Testing real-world digital mobile phone systems can become very complex very quickly. Both frequency and time-domain signal aspects must be considered when attempting to simulate the conditions of the operating environment. And the situation is further complicated by the existence of many different standards.
In fact, there are so many different standards that few mobile phones can truly be used throughout the world. One exception is the Iridium satellite system, although currently it is expensive.
Historically, both North America and Europe developed analog wireless communications before they moved on to more bandwidth-efficient digital systems. But when GSM was specified and developed in Europe, no attempt was made to maintain compatibility with existing analog systems.
This was not the case with North American-developed TDMA and CDMA phones. The resulting incompatibility of what are effectively geography-specific standards is being attacked on a number of fronts.
A comprehensive approach is being taken by the ITU, which aims to develop a worldwide wireless standard called International Mobile Telecommunications-2000. The 3G wireless systems that ultimately emerge may use wideband packet CDMA technology.
This terminology has resulted from the merger of two rival American proposals and features fast packet communications similar to that used to access the internet. There are other proposals, most notably wideband CDMA proposed by Europe and Japan. Regardless of the final 3G protocol specified, the present 2G systems won’t go away soon.
Encoding Differentiates Wireless Systems
In general, digital modulation schemes for wireless phones attempt to improve the use of the available spectrum. They do this in several ways, but the most popular approaches are grouped under the TDMA and spread-spectrum headings. CDMA falls into the spread-spectrum group.
Mobile systems also use various geographical tessellation schemes that subdivide the served area into cells that require less transmitter power. Lowered RF power is a major factor in these systems for two reasons. First, a mobile phone uses a battery, so lower power extends battery life. Second, reducing transmitted power lessens the amount of interference each phone causes to others within the same or adjoining cells. This allows frequency bands to be reused by cells that are closer together, increasing spectral efficiency.
DECT and GSM Represent Europe
One of the simpler TDMA protocols is DECT. In this system, cells can be very small indeed—only a few hundred feet wide. The frequency band between 1,880 MHz and 1,900 MHz contains 10 equally spaced, 1-MHz-wide channels. Modulation is accomplished by FSK, resulting in a +288-kHz carrier shift for a 1 and -288-kHz for a 0.
Twenty-four time slots occupy each 10-ms frame. A packet comprising data, signaling, and error-correction bits, suitably shaped by Gaussian filtering, modulates the carrier within the time slots assigned to a specific user. This means that DECT transmissions occur in bursts. The control of the ramp-on and ramp-off slew rates, corresponding to the leading and trailing edges of the output power burst, is critical. Transitions that are too slow reduce the amount of data that can be reliably transmitted during the fixed burst time. Transitions that are too fast increase interference in adjacent frequency channels.
As with DECT, GSM systems also use power ramping to minimize interference due to switching, but they add frequency hopping to reduce the effect of fading upon the signal. GSM systems also use GMSK instead of FSK modulation. Testing a GSM receiver requires simulation of transmitted bursts with the correct ramp-on and ramp-off timing.
“The simulation of GSM frequency-hopping signals requires good signal purity, fast tuning speed, and complex modulation capability,” said Steve Stanton, product manager at Tektronix. “With options B10 and B11, the Rohde & Schwarz SMIQ series of signal sources can generate correctly power-ramped GSM signals with user-defined data in the occupied time slots. A frequency and amplitude list is used to create the frequency hops desired for the simulation.”
CDMA Appears in the Challenger’s Corner
In a CDMA spread-spectrum system, there are separate frequency bands, but several users may occupy any one at the same time. The key to CDMA performance is the use of mathematically orthogonal pseudorandom sequences to modulate the digital data.
Theoretically, these so-called Walsh code sequences don’t interfere with each other when two or more are active within a given frequency band. In practice, there is measurable interference, and some test instruments allow you to simulate the effect of several Walsh codes simultaneously for more realistic system testing.
A pseudorandom series clocked at the 1.23-MHz chip rate, much higher than the data rate, effectively spreads the power in the data signal across a wide frequency band. Demodulation is done by correlating the received signal with the filter characteristics known to have been assigned to a particular user.
The advantages of CDMA include high spectrum utilization and demodulation gain. Very low transmitter power is used in these systems; but although the signal may be many decibels below the noise floor, it still is accurately recovered.
The complex waveforms present in a CDMA spread-spectrum system can be generated by the Hewlett-Packard ESG series of signal generators. “We offer a personality module that builds the waveforms internally,” explained Eric Worthington, a product marketing manager at HP. “In the ESG generator, there’s a DSP on the arbitrary waveform generator board that produces IF-95 type CDMA or multicarrier CDMA. It also generates multitone and AWGN. There’s a separate option for generating wideband CDMA.”
What About Other Test Requirements?
“For TDMA testing, error vector magnitude and adjacent channel power are important tests for manufacturers. Phase error for GSM and frequency stability for DECT are fundamental tests,” Mr. Worthington continued. “Receiver tests require the data to be specified within each time slot. You can specify data from the signal generator or use externally generated data for this test.” BER and FER tests also are used to evaluate system performance, with or without deliberate impairments.
Going beyond this level of functionality, especially for receiver testing, requires the use of a format-specific single-box tester. These instruments may incorporate actual base-station chip sets and specialized firmware to accurately simulate a real phone channel. In the case of a CDMA system, this implies generation of a pilot signal, the sync signal to align the pseudorandom sequence, paging, and the actual traffic channel together with the required two-way call-setup protocol.
An example of a single-box tester is IFR’s Model 2967 GSM 900/1800/1900 Radio Test Set. In addition to dual-band capability, the tester includes BER/RBER/FER receiver test measurements and support for AMPS, NMT, and MPT 1327. There are similar test sets available for other TDMA standards and CDMA.
When testing phone components, less complex and lower cost equipment often can be used. For example, IMD measurements on amplifiers are meaningful even though the amplifiers may be used in a wireless application. Two CW sources or a single source having multiple outputs with low IMD themselves are required.
Peter Spurr, signal source product manager at IFR, commented that some customers were using SINAD measurements for subassembly testing. They had found sufficient correlation between SINAD values and desired system performance to allow the use of simple analog tests at the subassembly stage. Generally, Mr. Spurr said, simpler tests resulted from an analysis of the actual testing requirements, and often it was possible to avoid simulation of a full air interface.
There are, however, conflicting views about the use of analog test methods in certain situations. CDMA amplifiers, in particular, are subjected to dynamically changing signals. Users join and leave the traffic channel, base-station power control messages are sent as the voice activity factor changes, and cancellation and reinforcement of the simultaneous pseudorandom codes occur. Trying to correlate actual performance to results from simple tests is difficult with complex modulation schemes like CDMA.
“Some customers are testing CDMA amplifiers with multitone generators. CW tones with different phases are summed and used to drive the DUT,” explained Mr. Worthington of Hewlett-Packard. “Because of the phases of the tones, peaks are occasionally generated, and the randomness of the peaks can simulate a real CDMA signal. But, in the frequency domain, there are only a few CW lines instead of the near-continuous CDMA spectrum.
“CDMA testing then started using AWGN to provide a better simulation,” Mr. Worthington continued. “Although that’s fine in the frequency domain, in the time domain it doesn’t result in the same peak-to-average differences you see in a real signal environment.” So, in this example, simplified testing may not yield the desired result.
When a phone is returned for repair, many activities must be performed in addition to comprehensive testing. “Market research reveals that the test function only represents 30% of the tasks that a GSM service center undertakes when processing a mobile phone,” said Tony Brown, an IFR product manager.
The IFR Model 2935 with the PhoneTest software package allows service centers to generate warranty claim forms. The product also provides mobile repair histories, repair hints, and workshop statistics.
It’s easy to see that frequency switching speed is important if many tests need to be performed sequentially in an ATE environment. According to Steve Reyes, marketing manager at Giga-tronics, “Using an earlier generation RF/microwave source with a typical 25-ms switching speed, it took 12.5 s to characterize the gain of an amplifier from 8 to 12 GHz in 10-MHz steps. Using our 12000A Synthesizer with switching speeds between 100 and 500 m s, the same test takes 200 ms.”
The difference in this case is more than the obvious saving of 12.3 s. If the amplifier required tuning, the new generator provides five display updates a second, so the alignment can be done in real time.
Summary
The underlying fact that modern digital mobile phones use complex modulation schemes cannot be overlooked when conducting performance tests. But, neither do you always require the most flexible signal source with arbitrary waveform modulation capabilities to get meaningful results. A wide range of signal sources is available.
Taking a structured approach to your test problem will identify areas where familiar, low cost analog tests are applicable. And having a sound understanding of the many possible pitfalls should help you determine the tests that must be performed more rigorously.
RF/Microwave Test Products
Test Set Covers Analog,
Digital, and Packet Formats
The Model 2959 Advanced Multi-Mode Cellular Phone Test Set handles AMPS, NAMPS, TDMA (IS-54/IS136A) cellular, and CDPD formats. VSELP and ACELP vocorder testing, IQ modulation analysis, handset DTMF signaling, and TDMA time-slot change measurements are supported. Tests include call registration, hand-offs, page responses, SAT/ST frequency error, and antenna/cable loss. You have control of the test level from quick confidence checks to comprehensive RF signaling and analysis. Using the automatic mode, you can select the tests to be performed and the pass/fail criteria for each. $16,995. IFR Americas, (316) 522-4981.
CDMA Instrument Performs Base
And Mobile Station Functions
The Model MT8802A Radio Communication Analyzer evaluates CDMA phone transmission and reception systems based on IS-95 and ANSI J-STD-008. The instrument contains a cellular/PCS base station simulator with an integral AWGN generator; a digital modulation analyzer covering 700- to 1,100-MHz and 1,400- to 2,300-MHz bands; a 3-GHz RF spectrum analyzer; and a thermocouple-based power meter. A digital modulation generator covering 300 kHz to 3 GHz with Rho >0.99 and ±0.1-dB level accuracy relative to other channels; an audio analyzer/generator; an analog modulation analyzer; and a frame/bit error rate tester also are included. $53,000. Anritsu, (800) 267-4878.
Meter Measures and Displays
Complex Signals
The 4530 Series RF Power Meters can make high-speed power measurements from 10 kHz to 40 GHz with the proper sensor. Statistical features such as histograms and CDFs accommodate fast analysis of CDMA, TDMA, and GSM signals. The single-shot sampling rate is 1.25 MS/s, increasing to 50 MS/s for repetitive signals. When used with a peak power sensor, the meters can display power as a function of time, similar to an oscilloscope’s display of voltage vs time. GPIB and RS-232C ports, an RF envelope output, and a programmable DC recorder output are standard. Boonton Electronics, (973) 386-9696.
Arbitrary Waveform Generator
Future-Proofs Modulation
The Rohde & Schwarz AMIQ Modulation Generator is a dual-channel, 100-MS/s arbitrary waveform generator with 14-bit vertical resolution. It supplements the SMIQ vector signal generator by providing virtually unlimited IQ modulation. Included WinIQSIM software automatically calculates a range of IQ-Baseband- and IF signals for established standards as well as emerging ones such as W-CDMA. WinIQSIM can simulate multicarrier formats, gives users access to externally stored waveforms, and can be used to remotely control the AMIQ. Alternatively, the AMIQ can be controlled from the SMIQ generator. Fast local access to stored waveforms is via a built-in hard disk. From $14,950. Tektronix, (800) 426-2200, press 3, code 1086.
Microwave Synthesizer Features
Linear Frequency Sweep
The Model 12000A Series Microwave Synthesizer provides an output range of 10 MHz to 20 GHz based upon a tunable YIG oscillator. The oscillator switches frequency in <500 m s, and the maximum leveled output power is +15 dBm from 10 MHz to 20 GHz. The phase-locked loop controls the analog ramp sweep to within an accuracy of a few hertz. Harmonics are <-65 dBc at +6 dBm and $25,000. Giga-tronics, (800) 726-4442.Comprehensive Options
Extend Generator Capabilities
The ESG Series of analog RF signal generators and the ESG-D Digital Series each have four models. The models are distinguished by their frequency range: 1, 2, 3, or 4 GHz. Features include ±0.5-dB level accuracy, up to 40-MHz FM deviation, and a 10-MHz FM rate; a function generator with dual-tone sine output; and a step sweep mode with programmable frequencies, power levels, and dwell times. Digital capabilities address TDMA, CDMA, and W-CDMA standards through a built-in IQ modulator. Options provide a dual arbitrary generator, a real-time IQ baseband generator, single- and multi-channel CDMA personalities, and BER testing. Hewlett-Packard, (800) 452-4844.
Glossary
2G—second generation
3G—third generation
ACELP—algebraic code excited linear predictive
AMPS—advanced mobile phone system
AWGN—additive white Gaussian noise
BER—bit error rate
CDF—cumulative distribution function
CDMA—code division multiple access
CDPD—cellular digital packet data
CW—continuous wave
DECT—digital European cordless telecommunication
DTMF—dual-tone multifrequency
DSP—digital signal processor
FER—frame error rate
FSK—frequency shift keying
GMSK—Gaussian minimum shift keying
GSM—groupe speciale mobile, global system for mobile communications in North America
IS-95—Interim Telecommunications Industry Association Specification for CDMA
IMD—intermodulation distortion
IQ—inphase/quadrature phase
ITU—International Telecommunications Union
MPT—Ministry of Posts and Telecommunications (U.K.)
NAMPS—narrowband advanced mobile phone system
NMT—Norsk Mobile Telephone
PCS—personal communication service
PM—pulse modulation
RBER—residual bit error rate
SAT/ST—supervisory audio tone/signaling tone
SINAD—signal-to-noise-and-distortion ratio
TDMA—time division multiple access
VSELP—vector sum excited linear predictive
W-CDMA—wideband code division multiple access
Copyright 1999 Nelson Publishing Inc.
February 1999