Printer Friendly

How to evaluate a multipath fading emulator.

Multipath fading emulators have become indispensable for testing wireless communications equipment. However, these devices generally are not well understood and data sheets typically do not reveal the information needed to compare instruments or determine the suitability of a unit for a particular application. This article provides a systematic approach for evaluating multipath fading emulators against the criteria that are important in a typical test environment.

The performance of a wireless communications network depends not only on the types of mobiles and base stations in use, but also on the deployment of the base stations and any factors that affect the local RF propagation conditions. While field testing can provide a rough approximation of how well a given mobile or base station will operate under various conditions, its utility is mostly limited to acceptance testing performed by service providers on their own networks. Eliminating interdependencies between the network infrastructure, subscriber units and RF environment requires that comprehensive testing be performed on each individual element.

Every communications standard identifies performance requirements that must be met for a product to be considered compliant, and compliance with the standard is supposed to guarantee interoperability of the various network elements. The majority of testing is performed using a coaxial connection directly from the radio test set to the device under test (DUT). However, rigorous testing of the receiver section of the DUT must be performed in a way that emulates real-world conditions since consumers rarely will be linked to the base station via a perfect hardwired connection.

The actual communications channel between a mobile and a base station is subject to various imperfections for which the respective receivers must compensate. Most digital mobile radio standards include requirements that both base stations and subscriber units be tested under various interference and multipath fading scenarios, which have been selected to model the RF environments likely to be encountered by users.

An ideal RF communications channel consists of a single line-of-sight path of constant distance through a uniform propagation medium with little attenuation. In a wireless network, a direct path may exist between a transmitter and receiver, however, the signal also may arrive via multiple indirect modes of propagation such as reflection, diffraction and scattering. The receiver detects the direct signal (if present) followed, in turn, by each indirect signal. The time delay for each indirect signal may be independent and depends on the distance traveled and the characteristics of the transmission medium encountered. Large fluctuations in the amplitude and phase of the desired signal may occur when the direct and delayed signals are combined in the receiver. This phenomenon is referred to as multipath fading.

Modeling the effects of multipath fading requires real-time processing and complex distortion of a live, unknown signal. A multipath fading emulator is a specialized piece of test equipment that allows a user to simulate a comprehensive variety of propagation effects while maintaining a tightly controlled test environment. Each channel of a multipath fading emulator represents the overall RF link between a transmitter and receiver and can consist of one or more paths. (Each path in a channel simulates a unique propagation mechanism for the input signal.) The functional block diagram for one channel of a typical multipath fading emulator is shown in Figure 1.

BANDWIDTH

The bandwidth of the multipath fading emulator must be sufficient to accommodate all signals that need to be processed concurrently, including the primary test signal as well as any desired pilot, control and interfering signals. The bandwidth must also be wide enough to accommodate any frequency hopping requirements. The likelihood that a spurious signal or noise originating inside the instrument will be passed along to the output and detected by the DUT increases as the bandwidth of the multipath fading emulator increases.

It is desirable that the multipath fading emulator not introduce amplitude and phase distortion to the input signals. This performance restriction requires the gain and group delay to remain constant across the needed bandwidth. The bandwidth of the multipath fading emulator can be evaluated using a vector network analyzer. The multipath fading emulator should be set to the desired center frequency and the automatic gain control should be disabled. In addition, the network analyzer should be set to measure magnitude and group delay through the multipath fading emulator.

DYNAMIC RANGE

The dynamic range of a multipath fading emulator generally is defined as the ratio between the largest and smallest signals that can be processed and reproduced simultaneously. In practical terms, this ratio usually equates to the difference between the amplitude of the desired signal and that of any spurious or noise signals produced by the multipath fading emulator and detected by the DUT. It is important to note that the available dynamic range of a given multipath fading emulator may vary depending upon the application.

Dynamic range is one of the most important factors to consider when evaluating the suitability of a multipath fading emulator for use in a given application. Generally, receiver testing is performed by measuring the bit error rate (BER) at a variety of carrier-to-noise (C/N) or carrier-to-interference (C/I) ratios. The BER specification may be stated in terms of frame error rate or frame erasure rate, and the C/N may be expressed in alternate terms such as energy per bit to noise density ([E.sub.b]/[N.sub.o]), but the principle remains the same. The multipath fading emulator must allow enough dynamic range after fading to test the BER at all desired C/N and C/I ratios. The dynamic range requirements for a multipath fading emulator can be derived from the upper limits of C/N and C/I at which system performance must be verified. A minimum dynamic range of 30 dB is required to reproduce the nulls that can occur during Rayleigh fading.

INTERFERENCE

A receiver cannot differentiate between desired and undesired signals or even between discrete signals and noise. Thus, all undesired signals may be classified generically as interference. Measurement and calculation methods differ according to the signal type, but the effects on the DUT are indistinguishable. The dynamic range of a multipath fading emulator is determined primarily by the amount of interference it generates. A multipath fading emulator produces three general types of interference signals: in-band spurious, out-of-band spurious and noise.

In-band Spurious

In-band spurious signals are undesired signals produced by the multipath fading emulator within the bandwidth of the DUT. These signals are detected by the receiver as distortion to the desired signal and cannot be removed or attenuated from the desired signal without causing further signal degradation.

Motion of a transmitter, receiver or reflector can cause a Doppler frequency shift in an RF carrier. Two spurious signals result from the vector modulation used to implement Doppler shifts in a multipath fading emulator: signal feedthrough and Doppler image. The signal feedthrough and Doppler image often will be the largest spurious signals generated by a multipath fading emulator and directly impact the available dynamic range directly because they always fall within the bandwidth of the DUT. The Doppler offset frequency will be

[Delta][f.sub.d] = v/c x [f.sub.c]

= v/[Lambda]

where

v = relative velocity of transmitter with respect to receiver

c = speed of light ([approximately equal to]3 x [10.sup.8] meters/second)

[f.sub.c] = carrier (and signal feedthrough) frequency

[Lambda] = carrier wavelength

The Doppler shifted carrier frequency will be

[f.sub.Doppler] = [f.sub.c] + [Delta][f.sub.d]

The Doppler image carrier frequency will be

[f.sub.Doppler image] = [f.sub.c] - [Delta][f.sub.d]

Intermodulation products are generated when two or more signals are passed through a multipath fading emulator. For any pair of signals, the third-order intermodulation products generally are the most problematic and occur at

[f.sub.IM1] = 2[f.sub.1] - [f.sub.2]

[f.sub.IM2] = 2[f.sub.2] - [f.sub.1]

where

[f.sub.1],[f.sub.2] = input signal carrier frequencies

[f.sub.IM1],[f.sub.IM2] = third-order intermodulation product frequencies

Intermodulation distortion also appears as spectral regrowth when a spread spectrum signal (such as a code-division multiple access (CDMA) signal) is passed through the multipath fading emulator. Spectral regrowth appears as a noise pedestal around the spread spectrum signal. The increased noise floor caused by the spectral regrowth decreases the dynamic range of the multipath fading emulator and causes uncertainty in the C/N (or [E.sub.b]/[N.sub.o]) detected by the receiver.

Digital signal processing phenomena such as undersampling, aliasing and quantization error contribute to distortion of the baseband signal that is transferred to the RF output. The analog-to-digital (A/D) and digital-to-analog (D/A) stages of the multipath fading emulator must offer an optimum combination of sampling rate, vertical resolution and filtering to faithfully reproduce the desired signal. The processing stages also must minimize abrupt signal transitions, which can cause impulse noise and artifacts.

Nonlinear devices can introduce harmonic distortion to a signal. In a multipath fading emulator, harmonics are of particular concern if they are generated at a baseband and within the processing bandwidth.

Out-of-band Spurious

Spurious signals generated by a multipath fading emulator outside the bandwidth of the DUT can be categorized as out-of-band spurious. These signals generally will not be detected directly by the receiver unless the DUT provides inadequate filtering. However, the presence of large, unexpected signals can cause gain compression, overload or intermodulation problems in the DUT or other devices in the test system. Out-of-band spurious can also cause uncertainty in power measurements since most power meters are inherently broadband devices. The effects of any known out-of-band spurious signals generated by a multipath fading emulator must be considered carefully.

In the output stages of the multipath fading emulator, the RF output signal is produced by mixing the processed baseband signal with a series of LOs. Each upconversion stage generates two significant spurious products that must be considered: the LO feedthrough and image. If lower sideband conversion is used in the mixing process, the spurious products occur at

[f.sub.LO] = [f.sub.c] + [f.sub.IF]

[f.sub.image] = [f.sub.c] + 2 x [f.sub.IF]

where

[f.sub.LO] = LO frequency

[f.sub.image] = image frequency

[f.sub.c] = multipath fading emulator center frequency

[f.sub.IF] = intermediate frequency (IF)

The amplitudes of these mixing products can exceed the amplitude of the desired signal so filtering generally is used after each frequency conversion. If the mixing products are not filtered sufficiently, they will be impossible to remove in subsequent conversion stages and will appear as spurious signals at the output of the multipath fading emulator. The spurious products that may be present at the output of a dual-conversion multipath fading emulator due to LO feedthrough and image signals are shown in Figure 2 where the carrier frequency is 860 MHz and the first and second IFs are 140 and 15 MHz, respectively.

Noise

Since a receiver will integrate all signals within its bandwidth, the effect of any noise contribution from the multipath fading emulator increases in direct proportion to the DUT bandwidth. Excess noise generated by the multipath fading emulator reduces the available measurement dynamic range and may cause distortion on any signals being processed.

In the context of receiver performance testing, interference signal levels rarely are expressed individually - the [TABULAR DATA FOR TABLE I OMITTED] important parameter is the amplitude of the interference in relation to that of the carrier. Although the C/I ratio generally is used when discussing receiver performance, it is more convenient to refer to the interference-to-carrier (I/C) ratio when evaluating distortion of a signal due to interference because expressing interference in terms of I/C allows the interference components to be summed directly.

Specifications for spurious products generally are measured and expressed in decibels below the carrier (dBc), which represents a ratio between the power of the interference and that of the carrier. However, a receiver does not detect power; it detects the voltage induced across its input impedance. The formula used to convert from units of dBc to the equivalent I/C voltage ratio that will be detected by the receiver is

I/C = 10 dBc/20

= 10 [P.sub.s] - [P.sub.0]/20

where

dBc = IC (dB)

[P.sub.s] = spurious signal power (dBm)

[P.sub.0] = carrier power (dBm)

Noise specifications and measurements generally are expressed as power spectral densities in units of dBm/Hz or dBc/Hz. Assuming the noise bandwidth is greater than the receiver bandwidth and the noise is evenly distributed across the receiver bandwidth, I/C can be calculated as

I/C = 10 dBc/Hz + 10log (B[W.sub.r])/20

= 10 dBc/Hz - [P.sub.0] + 10log(B[W.sub.r])/20

where

dBc/Hz = multipath fading emulator noise-floor-to-carrier ratio (dBc/Hz)

B[W.sub.r] = receiver bandwidth (Hz)

dBm/Hz = multipath fading emulator noise floor (dBm/Hz)

[P.sub.0] = carrier power (dBm)

The net I/C ratio simply will be the sum of contributions from all interference components. Several values can be derived from the net I/C ratio:

dynamic range = -20log (I/C)

error vector magnitude = I/C

magnitude [error.sub.s] = I/C/[-square root of 2]

phase [error.sub.s] = [tan.sup.-1](I/C/[-square root of 2])

amplitude [ripple.sub.k-t-k] = 20log[1 + [(I/C).sub.k]/1 - [(I/C).sub.k]]

Table 1 lists the modulation errors and amplitude ripple that a receiver detects in the presence of various levels of interference. Even relatively low levels of interference can cause uncertainties large enough to detract from the performance margin that has been designed into the DUT. If the design margins in the DUT are not sufficient to allow for the measurement uncertainties introduced by the test system, erroneous and unrepeatable test results will be recorded. On the other hand, it is difficult to justify expending significant development resources to build additional margin into a design solely to compensate for excessive test system uncertainties.

Evaluation of the dynamic range of a multipath fading emulator can be performed using a signal generator and spectrum analyzer. It is easiest to perform the testing using a CW input signal because the measurements can be made directly using marker functions on the spectrum analyzer and there is no modulation to mask in-band spurious signals. It is important to keep in mind that all spurious signals found using a CW input signal are present regardless of whether they are masked on a spectrum analyzer display by fading effects or by modulation of the carrier signal. The receiver detects the hidden spurious signals and its performance is affected accordingly.

The multipath fading emulator should be set to produce a Doppler shift of the test signal. The span of the spectrum analyzer should be set to at least three-times the selected Doppler offset and the resolution bandwidth should be set to no more than one-fifth the Doppler offset. The level of the desired signal should be observed as the Doppler offset is varied throughout the range of interest. The desired signal level should remain constant and be used as the carrier power for all subsequent measurements. The maximum levels of signal feedthrough and Doppler image should be recorded.

In addition, the span and resolution bandwidth of the spectrum analyzer should be increased progressively so that all undesired signals can be measured and recorded. Finally, the noise floor within the bandwidth of the multipath fading emulator should be measured and recorded. The maximum dynamic range of the multipath fading emulator can be calculated by summing the I/C ratio of each in-band spurious and noise signal.

AN EXAMPLE

A model MP2700 multipath fading emulator was tested to allow estimation of the available dynamic range for various applications. The amplitudes of the signal feedthrough and Doppler image were determined to be 53 and 62.7 dBc, respectively. No other significant in-band spurious signals were found. The in-band spurious products are shown in Figure 3.

The noise floor was measured at -146.2 dBm/Hz using the marker function on the spectrum analyzer. However, further analysis showed that this reading was only 1.75 dB above the spectrum analyzer noise floor. Since the measured noise floor is actually the sum of the noise floors of the model MP2700 multipath fading emulator and spectrum analyzer, the instrument's noise floor was actually only -151 dBm/Hz. The emulator's noise floor is shown in Figure 4.

Since no other significant signals were found, the dynamic range of the model MP2700 emulator can be estimated by summing the interference caused by the signal feedthrough, Doppler image and noise floor. Table 2 lists the calculated dynamic range for several wireless communications standards including wideband CDMA (W-CDMA) and future W-CDMA.

TEST SYSTEM COMPATIBILITY

While performance issues largely determine whether a given multipath fading emulator can be used for a particular application, several additional factors determine how easily it will adapt to different test environments
TABLE II

THE MP2700 MULTIPATH FADING EMULATOR'S ESTIMATED DYNAMIC RANGE

                   System Bandwidth      Dynamic Range
Standard                (MHz)                (dB)

TDMA                    0.0250                50
GSM                     0.2000                48
CDMA                    1.2288                46
W-CDMA                  4.0960                43
Future W-CDMA          20.0000                38


Frequency Range

All multipath fading emulators operate throughout a limited frequency range. It is fairly simple to use an additional stage of frequency conversion to cover frequencies above the specified range if necessary. A more difficult situation occurs when it is necessary to work at frequencies not covered by the final output filter. Typically, the tunable output filter of a multipath fading emulator cannot be used at frequencies below approximately 800 MHz, and this limitation can cause problems in paging or digital television applications. If no filtering is performed on the signal, the LO feedthrough and image signals will be amplified by the final gain stage of the multipath fading emulator and pass directly to the output. Even worse, the presence of the spurious signals may force the output amplifier into compression and severely distort the signal. Even if filtering is performed at the output of the multipath fading emulator, distortion to the signal cannot be reversed. It is important that the multipath fading emulator offer a solution to this problem.

Input Signal Level Range

It is desirable that the multipath fading emulator be able to accommodate a wide range of input signals without the need for external amplification or attenuation. Automatic gain control (AGC) circuitry can condition the input signal so that the full dynamic range of the A/D converter is used, regardless of the signal type. AGC allows constant output power to be delivered from the multipath fading emulator even if the input signal level changes. It is important that the multipath fading emulator offer the ability to disable the AGC for power control or other tests that require the output power to track the input power.

Output Signal Level Range

It is desirable that the output signal level from the multipath fading emulator be compatible with the input level required by the DUT or the next downstream piece of test equipment. The most important consideration is the maximum output power available from the multipath fading emulator. It is fairly simple and inexpensive to decrease the output level from the multipath fading emulator by inserting external attenuation, but it is much more complicated and expensive to increase the output signal level.

Output Signal Stability

It is desirable for the output power of the multipath fading emulator to remain stable and accurate regardless of the fading characteristics selected. Otherwise, the necessary calibration steps can be complicated and time-consuming.

Baseband (I/Q) Inputs and Outputs

It is often necessary to isolate the performance characteristics of the baseband and RF subassemblies in a receiver. This capability can be useful when debugging a design or determining system margins and becomes indispensable when the RF section of the design is either unavailable or incomplete. The performance of the incident and quadrature (I/Q) interface should be at least as good as that of the RF section to be used in the completed design.

Flexibility in Configuration

All multipath fading emulators provide for simple single- and multichannel operation. However, special channel and path configurations are often required. It is desirable for the multipath fading emulator to be able to address interference (two input signals/one output) or diversity (one input signal/two outputs) applications - or even a combination of both - with no external hardware.

CONCLUSION

Determining the best multipath fading emulator for use in a given application requires a thorough understanding of the application as well as careful consideration of the remainder of the test system. While the data sheets make most of them appear to be similar in performance and functionality, multipath fading emulators are complex instruments and significant differences exist. Making a well-informed decision requires the user to look beyond the data sheet when selecting a multipath fading emulator; the criteria discussed should provide a good starting point for a meaningful evaluation.

Patrick Weisgarber received his BSEE from Rensselaer Polytechnic Institute. Currently, he is a senior account manager for Noise Com and can be reached at (201) 760-1000.
COPYRIGHT 1998 Horizon House Publications, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Wireless Report
Author:Weisgarber, Patrick
Publication:Microwave Journal
Date:Oct 1, 1998
Words:3522
Previous Article:ASTOR contractors reveal radar sensor details.
Next Article:Third-generation wireless test equipment requirements.
Topics:

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters