θi, is equal to the final signal phase, Θ0. Because of this law, it is possible to determine the difference in distance (for a given wavelength) between the paths of the line-of-sight (LOS) signal component (Elos) and the ground-reflected signal component (EG). With some simple math, it is possible to determine the difference in distance, d, as shown in Eq. 1.
For RF and microwave signals, even a small difference in the signal propagation path ( d,) between a LOS and a reflected signal can produce a substantial effect on the phase of the aggregate of the signal observed by the receiver. In fact, it is possible to calculate the difference in phase between the two paths with Eq. 2.
Equation 2 illustrates that especially for high-frequency signals, where the wavelength (λ) is short, even small changes in the distance between the transmitter and receiver can have substantial changes in phase of the aggregate signal. Moreover, because the LOS signal and the reflected signal are 180 deg. out of phase with one another, the two waves will periodically add or cancel one another, depending on the distance between the transmitter and receiver.
Over long distances, the antenna height is negligible, and the large-scale fading observed in these scenarios is not significantly affected by ground reflections. However, over shorter distances, ground reflections can have a substantial impact on the power observed at the receiver.
Diffraction is a phenomenon by which electromagnetic (EM) waves can propagate around buildings, mountains, and other physically large objects. Like reflection, diffraction can also produce substantial fluctuations in the signal strength observed by a wireless receiver. In addition, this effect becomes more prominent in urban environments, where a large number of buildings are positioned close together. To understand this, consider how diffraction can affect signal strength (Fig. 2). A large object may prevent LOS signal propagation, but EM waves will actually wrap around the object, producing an attenuated but usable signal. Some theoretical models may help to understand just how much signal attenuation can occur as a result of diffraction.
One of the more popular models for analyzing the effects of diffraction on EM waves is the Knife-edge diffraction model.1 It provides a mechanism for estimating signal strength as a function of wavelength, object height, and distance between the transmitter and the receiver. Based on a geometric model of the physical environment, it is possible to calculate a Fresnel-Kirchoff diffraction parameter using Eq. 3.
As Eq. 3 shows, the diffraction parameter (and hence loss) increases with the height of the obstructing object. In addition, since wavelength, λ is in the denominator, higher-frequency microwave signals are much more susceptible to attenuation from physical objects. Based on the calculated Fresnel-Kirchoff diffraction parameter, it is possible to calculate the expected loss in signal strength as a result of diffraction. William C. Y. Lee, a pioneer in wireless communications technology, actually approximated the loss/gain due to diffraction, Gd (dB), as a function of the Fresnel- Kirchoff diffraction parameter, v. His derivation follows in Eq. 4.
Figure 3 graphically depicts the diffraction gain/loss, Gd (dB), as a function of the Fresnel-Kirchoff diffraction parameter. Based on the equations above, consider a scenario where a transmitter and receiver are 1 km away (d1 = d2 = 500 m) and a 100-m object (such as an urban office building) is equidistant between them. Based on these distances, a 1 GHz signal (λ = 0.3) will have a Fresnel- Kirchhoff diffraction parameter with a value of 11.457. Thus, the receiver will observe a signal strength loss of approximately 34.2 dB. As this exercise illustrates, even a single object (if tall enough) can result in significant loss of signal strength. Thus, compensation for fluctuations in signal strength is an important requirement of wireless receiver design.
While large-scale channel fading often causes rapid changes in signal strength, small-scale channel fading produces distortion in either the phase or amplitude of an EM wave. In a typical environment, a receiver will pull in a signal comprised of signal components from many different signal paths. In this case, an EM wave from each signal path will arrive at the receiver at a different time. Unfortunately, such multipath propagation causes intersymbol interference (ISI) in which symbol n - 1 will distort the phase and amplitude of symbol n. While ISI does not significantly affect the overall power level of a received signal, it does affect the modulation quality.
Two models commonly used to emulate multipath fading are the Rayleigh and Rician fading models. The Rician model includes both LOS propagation between the transmitter and the receiver and non-LOS signal propagation. In this model, a K parameter represents the power of the LOS transmission relative to the aggregate sum of the multipath signal products. When the K parameter is large, multipath products are minimal and the ISI is reduced. When K is small, ISI becomes more significant. Figure 4 shows this effect, comparing the K parameter for a 16 QAM signal with a symbol rate of 3.84 MSymbols/s and the influence of a Doppler frequency shift of 5 kHz.
As Fig. 4 shows, higher ISI significantly reduces the modulation quality, as measured in terms of error vector magnitude (EVM), in situations where LOS signal propagation is not possible. As a result, receiver validation often requires multipath channel emulation to ensure that the receiver will behave as expected in its actual operating environment.
Both large-scale and small-scale fading pose design challenges. To maximum a receiver's dynamic range, most receivers implement an automatic- gain-control (AGC) circuit to compensate for rapid changes in received signal strength. The ADC is typically placed immediately following a preselector filter in a wireless receiver (Fig. 5).4 The AGC ensures that the mixer and intermediate-frequency (IF) levels remain relatively constant, allowing the receiver to achieve adequate dynamic range even with rapid fluctuations in signal strength.
Continue to page 2
In addition to the design challenges created by largescale and small-scale fading, they also make it difficult to test a wireless receiver for such conditions. For realistic evaluation, engineers must provide a receiver under test with a stimulus that most accurately represents the deployment environment. Traditionally, receiver design validation and verification has been a difficult, expensive, and often time-consuming process. Today's test engineers generally use one of three approaches for receiver validation: channel emulation, drive testing, and signal record and playback.
Channel emulation can be performed with a channel simulator designed to reproduce various types of fading. Typically, these instruments use statistical models for various frequencies and types of environment (urban, rural, mountains). With a channel simulator, as many as 6 to 12 different signal paths can be used to recreate large-scale or small-scale multipath fading.
However, channel simulators have their drawbacks. Because receivers often experience as many as 40 or more different signal paths in their deployment environment, even the most sophisticated simulators cannot recreate the full complexity of a realworld signal. In addition, simulators usually only have limited options for simulating interference. While some simulators provide a second interference channel, they cannot typically provide multiple sources of interference at varying levels of power intensity, as commonly found in a deployment environment. A channel simulator provides the comfort of performing measurements in a lab, but may lack the capability to simulate the actual operating environment.
For this reason, drive testing is another method used for evaluating a receiver's performance under real-world conditions. Drive testing ensures that a receiver can adapt to both largescale and small-scale channel fading. By driving through a typical deployment environment, engineers can determine if their receiver can adapt to even the most drastic power changes and multipath fading introduced in the channel. In some cases, engineers will choose a particularly difficult environmental scenario in which many receivers are known to fail or give poor performance.
But even drive testing has drawbacks. Not only are drive tests expensive and time consuming, they often are not repeatable. Because environmental factors such as weather and humidity can substantially impact signal propagation during a particular test, it is often difficult to recreate those same conditions when needed. For this reason, substantial interest is growing in technology that enables engineers to record an RF signal of interest and play it back in a laboratory environment.
By recording actual signals, it is possible to capture a wide range of channel impairments such as large-scale channel fading, multipath propagation, and interference. The process of capturing actual RF communications signals involves literally taking an RF recording system on a drive. It is driven through an environment known to be difficult for receivers. Then, once a large amount of data is captured (hours or more of signal data), the recorded signal can be regenerated in a controlled laboratory environment and directly connected to the receiver for evaluation.
The use of recorded waveforms for receiver testing has many advantages over traditional test methods. Compared to channel simulators, this method enables a receiver to be evaluated with more natural impairments as opposed to simulations. This approach also produces test results that are perfectly repeatable. Because a recorded waveform will not change from one test to the next, the receiver can be characterized throughout multiple stages of the design cycle using the same stimuli data set. Finally, RF record and playback solutions are often more cost effective than the alternative methods. As an example, consider testing a receiver in different worldwide locations. While drive testing requires transporting a receiver to each location for evaluation, recorded signals from multiple locations can be stored on file and generated with the same system.
Actual implementations of RF record and playback systems are not without significant design challenges. In fact, two key innovations have enabled these systems to effectively address the mainstream of receiver applications: improvements in digital bus technology and AGCs with gain reporting.
Traditional RF instruments use embedded random access memory (RAM) as a mechanism for waveform storage, limiting the maximum waveform size to several hundred megabytes. However, the evolution of faster bus speeds enables instruments based on the PCI Extensions for Instrumentation (PXI) bus format to use high-speed external redundan-array-of-inexpensive-disks (RAID) memory to store and replay much larger waveforms. Using external RAID memory, PXI instruments can generate or acquire waveforms as much as several terabytes in length at the full data rate of the instrument.
Using LabVIEW signal-processing software from National Instruments, RF stream-to-disk applications achieve best results with a concurrent loop structure called producer-consumer loop. In this case, the producer loop acquires baseband data from a vector signal analyzer (VSA) and passes it to a queue structure. The queue structure passes baseband data to a second loop (the consumer), which writes it to disk. Figure 6 shows how this system can be configured in software. The approach can be used at the full bandwidth of a PXI instrument. As an example, the NI PXIe-5672 2.7-GHz vector signal generator (VSG) from National Instruments can support continuous generation of bandwidths as wide as 20 MHz for five hours or more.
A second design challenge of RF record and playback designs is the difficulty in maintaining the full dynamic range of the acquisition instrument. While the dynamic range of one recording instrument, the PXI-5661, operates with as much as 80 dB of dynamic range, it is difficult to maintain that dynamic range through an entire drive test without an AGC. For example, real-life FM signals range from less than 0 dB V to +110 dB V. Moreover, even at 0 dB V, the noise floor is around 13 dBV. Thus, a VSG would need more than 120 dB of dynamic range without external signal conditioning to record large changes in signal strength.
A possible solution is to add a simple AGC at the front end of the recording instrument, such as an AGC available from Averna Technologies and shown in block diagram form in Fig. 7. The AGC ensures that the recording instrument observes a relatively constant power level. By maintaining constant power at the VSA, it is possible to use the full dynamic range of the instrument even as the power of the recorded signal is constantly and often significantly changing.
This approach provides the capability to store the gain data supplied by the AGC. By storing this information, the reverse gain (attenuation) can be applied to the signal upon playback with a vector signal generator. As a result, a record and playback system with an AGC can be used to accurately capture and reproduce RF signals even when rapid changes in signal strength occur. To illustrate the effectiveness of an AGC when used with a recording instrument, Fig. 8 compares the power of a record-and-playback signal with and without the AGC on the front end. As Fig. 8 illustrates, an AGC can be used to recreate the large-scale fading of the recorded RF signal with less than 2-3 dB of variation from the actual signal strength. In addition, this can be done without sacrificing the dynamic range of the recording instrument.
Although EM waves for communications can undergo a great number of impairments in the real world, modern RF record and playback systems offer a practical solution to these changing signal conditions. With AGC-enabled instrumentation, test engineers can now accurately recreate challenging physical channels in a laboratory environment. Even waveforms with complex modulation formats, including advanced digital modulation formats, can be stored in PXI instrument memory and manipulated in terms of fading and other channel impairments in order to precisely emulate the signal environment that a communications receiver is likely to face under actual working conditions. Thus, receivers can now be tested with greater repeatability and at lower cost than ever before.
1. Theodore S. Rappaport, Wireless Communications: Principles and Practice, Prentice-Hall PTR, Saddle River, NJ, 2001.
2. W. C. Y. Lee, Mobile Communications Engineering, McGraw-Hill, New York, 1982.
3. Jeffery H. Reed, Software Radio: A Modern Approach to Radio Engineering, Prentice-Hall, Saddle River, NJ, 2002.
4. Quzheng Gu, RF System Design of Transceivers for Wireless Communications, Springer, New York, 2006.