Introduction This application note is written for those who desire an understanding of the test system configuration and testing of Wireless LAN (WLAN) devices and some of the issues that arise in connection with it. Further detail on many of the topics covered herein may be found in the Appendices. The use of wired Local Area Networks has become ever more commonplace, even in situations where only a few computers need to be connected together. Price reductions have helped stimulate use and so have easier sys- tem configuration and increasing robustness. A number of applications can benefit from the removal of the cable connec- tions needed by a fixed LAN. Remote database access is a good example– from warehouses to retail stores to college campuses. WLAN cards will also be used soon for public Internet access in certain "hotspots," such as airports and hotels, where users are largely stationary and need access to a variety of medium- and high-speed digital services. The IEEE 802.11 Wireless LAN specifi- cation was written to extend the func- tionality provided by the IEEE 802.3 Wired LAN standard. A radio interface adds considerable complexity; howev- er, advances in highly integrated radio circuitry have made it possible to bring the cost of wireless devices down to affordable levels. The ETSI BRAN HiperLAN/2 is an alternative specification for WLAN, with more extensive services, but diminishing commercial support. Its radio frequency (RF) operates in a similar way to 802.11a, although the allocation of transmission time-slots is quite different. Increasing collabo- ration is now taking place between those involved in the two standards. While wired LAN already uses numerous techniques to deal with multiple users who must access a central server, additional steps must be taken to deal with the vagaries of WLAN links. A WLAN link has many less-than-ideal transmission charac- teristics, such as the dependency of signal errors on physical position and the ability of nearby RF devices to "eavesdrop" or interfere. Security is always an important issue in radio transmissions. Considerable effort is being made to ensure that security for WLAN is both adequate and straightforward to apply. This application note begins with a brief description of an IEEE 802.11 Wireless LAN system, emphasizing the radio or physical layer. Consistency at this level provides the basis for wide- spread device interoperability. Comparisons are made to cellular radio systems to highlight the signifi- cant differences in the operation of the two links. Transmitter and receiv- er measurements needed to verify conformance with the IEEE specifica- tion are described, along with infor- mation on how to set up the Device Under Test (DUT) and the test equip- ment. Appropriate equipment from Agilent Technologies is highlighted in Appendix A. Finally, Appendices B and E provide a wide range of refer- ence and learning material. IEEE 802.11 Wireless LAN PHY Layer (RF) Operation and Measurement Application Note 1380-2
Transcript
Introduction
This application note is written forthose who desire an understandingof the test system configuration andtesting of Wireless LAN (WLAN)devices and some of the issues thatarise in connection with it. Furtherdetail on many of the topics coveredherein may be found in theAppendices.
The use of wired Local Area Networkshas become ever more commonplace,even in situations where only a fewcomputers need to be connectedtogether. Price reductions have helpedstimulate use and so have easier sys-tem configuration and increasingrobustness.
A number of applications can benefitfrom the removal of the cable connec-tions needed by a fixed LAN. Remotedatabase access is a good example–from warehouses to retail stores tocollege campuses. WLAN cards willalso be used soon for public Internetaccess in certain "hotspots," such asairports and hotels, where users arelargely stationary and need access toa variety of medium- and high-speeddigital services.
The IEEE 802.11 Wireless LAN specifi-cation was written to extend the func-tionality provided by the IEEE 802.3Wired LAN standard. A radio interfaceadds considerable complexity; howev-er, advances in highly integrated radiocircuitry have made it possible tobring the cost of wireless devices downto affordable levels.
The ETSI BRAN HiperLAN/2 is analternative specification for WLAN,with more extensive services, butdiminishing commercial support. Itsradio frequency (RF) operates in asimilar way to 802.11a, although theallocation of transmission time-slotsis quite different. Increasing collabo-ration is now taking place betweenthose involved in the two standards.
While wired LAN already usesnumerous techniques to deal withmultiple users who must access acentral server, additional steps mustbe taken to deal with the vagaries ofWLAN links. A WLAN link has manyless-than-ideal transmission charac-teristics, such as the dependency ofsignal errors on physical positionand the ability of nearby RF devicesto "eavesdrop" or interfere.
Security is always an important issuein radio transmissions. Considerableeffort is being made to ensure thatsecurity for WLAN is both adequateand straightforward to apply.
This application note begins with abrief description of an IEEE 802.11Wireless LAN system, emphasizing theradio or physical layer. Consistency atthis level provides the basis for wide-spread device interoperability.
Comparisons are made to cellularradio systems to highlight the signifi-cant differences in the operation ofthe two links. Transmitter and receiv-er measurements needed to verifyconformance with the IEEE specifica-tion are described, along with infor-mation on how to set up the DeviceUnder Test (DUT) and the test equip-ment. Appropriate equipment fromAgilent Technologies is highlighted inAppendix A. Finally, Appendices Band E provide a wide range of refer-ence and learning material.
IEEE 802.11 Wireless LAN PHY Layer (RF) Operation and Measurement
Section One: Basic Concepts of IEEE 802.11 Wireless LAN 3
1. BASIC CONCEPTSOF IEEE 802.11 WIRELESS LAN
As the name implies, Wireless LANwas designed to extend the datatransfer function of a Wired LAN.The heritage is important—stan-dards which define how it workscontinue to evolve, but at its heart,WLAN is a system for transferringpackets of digital data wirelessly andwithout error whenever an originat-ing computer can send them. In thisrespect, it is like an asynchronousBluetooth link but unlike a synchro-nous cellular voice connection,which is based on analog transmis-sion. Transmissions take place aftera device has first listened to makesure the channel is clear, a methodcalled Carrier Sense MultipleAccess/with Collision Avoidance(CSMA/CA). It is fundamentally dif-ferent from the rigorous timeslotallocations used in cellular andcordless phones. This may causeconfusion for engineers migratingfrom other technologies.
Software is used to adapt LAN pack-ets for transmission over the radiopath, which, of course, is far lesspredictable than a wired path. Thisprotocol is called Logical Link
Control (LLC) and Medium AccessControl (MAC). User data is encod-ed, headers are added, and longerpackets are broken up (fragmented)before transmission.
The most widely used WLAN sys-tems involve Network InterfaceCards (NICs, also referred to as STAtions), and Access Points(APs). These allow the users tocreate individual wireless-to-wiredLAN links, called InfrastructureBasic Service Sets, or BSS. AnExtended Service Set (ESS) entailsthe construction of a complete net-work. The access point acts notonly to transfer data from wired towireless devices, but is also respon-sible for allocation of the radiochannel to the clients it serves.
In the absence of other users inthe license-exempt bands used byWLAN, a BSS can be configured to make efficient use of the radiospectrum. Even in this situation,throughput for individual users isusually only a fraction of the peakdata rates which WLAN systemsare known by. Typically, the maxi-
mum throughput is half the nomi-nal peak rate and drops even moreas clients are added (there may beno predetermined limit), or as theNIC moves away from the AccessPoint. Coverage depends on boththe physical environment and thefrequency band; however, in manycases the transfer time for a fileover WLAN will be as low as that of a wired LAN because of bottle-necks occurring elsewhere.
Two other WLAN schemes exist. A Peer-Peer (ad hoc) IndependentBasic Service Set, or IBSS, involvesthe use of two NICs. Interoperabilitybetween such cards may be limiteddue to different vendor designs, butotherwise this can be one of themost straightforward ways of effect-ing wireless data transfer betweenmutually friendly devices. Extendersare more specialized system compo-nents which deal with radio propa-gation problems while not acting asspecific network end-points.
1.1Use of Radio Carriers and Modulation
Radio is not the only mediumaddressed by the 802.11 specifica-tion, but it will be the focus of thisapplication note.
Data must be applied to a radio car-rier before transmission. The carriercan be used in several ways:
1.1.1Modes of Carrier Operation
• FHSS – Frequency Hopping Spread Spectrum
A single carrier switches frequencyto reduce the likelihood that it willinterfere with, or be interfered by,other carriers.
Table 1: System frequency bands, data rates,and modulation schemes
TransmitScheme
System
802.11b802.11g802.11a,hHiperLAN/2HiSWAN
2.4GHz2.4GHz5GHz5GHz5GHz
115436, Opt to 545454
OPTION DIFF DIFF
OPTION
NOTE: DIFF=differential modulation encoding.802.11g includes an option for mixed CCK-OFDM.
DIFF DIFF
FrequencyBand
Max DataRate (Mbps)
Modulation
64 QAM
16 QAM
QPSK
BPSK
OFDM
PBCC
CCK
4 Section One: Basic Concepts of IEEE 802.11 Wireless LAN
• DSSS – Direct Sequence Spread Spectrum
The energy in a single carrier isspread over a wider spectrum bymultiplying data bit(s) with a special11-bit pattern, called a Barker key.This is done at a chip rate of 11MHz.This technique can help reduceinterference from narrow-bandsources. The IEEE 802.11b -1999specification uses an 8-bit key.
Two schemes are used by 802.11b to spread the spectrum of a singlecarrier. CCK (Complementary CodeKeying) is mandatory, while PBCC(Packet Binary Convolutional Coding)may be added. Channel agility mayalso be added as an option.
• CCK – Complementary Code Keying
This is used to increase IEEE802.11b's peak data rate from 2 to 11 Mbps, while still using QPSK(Quadrature Phase Shift Keying)modulation. It does this by firstincreasing the data clock rate (sym-bol rate) from 1Mbps to 1.375Mbps,then taking data in 8-bit blocks(8*1.375 = 11). Six of the 8 bits areused to choose 1 of 64 complementa-ry codes, which are 8 chips long andclocked out at 11MHz. Thus all 8chips are "used up" in (1/1.375) µs –the time before another byte is ready.The other 2 bits are combined withthe code in the QPSK modulator.
• PBCC – Packet Binary Convolutional Coding
This scheme is optional for IEEE802.11b and g. It makes use ofForward Error Correction to improvethe link performance when noise isthe limitation. Scrambled data is fedinto a convolutional encoder. Theencoder consists of a 6-stage memory,with specific taps combined to givetwo outputs. The four possible out-put states (00,01,10,11) are mappedinto two possible QPSK states(11Mbps). A codeword controlshow the chosen state alternates
over time. The RF modulator isdriven from this point.
IEEE 802.11a, HiperLAN/2 andHiSWAN use OFDM.
• OFDM – Orthogonal Frequency Division Multiplexing
OFDM uses multiple carriers, ofwhich there are 52, spaced 312.5kHzapart. Data is sent on 48 carrierssimultaneously, with 4 used aspilots. The time to transmit eachbit increases in proportion to thenumber of carriers. This makesthe system less sensitive to multi-path interference, a major sourceof distortion.
This note concentrates only onDSSS and OFDM systems. Some sys-tems, such as 802.11g, may use bothmethods during the same RF burst,making them more compatible with802.11b systems.
• Modulation
The RF carrier(s) must be modu-lated. All the WLAN systemsdescribed in this note use a formof phase-shift keying for the pre-amble. More complex schemes,such as 64QAM (Quadrature
Amplitude Modulation) give fasterbit rates for user data, but requirebetter radio performance and lessnoise to work to their full poten-tial. BPSK (Phase Shift Keying),QPSK, and QAM are described instandard RF texts.
Often, the modulation formatchanges during the transmission.This is because the early part of the burst contains importantinformation about the burst, includ-ing analog characteristics such asfrequency, and digital informationsuch as burst length. Simpler mod-ulation formats are less prone tobit errors, and thus are more suit-able to use early in a burst.
1.1.2Frequency Bands and Power Levels
WLAN systems operate in one of thefrequency bands shown in Figure 1below. The maximum transmit pow-ers are also shown. Transmit PowerControl and Dynamic FrequencySelection, part of the HiperLAN/2specification, will be added to802.11a operation to satisfy Euro-pean regulatory requirements.
Figure 1:
Major channel allocations and power levels
Section One: Basic Concepts of IEEE 802.11 Wireless LAN 5
1.2Anatomy of a WLAN device
Physically, the most common formatof a WLAN device is that of a PCCard (PCMCIA), suitable for directconnection to a laptop computer.Access points may simply have a PCCard mounted on a motherboard.
Electrically, the WLAN card is splitinto two major sections: the analog RF(PHY layer) and digital Baseband(MAC, or Medium Access Control)processing. The connection to the hostcomputer is generally through the PCCard or Compact Flash interface.
In an access point, additional digitalcircuitry is used for the interfacewith a wired LAN (i.e., via Cat 5/Cat6 cabling). Some designs providepower from the LAN wiring ratherthan from a separate power supply.
1.2.1Description of Operation
Figure 2 below is a generic blockdiagram of a radio system. As withmost electronic systems, newer radiodesigns have higher levels of integra-tion, although performance trade-offsmust be made. This applies particu-larly to the receiver. Some systemswill not use, or provide access to,the intermediate signals discussedin this note. Readers are advised tocarefully study the block diagramof the system they must test.
The Local Oscillators (LOs) shareboth transmit and receive functions.Frequency doubling or tripling isused (although not shown in the diagram) to give better VoltageControlled Oscillator (VCO) performance and to isolate the RF output from other signals.
1.2.2 Data Reception
Diversity reception is used to reducethe effect of nulls on signal levels. AReceive Signal Strength Indication(RSSI) test, made during the shorttraining sequence, determines whichpath is switched in for a particularburst. The chosen signal is fedthrough an amplifier/downconver-sion chain before being mixed into apair of quadrature signals which aredigitized. Analog gain control meansthat 6 to 8 bits is usually sufficientfor the A/D conversion. Some hybridschemes use DSP (Digital SignalProcessing) for the IQ (In-phaseQuadrature) separation and there-fore need only a single connectionfrom the analog circuit.
An equalizer and other componentsof digital circuits are able to reducethe effect of distortions such as frequency error or amplitude varia-tions, but the design itself mustensure that others errors—such ashigh-frequency local oscillator phasenoise—are low enough to guaranteethe needed link performance.
The RF portion of 802.11b, whichis difficult to make small and inex-pensive, is not so challenging interms of Bits/Hz. However, thehigher data rates of 802.11a andthe doubling of channel frequencymake it far more difficult to designand manufacture.
Receiver sensitivity is importantbecause it determines the maximumrange over which a WLAN link canoperate. There are secondary sys-tem benefits as well. If one linkcompletes a transmission fasterthan another because the PacketError Rate is lower, battery con-sumption will be reduced and lessinterference will occur to otherusers. In a real-world Industrial,Scientific and Medical (ISM) orUnlicensed National InformationInfrastructure (UNII) environment,interference suppression and lineari-ty will directly affect the perform-ance of the radio. They are thereforeimportant test parameters. It maybe more difficult to distinguishbetween causes of poor performanceas hardware becomes more andmore integrated and if special testmodes are not available.
1.2.3Data Transmission
Transmitter performance require-ments usually necessitate an externalpower amplifier (PA). Cost, currentconsumption, and linearity combineto demand considerable attention todetail in this choice. Pre-distortionof the signal from Baseband process-ing may allow less stringent PAdesign, but it may also limit thechoice of devices used if it relies
Figure 2:
Block diagram of a WLAN card
6 Section One: Basic Concepts of IEEE 802.11 Wireless LAN
heavily on a particular performancecharacteristic. Diversity transmis-sion may also be applied (theswitching in Figure 2 does not allowfor this).
Designs with SAW (Surface AcousticWave) or Dielectric Bandpass filter-ing in the IF (Intermediate Fre-quency) are suited to normal TDD(Time Division Duplex) operation,but also have fewer options forinternal loop-back paths for testingand self-calibration. Self-calibrationbecomes important when the per-formance of the analog circuit isaffected by temperature.
Differential signal paths are becom-ing more common as power supplyand signal voltages decrease andbackground noise becomes moreprominent. Balun transformers canbe used when single-ended signalsare needed.
Although the analog hardware canbe tested in isolation, it needs to becombined with DSP (Digital SignalProcessing) of the Baseband circuitin order to comprise a completetransceiver. Care is needed whenmodeling total system performancebecause a number of error contri-butions may not be just simplearithmetic additions, but resultfrom analog and other phenomena.
Algorithms within the DSP playan essential role in both transmis-sion and reception. Figure 3shows an example of the mainprocessing blocks needed for802.11a/ OFDM system.
1.3Time Division Duplex and FrameStructure
A WLAN device can only transmitor receive at a single time. Trans-missions occur as bursts (frames)which vary in length and spacing,usually in the range of a few hun-
dred microseconds to one milli-second. The 802.11b CCA (ClearChannel Assessment) receiver testspecifies the longest possible 5.5Mbpsframe—3.65ms.
The basic structure of the frames isshown in Figures 4 and 5. The pre-amble is used by the receiver toadapt to the input signal. This mayinvolve frequency and phase error
equalizing, as well as time alignment.The header contains a wealth ofinformation, including the destina-tion address and the format of theremainder of the burst. User data istransferred from the original pack-ets, which are fed into the MAClayer. Long packets may be fragment-ed (broken up) if the radio deter-mines that this will improve linkperformance.
Figure 4: IEEE 802.11a frame structure
Figure 5: IEEE 802.11b frame structure with short/long sync
Figure 3: Digital processing blocks for an OFDM transceiver
Section One: Basic Concepts of IEEE 802.11 Wireless LAN 7
Five time periods in the 802.11 spec-ification determine the spacingbetween transmissions. The physicalvalues vary according to the stan-dard used and are shown in Table 2.
The combined effect of using CSMA/CA, data which is sent when readyand different time intervals betweenframes, is to produce seeminglyrandom spacing between bursts.
1.4 The Medium Access Control Layer
The Medium Access Control (MAC)layer provides an asynchronous datapacket delivery service to the LLCsoftware that uses it. This means thatit is impossible to predict exactlywhen transmissions will take place.The MAC software takes the data and,identifying the PHY layer with whichit is working, transports the data tothe MAC layer software in the client’sreceiver. Three types of MAC Framesare used to provide this service:
• Management Frames
Eleven sub-frame types are used forlink management. They provide themeans for establishing and terminat-ing a link—beacon transmission andprobe requests, authentication, andassociation.
• Control Frames
Six sub-frame types are used tomake sure the link functions cor-rectly: Request To Send, Clear ToSend, ACKnowledged, Power Save,Contention Free, and CF–End +CF–Ack.
• Data Exchange Frames
Eight sub-frame types are usedand all except one contain user
data. Most of them make the linkmore efficient by adding link con-trol information.
Figure 6a (below) shows how theseframes are used (not to scale). Figure 6b shows the major frametiming options.
1.5Establishing Contact
Starting when a device is poweredup, software above the MAC layermust stimulate the device to estab-lish contact. Either active or passivescanning is used. The IEEE specifi-cation allows for different implemen-tations, so characteristics may differbetween devices.
1.5.1 Active Scanning
Active Scanning is the fastest way to establish contact, but consumesmore battery power. Listening for a
clear channel, the device whichseeks to establish contact sends aProbe Request. If the Service SetIDentity matches, the recipient thensends a Probe Response. The scan-ning device uses this information todecide whether or not it will join the(I)BSS, but there is no further trans-mission at this point.
1.5.2Passive Scanning
In Passive Scanning, Beacons andProbe Requests are used. Afterselecting a channel, the scanningdevice listens for Beacons or ProbeRequests from other devices. ABeacon is transmitted from anAccess Point in a BSS or ESS. It con-tains information about the AP anda timing reference. Beacon transmis-sions occur on a 1024µs time grid,spaced roughly every 100ms. Likeother transmissions, they are subjectto a clear channel test and so maybe delayed.
Time Interval 802.11a 802.11b
SIFS - Short Inter Frame Space 16µs 10µs
SLOT 9µs 20µs
Priority IFS = SIFS + SLOT 25µs 30µs
Distributed IFS = SIFS + 2 SLOTS 50µs 50µs
Contention Window Min 15 slots 31 slots
Extended FS - much longer Variable Variable
Table 2: IEEE 802.11 Timing Intervals
Figure 6a: Outline process to send data (IEEE 802.11)
Figure 6b: Frame timing (IEEE 802.11)
8 Section Two: PHY Layer (RF) Test Suite
1.5.3Authentication
Before any user data is transferred,the Sender and Receiver must agreethat they are ready to talk. Theseare processes of authentication(which happens first) and associa-tion. There are several others (e.g.,random exponential transmissionback-off) which can be used if adevice finds that the channel is notclear when it is ready to transmit.
Earlier, a list of modulation rateswas shown for the different 802.11standards. The specifications do not define exactly how these areselected. Different designs willemploy different proprietary algo-rithms. Currently, there is no "qual-ity of service" standard in the MACframes. However, the transmittingdevice is able to gauge the fidelityof the link based on the regularityof ACK frames coming back fromthe device it is seeking to contact.
1.6Exchanging Data: Two Methods
When two WLAN devices are readyto exchange data, they must chooseone of two methods. The decisiondepends on the expected perform-ance of the radio link.
1.6.1Two-step ExchangeTwo-step data exchange is simplythe sequence:• Send • Acknowledge
This is good for short packets or a sparsely-used RF environment.
1.6.2Four-step ExchangeMore typically, however, exchangeis a four-step process:• Request To Send (RTS)• Clear To Send (CTS)• Send• Acknowledge
This is used for long frames, wherethere is a higher likelihood of inter-ference with, or from, an RF-noisyenvironment. A special MAC signal(dot11RTSThreshold) is used tochoose between frame lengths.
2. PHY LAYER (RF) TEST SUITE
The measurements described beloware used mainly to determine if aWLAN device conforms to a relevantstandard. The only mandatory RFtesting is in the regulatory statutesof various countries.
Transmissions using an antenna orlive network may be unpredictable.While a number of tests describedhere can be performed live, ascreened RF connection and/orenvironment is normally used. Thisis essential for repeatable receivermeasurements. Signals at 2.4GHzand 5GHz act very differently fromthe audio and digital signals withwhich most people are familiar.
Readers are encouraged to seekcompetent technical advice if theywish to perform RF measurementsand are new to the subject.
3. TRANSMITTER MEASUREMENTS
If not controlled, many transmitterparameters can reduce the perform-ance of WLAN systems, or evenprevent RF devices from workingtogether. Tests have been devised to prevent this from happening.Table 3 provides a summary.
Transmitter tests are described firstbecause some transceiver problemscan be found quickly by analyzing
Table 3: Summary of IEEE 802.11a,b transmitter tests and configurations
IEEE Test Packet Type Payload Instrument ConfigurationRef.#
18.4.7.1 Transmit Power Longest Framed PN9(15) Edge Trigger, using Trigger hold-off18.4.7.2 ≥1024 byte (set to frame length)17.3.9.1 payload
Transmit Power Longest PN9(15) Measurement BW ≥18MHz to capture peak or “Properly adjusted” for limitations in measurement BW
Power Density Longest PN9(15) RBW 1 MHz802.11a Detector type, Sweep Time: See notes
18.4.7.6 Power Rise/Fall Longest RBW ≥18MHz, VBW ≥1 MHz,802.11b to suit <2µs rise time
18.4.7.7 RF Carrier Longest 0101 DPSK RBW 100kHz, VBW 100 kHzSuppression Framed Scrambler OFF Detector type: See notes802.11b
17.3.9.6.1 Center Frequency Longest Reference and result both measured Leakage Framed during channel estimation part of burst802.11a
17.3.9.6.2 Spectral Flatness Longest PN9(15) Use Channel Estimation from pre-802.11a Scrambler ON measurement equalizer.
18.4.6.8 Transmission Longest PN9(15) RBW > 1MHz. Spurious Scrambler ON If smaller, integrate result to be equiv-
alent of 1MHz [Quasi] peak detector
17.3.9.4 Center Freq. Longest PN9(15) 802.11a requires Tx clock & symbol 18.4.7.4 Tolerance Framed Scrambler ON clock to come from the same oscillator.
Method depends on Test Mode
17.3.9.5 Symbol Clock 802.11a Tested by inference from Freq. Tolerance RF Center Frequency
17.3.9.6.3 Constellation 802.11a PN9(15) An equalizer is used before theError 20 frames Scrambler ON measurement. Works on Short 802.11a ≥16 OFDM or Long training sequence
symbols
18.4.7.8 Error Vector Unframed 1111 Equalizer removes frequency error Magnitude Scrambler ON before measurement802.11b
LongestUse unframedTest mode signalor time-gating
PN15Scrambler ON
Average
Peak
Note: Packet types, payloads, and measurement configurations, shown in italics, are recommendations for testing where a specification is unclear. Test refrence numbers which start with 17 apply to IEEE 802.11a. Test refrence numbers which start with 18 apply to IEEE 802.11b.
Section Three: Transmitter Measurements 9
transmitted output first. Exami-nation of the diagram in Figure 2(pg 5) shows why this is the case:The local oscillator(s) (LOs) for fre-quency up- and downconversionare shared, so many impairmentson an LO which could affect thereceiver will be immediately evidentin the transmissions.
3.1Test Conditions and Measurement Setup
Two main configurations are usedfor testing the transmitter path.They are distinguished by the signalinterfaces and the way the device iscontrolled. One is suitable for onlyRF/analog circuitry, while the otheris applicable to a complete WLANdevice. Figure 7 (right) shows theconfiguration for the RF/analogcase. Control of the circuit requiresproprietary hardware, but the IQBaseband signal for driving thetransmit path can be provided by an ESG-C signal generator.
Some WLAN designs have an inter-mediate signal available, usually inthe 100s of MHz range. It may beuseful for selecting a signal analyzer/signal generator with coverage atthis frequency.
The antenna in a real-world systemis often designed to focus transmitpower in a certain direction andwill have a radiation efficiency thatdepends on the exact implementa-tion. This can make it difficult tocompare the performance of differ-ent pieces of RF hardware. Thussome measurements refer to EffectiveIsotropic Radiated Power (EIRP).Physical measurements involve theuse of a remote antenna for testing,which can be very impractical.Instead, theoretical calculations canbe used to provide a correction coef-ficient for measurements made overa direct cable connection.
3.1.1Measurement Triggering
Measuring framed RF signalsrequires the use of a trigger signal.Many test instruments have this asan internal feature. In 802.11a,HiperLAN/2, and 802.11b, a compli-cation occurs because of significantvariation in signal level due to mod-ulation. Careful selection of thetrigger level may solve the problem.Alternatively, the Trigger Hold-offfunction may be used if the frameperiod is fairly regular. Trigger Hold-off acts to disable the trigger circuitfor a defined period of time after atrigger signal has been acted on.
3.1.2Interaction with DSP
Often, a complete WLAN card mustbe tested. It is only at this point thatproblems of interaction betweenthe Digital Signal Processing (DSP)and analog circuitry are uncovered.These can be as straightforward aspower supply de-coupling or as com-plex as electromagnetic coupling.Figure 8 shows the configurationneeded to test a complete WLANdevice. It is simpler than Figure 7because the MAC processor providesthe stimulus. The results are there-fore the combined effect of DSPcontrol in the MAC and the analog
Figure 7: Transmitter test configuration for RF/analog circuitry
Figure 8: Transmitter test configuration for a complete WLAN card
10 Section Three: Transmitter Measurements
circuit performance that follows it.Tests may be constrained, in runtime as well as circuit evaluation,by MAC control software availableto the user.
3.2Test Modes
The IEEE 802.11 specificationdescribes various Test Modes, whichcontrol the operational state of theradio and a number of transmitparameters. These are more likelyto be used when testing a completeWLAN Card. They are shown inTable 4. Other parameter optionsmay become available for 802.11a.
The IEEE 802.11 specification doesnot include any over-the-air controlof Test Mode functions. Hence it isoften not a straightforward task toprovide the complex, integrated test-ing expected of modern wirelessdevices. Proprietary device controlsoftware may also be needed. If TestModes are supported, a secondWLAN device, shown in Figure 8,is usually not needed. However, itmay be a useful addition to handleoperational (functional) testing.
A secondary effect of the inde-pendence of test equipment anddevice control is that system con-trol software must pay specialattention to the triggering andtiming of measurements.
3.3 Transmitter Power
The power measurements describedbelow will be affected by loss and theVoltage Standing Wave Ratio (VSWR)of cables and other RF componentsused in measurement. It is impor-tant, both at 2.4GHz and 5GHz, touse components suitable for thefrequency range.
The 802.11 standard does notspecify a tolerance limit for theimpedance of the antenna ports orthe frequency range to be tested.A port match of 10dB (VSWR~2:1)is often used, representing animpedance variation of between25 and 100 ohms. Combined withat the antenna, but no other losses,this can produce signal variationsof up to ±1dB. In addition, mis-matches at harmonic frequenciesmay cause amplifier non-linearityand produce modulation qualityproblems. These should be ana-lyzed during design.
3.3.1Average Output Power
If a device’s transmitter and receiverare working correctly in all otherrespects, then average transmitpower will be the main factor affect-ing the WLAN's coverage area.
All normal WLAN transmissions areframed (bursted). If a fixed mark-space ratio can be generated by thetest software, it is possible to per-
form a power measurement usingdetectors with average responses. A correction factor can be used tocorrect the lower reading obtained.This technique can be a little labori-ous, however. The measurement inFigure 9 was triggered from the burstedge and is thus simpler to make.
It is common to make just an aver-age power reading using unframeddata. This gives an indication of theoperating level of the device, butremoves all information about thedynamic performance of the trans-mitter path, unless done in conjunc-tion with a modulation quality orspectrum test.
3.3.2Peak Output Power, ComplementaryCumulative Distribution Function (CCDF)
In a system using bursted RF, powerpeaks are often associated withturn-on spikes. As Figure 10 shows,for IEEE 802.11a and HiperLAN/2,significant peaks can be seen duringthe entire burst. In a standard sig-nal, the peaks can be as much as11dB higher than average over a sin-gle frame. If the peaks are not trans-mitted correctly, the receiver willrecord Bit Errors or Packet Errors.Thus the overall link quality will bereduced.
IEEE 802.11b also uses modulationformats which cause variations of2-3dB above average power through-out the burst. These variations can
Table 4: Description of 802.11 test modes
Name Type Valid Range Description
TEST_ENABLE Boolean True,false True enables test mode
Figure 9: Average power measurement of IEEE 802.11a, using EPM-P
Section Three: Transmitter Measurements 11
be very short, so a wide measure-ment bandwidth is needed to accu-rately characterize the signal. If themeasurement bandwidth is less thanthe signal bandwidth, it is impossi-ble to track peaks accurately. Forexample, if we assume that signalpower distribution is Gaussian andthat video measurement bandwidthis 5MHz, the peak power level willread low by 2-3dB. The averagepower, however, will be correct towithin 0.5dB. Regulatory require-ments usually stipulate a "true peakmeasurement for the emission inquestion over the full bandwidth ofthe channel." This implies a mini-mum capture-bandwidth of 18 MHz.
Fortunately, measurement methodshave been developed to deal with thevariable amplitude formats of 802.11.If we look at the Power vs. Time plotin Figure 10, we can see that the
highest powers occur relatively infre-quently. The probability that the sig-nal will exceed a certain valuedecreases as the power thresholdincreases. If some peaks are clipped,it is useful to know how often thishappens in order to allow adjust-ment of amplifier bias currents.
Figure 10 also shows a plot of PowerLevel (horizontal axis) vs. Prob-ability (vertical axis). This is knownas a Complementary CumulativeDistribution Function (CCDF) plot.The left-hand power reference isautomatically set to the average ofthe measured signal—in this case, -22.858dBm.
The gray curve is the equivalent of Gaussian distributed noise and serves as a guide to possibleproblems.
When required, calibration of theabsolute power level of the signalanalyzer may be carried out usinga CWRF (Continuous Wave RadioFrequency) signal and powermeter. Automated functions maybe available in the test equipmentand software.
3.3.3Transmitter Power Control
Control of the transmit power levelis part of the HiperLAN/2 specifica-tion and of IEEE 802.11b for trans-mit powers greater than 100mW. It isnot part of 802.11a, but will be intro-duced with the 802.11h specifica-tion. This is to meet EuropeanDynamic Transmit Control require-ments, which are intended toaddress some of the needs of otherusers operating at 5GHz.
Note: Measurement gating must be used to avoid including a zero power reading between bursts.
Figure 10: Combined display plot, including IEEE 802.11a CCDF measurement
12 Section Three: Transmitter Measurements
HiperLAN/2 has different require-ments for the Access Point and theMobile Terminal (MT). The AP has16 power levels, ranging from+30dBm to –15dBm, in 3dB steps.The MT has transmit power intervalsof –15dBm to -9dBm, -9dBm to+9dBm, +9dBm to +18dBm, and+18dBm to +30dBm.
The configuration for testing powercontrol is the same as for outputpower, with the addition of triggersignals to ensure that measurementsare made only after the device hassettled at a new power level.
Note: WLAN systems may not havethe high-speed power control algo-rithms of some cellular systems.Separate test control may be requiredfor design and manufacture.
3.4Transmit Output Spectrum
3.4.1Input Attenuation Settings
The measurement bandwidths usedfor swept spectrum measurementsare considerably less than the signalbandwidth—100kHz, as compared to18MHz. This has the effect of reduc-ing the readings displayed for nor-mal markers. However, the full signalpower is fed to the input mixer ofthe signal analyzer, which will intro-duce distortion if it is overloaded.Therefore, when using manual con-trol of input attenuation, adjust-ments should be made on the basisof total signal power. This may beshown as Channel Band (or Total)Power on the instrument.
3.4.2Transmitter Spectrum Mask
This test is used to ensure that mul-tiple WLAN devices do not undulyinterfere with each other. It is asso-ciated with the adjacent channelreceiver tests in section 6.6.
The need for a linear transmitterpath was described earlier.Inevitably, a designer has to make atrade-off between average transmitpower and distortion. While not adirect measure such as Packet ErrorRate, the spectrum test can be agood indicator of deteriorating per-formance. Visible changes in thespectrum are likely to occur wellbefore Packet Errors occur, due tomodulation errors. A spectrum testmay be combined with Error VectorMagnitude measurements to get tothe root cause of a problem, but notethat some EVM errors get worse asthe spectrum improves.
The IEEE 802.11 standards do notspecify a transmit modulation filterfunction; however, filtering isimplied in the spectrum mask.Figure 11 shows an 802.11b signalwhere the modulation filtering iscorrect, but where the designer hasforgotten to include an anti-alias(reconstruction) function.
• Signal Framing
The IEEE 802.11b spectrum mask isthat of a continuous signal. Thedrawback with configuring a trans-mitter for continuous output is thatthe final result may not be represen-tative of actual performance underburst conditions. Some standardsare explicit in the need to useframed signals.
Several options are available formeasuring the spectrum in practice:
•The transmission may be modifiedto produce a continuous(unframed) signal
•A slow sweep period can be usedto give a complete display of aframed signal
•Some form of gated spectrumanalysis may be performed on theburst using a vector signal analyzer
Figure 11:
IEEE 802.11b signal failing spectrum test mask.
Regulatory tests effect outer limit line levels.
Section Three: Transmitter Measurements 13
• Detector Type, Sweep Time, Signal Payload, Reference Level Setting
Spectrum masks are relative, basedon a reference level measurement.Fast sweep times can cause varia-tions in the reference level takenfrom one sweep to the next. Thishappens if the display point doesnot include all the level variationsseen through the burst (see notes inPeak Power measurement, section3.3.2). If the payload data variesfrom one frame to the next, it willincrease the likelihood of referencelevel variations.
A 500ms sweep was used for thereference measurements in Figures11 and 12.
The IEEE 802.11 standard does notspecify the type of measurementdetector to be used. Empirically,the shape of the spectrum (for anunframed signal) is similar usingeither a swept spectrum analyzeror vector signal analyzer.
A number of other communicationstandards, including GSM andBluetooth, now use zero-span, multi-ple offset measurements for adjacentchannel testing. An average detectoris used. A continuous sweep, using arelatively slow sweep time, will givesimilar results
A swept spectrum analyzer withan average detector was used forFigures 11 and 12. The measurementis not gated. The spectrum responseto the complete transmission ismeasured, but note that the absolutelevel will vary for a framed signal,depending on the mark-space ratioof the signal. The Average Detectorfunction in the PSA/ESA spectrumanalyzer supercedes the techniquepreviously offered by Video Filtering.
Note: Manually setting the VideoBandwidth too low when usingAverage Detector will cause anUncal error message to appear.
A peak detector will produce asimilar-looking plot if there areno short transient effects. Theabsolute level will be higher, plusit is more sensitive to changes inthe reference level setting due todifferent data patterns.
A vector signal analyzer can alsoprovide good results, and can beprogrammed to synchronize themeasurements with specific pointsin the frames. This can result inmore stable reference level readings.
3.4.3Power Density
Given the "noisy" nature of IEEE802.11a, some measurements likethe reference measurement for thespectrum test are defined as powerspectral density. Expressed as powerwithin a specified bandwidth, theresult is nominally independent ofthe measurement bandwidth. Themeasurement should generally bemade under realistic operating con-ditions. Variations will occur if thesignal level varies with frequencybecause of the data pattern beingtransmitted. The data pattern orscrambling state should be specified.In Japan, the maximum transmittedpower is expressed as 10mW/MHz.In the United States, a conversion
Figure 12: IEEE 802.11a signal measured using Agilentperformance spectrumanalyzer
Figure 13: Power spectral density meas-urement made on an IEEE802.11a signal with 01 patternand scrambling off, showingripple due to data pattern
Note: If variations in individualchannel powers are suspected,they can be separately identifiedwith the gated channel responsemeasurement in section 3.4.6.
14 Section Three: Transmitter Measurements
factor of 16 is used to go from apower density of, say, 2.5mW/MHzto 40mW/MHz.
3.4.4IEEE 802.11a Center Frequency Leakage
Energy at the center frequency of the carrier can cause problemswith receiver designs which usezero frequency intermediate signals.IEEE 802.11a specifically avoidsusing the center carrier for trans-mission. The measurement is gatedover the 8µs-channel estimationsection of the preamble. This isthe part of the waveform whenevery 4th carrier is turned on.
3.4.5IEEE 802.11b Carrier Suppression
The normal spectrum for an 802.11bsignal may not show a noticeablelevel reduction at the center frequen-cy. A "01" test pattern, with scram-bling off, is used to create the rightconditions. The reference is definedas the value of the highest signalfound. The carrier must be at least15dB below the highest signal found.
3.4.6Spectral Flatness
This test applies only to the OFDMsignals in IEEE 802.11a andHiperLAN/2. Variations in carrierflatness will reduce demodulationmargins and degrade link perform-ance. It is measured during the8µs-channel estimation phase of theburst, with all 52 carriers are turnedon. It is 8µs after the start of a nor-mal burst.
Note: Filters in the measurementpath may affect spectral flatness.Calibration or normalization maybe used to remove linear errors.
Figure 14: Plot showing center frequency leakage of an IEEE 802.11a signal
Figure 15: Plot showing RF carriersuppression of an IEEE 802.11b signal
Figure 16 IEEE 802.11a spectralflatness plot
Section Three: Transmitter Measurements 15
3.5Modulation Tests
3.5.1Constellation Error
Directly referred to only in IEEE802.11a, the constellation error ismeasured as under EVM, describedbelow. It is reported as a decibel(dB) value, dependent on the modu-lation rate being tested. Interactionmay occur between problems whichcause high constellation errors andthose which cause Spectrum Maskfailures. Many modulator errorscause different effects when usingOFDM rather as opposed to a singlecarrier. For more detailed informa-tion, see the References.
3.5.2Error Vector Magnitude
IEEE 802.11 uses a metric calledError Vector Magnitude (EVM) as ameasure of modulation quality. Ithas become an industry standard fora wide variety of applications, fromcellular phones to cable television.The basic concept of EVM is that anyimpaired signal (usually a complex
one) can be represented as the sumof an ideal signal and an error sig-nal. Since an error signal cannot bemeasured directly, test instrumenta-tion determines the error signal byreconstructing the ideal signal basedon received data and then subtract-ing it from the actual signal.
The error signal encompasses allsources of error, including:
At any time, the error signal is rep-resented as a complex vectorextending from where we are in theIQ plane to where we want to be.Every chip has its own error vector.EVM is defined as the root meansquare (rms) over 1000 chips.
• Combining EVM Readings
The list below shows that some errorsources are noiselike, while othersare systematic. EVM is the combina-tion of many factors, so it is not pos-sible to define an exact mathemati-cal relationship between EVMs com-ing from different components in atransmitter or receiver. However,non-coherent errors will usually addin on a root-sum-of-squares basis.Device simulation is therefore need-ed to perform accurate analysis.
New techniques are being developedto more readily isolate the causes of distortions. Up-to-date informa-tion may be obtained from theAgilent web site, www.agilent.com/find/wlan, or from your local Agilentsales representative.
The EVM specification of IEEE802.11b is a very generous 35%. Thiswould be poor for a QPSK signal, butis reasonable for DSSS given becauseof the coding gain due to spectrumspreading. The situation is very different for 802.11a,g. Table 5shows how the EVM specificationvaries with bit rate. Data rates above24Mbps are optional for IEEE802.11a.
Figure 17 The elements used to define EVM measurement
PHASE ERROR (t)
Carrier Leakage
ERRORVECTORMAGNITUDE (t)Error (t)
Ideal (t)
Actual (t)
MAGNITUDE
ERROR(t)
Data Rate Relative Constellation EVM(Mbit/sec) Error (db) (%rms)
HiperLAN/2 802.11a
6 -19 -5 56.2
9 -19 -8 39.8
12 -19 -10 31.6
18 -19 -13 22.3
24 -19 -16 15.8
36 -19 -19 11.2
48 N/A -22 7.9
54 -24 -25 5.6
Table 5: Constellation error and EVM equivalent for IEEE 802.11a
16 Section Three: Transmitter Measurements
In the same way a Power vs. Time ora Gated Spectrum plot give a lotmore diagnostic information than anumeric power reading, so EVM vs.Time and Channel (for 802.11a) canalso provide more information aboutthe cause of problems. Figure 18 is acombination display indicating char-acteristics of a timing error. Thesolid lines across the bars showthe RMS (Root Mean Square) EVMvalue at each point.
• Center Frequency Tolerance
The method for measuring centerfrequency tolerance depends on theavailability of test modes. If modula-tion is turned off, it may be as sim-ple as using a frequency counter,but a more accurate and realisticresult would be gained from a normalmodulated signal. This is becausechanges in current consumption dueto analog or digital circuits changingstate can cause transient changes in the local oscillator frequencies.The equalizer in the receiver of areal device has to use part of thetransmitted preamble, so it isimportant that this be correct.
In Figure 19a the frequency erroris reported as a by-product of thedemodulation process. The binary
data shown is the content of the preamble. Spaces representcarrier 0. Figure 19b shows howthe frequency may change duringthe frame.
HiperLAN/2 and future 802.11specifications have requirements forDynamic Frequency Selection. Thisallows the WLAN system to adaptto the presence of RADARs. Thecarrier must switch within 1ms inHiperLAN/2.
The IEEE 802.11b specificationallows for a Channel Agility option.The method of operation is notmandated. As an indication of thesettling time required, the operatingchannel frequency has ≤224µs tosettle within ±60kHz of its final value.
3.6Transmitter Bit Error and PacketError Rates
The modulation tests described sofar require wide-band measurementequipment. An alternative transmit-ter test involves making PacketError Rate (PER) or Bit Error Rate(BER) tests with a good "golden" ref-erence receiver. While this test maybe simple and convenient, it has anumber of serious limitations:
•The result depends on the receiver’s analog circuit perform-ance, which may be difficult toreproduce in low-cost hardware.
•The result depends on thereceiver’s data recovery algorithms (e.g., Viterbi), which may varyfrom one design to the next.
•Packet Errors indicate that allother performance margins havebeen used up. This is not a goodway to get a warning of perform-ance deterioration!
EVM and spectrum measurementsprovide far more information abouttransmitter performance. For thesereasons, this application note willnot provide further detail on Trans-mitter PER or BER measurements.
Figure 18: EVM vs. Time and vs. Channel for an IEEE 802.11a signal with timing error
Figure 19a: Frequency error from an IEEE 802.11a device
Figure 19b:Frequency change during frame
Section Four: Timing Tests 17
4. TIMING TESTS
WLAN systems are Time DivisionDuplex (TDD), so switching goesfrom transmit to receive. The use ofClear Channel Assessment meansthat the transition from one state toanother needs to be short and well-controlled.
Every individual station starts withits own timing reference. This isused to establish both transmissionand reception intervals. To workeffectively within a network, sta-tions need to synchronize theirtimers. This is done prior to associ-ation and then for as long as thestations are within the BSS.
The interaction between analogcircuitry and Baseband processingis critical. A combination of testequipment may be used to measuresignals in both domains simultane-ously. An oscilloscope or logicanalyzer will allow complex trig-gering configurations to be defined.Figure 27 (pg. 21) shows an exam-ple of useful signal connections.
4.1Power vs. Time
IEEE 802.11a does not specify aPower vs. Time template, but this isan important measurement for allradio standards. The wide modula-tion bandwidth of WLAN signalsreduce the likelihood of spectrummask failures due to switching, evenwith sub-microsecond rise times.However, an 802.11a receiver has toadjust to the incoming signal usingonly 16µs of the burst. Problemshave arisen in other standardsbecause of having no definition ofwhat happens before the burst. Theplots in Figures 20 and 21 show twoeffects in 802.11b that could causeinteroperability problems.
In Figure 20, a short power burstappears before the main burst. InFigure 21, a clear step in RF leveloccurs at the beginning of the burst.Further analysis of this deviceshowed that the Power Amplifier
was turned on before data transmis-sion began.
HiperLAN/2 has more explicitrequirements for the shape of theburst. These are shown in Figure 22.
Figure 20: Example of unusualramp conditions inIEEE 802.11b devices
Figure 21: Example of unusualramp conditions inIEEE 802.11b devices
Figure 22: Power vs. Time mask for 802.11b and HiperLAN/2
18 Section Five: Transceiver Spurious Tests
4.2Spectrogram Testing
Spectrogram tests offer a fast wayto detect anomalies in complexsignals. Color or gray-scale isadded to the display to expressamplitude. Using time-capture ina signal analyzer, suspect signalscan be replayed more slowly tofully examine amplitude, frequency,and modulation transitions.
This application note does notoffer a full representation of thespectrogram signal. Please contactyour local Agilent representativefor a demonstration, or visit www.agilent.com/find/wlan for details.
4.3Transmitter-Receiver, Receiver-Transmitter Turnaround Time
Tx-Rx, Rx-Tx Turnaround Time testsare significant in the operation of aWLAN system, but they requiredetailed knowledge of the devicebeing tested as well as access tointernal test points. The equipmentconfiguration for the Clear ChannelAssessment test can be adapted tomake these tests.
5. TRANSCEIVER SPURIOUS TESTS
The use of high-speed digital cir-cuitry means that overall systememissions are often a combination ofanalog and digital effects. The tests,described only briefly here, are oftentime-consuming and require closeattention to measurement configura-tion. Control lines that are nominallydigital can easily become unintendedantennas when RF signals coupleinto them. Unexpected variations inresults often indicate that RF signalsare present on cables.
Transceiver measurements consistof performing out-of-band spuriousemissions tests. These confirm thatthe WLAN radio is operating withinregulatory limits.
Two types of emissions tests areperformed—conducted and radiated.Conducted emissions are a measureof the unwanted signals generated
by the DUT from its output connec-tor or from any cabling the deviceuses. Special signal coupling tech-niques are required for some meas-urements.
Radiated emissions are those ema-nating from the device and pickedup on external antennas. OfficialRFI testing often involves the use of an anechoic chamber to removebackground disturbances.
Separate standards are specifiedaccording to the region in whichthe equipment is to be used. TheUnited States follows Federal Com-munications Commission (FCC)standard, parts 15.205, 15.209,15.247, and 15.407. European coun-tries follow the European TechnicalStandards Institute (ETSI) ETS 300328 standard. In Japan, TELECdefines operating limits.
Spurious emission testing can beperformed using a spectrum ana-lyzer. Tests requiring compliancewith International Special Com-mittee on Radio Interference(CISPR) Publication 16 mayrequire electromagnetic compati-bility (EMC) spectrum analyzerswith quasi-peak detectors. Thesetests are not covered in this appli-cation note. Please contact yourlocal Agilent sales representativefor more information on Agilent'sEMC products.
Figure 23:Spectrogram showing change in use of carriers at the start of an IEEE 802.11a frame
Section Six: Receiver Measurements 19
6. RECEIVER MEASUREMENTS
Receiver design is often difficultbecause the designer has to allow formany different input signal condi-tions, some of which are difficult topredict. This is especially true whenoperating in an unlicensed band.This publication covers only themore common tests needed by appli-cation engineers; however, details ofmore sophisticated techniques canbe found in Appendices B and E.One such technique is to record alive RF signal, save it, and thenreplay it on demand, as shown inFigure 24.
Contact your local Agilent sales rep-resentative for more information onsuch specialized tests.
6.1Test Conditions and Setup
Two basic receiver test configura-tions are described below. The firstis a test of the analog circuit alone,while the second embraces the com-plete receiver.
Testing in IEEE 802.11a and 802.11bis generally done using a one-waysignal path, while HiperLAN/2designs may include options for sig-nal loop-backs. This allows external
test equipment to demodulate thereturned signal and do its own BERmeasurement.
A one-way signal path has the poten-tial for faster testing, because datadoes not have to be returned; howev-er, it places a greater burden on thedevice supplier and system integra-tor. Care is needed in the triggeringand sequencing of the measurement,since changes in level of the signal
source require time to settle beforefurther measurements are begun.
6.2Bit Error Rate
The WLAN standards do not directlyrefer to Bit Error Rate (BER) meas-urements. Unlike cellular (voice) sys-tems, WLAN transmissions do notnormally send unprotected bits. Ofcourse, Packet Errors are caused by
Figure 24: Signal record/replay of suspect signal using Agilent 89600Vector Signal Analyzer and ESG
Table 6: Summary of WLAN receiver performance tests
18.4.8.3 PN9(15) 802.11b: Test all other channels within the bandInterferer -35dBm, Wanted -70dBm, test as Min.
1000 frames Input Interferer is unsynchronized with wanted signal.
17.3.10.2 802.11a: See table 8 for levels.Interferer is unsynchronized with wanted signal.
17.3.10.3 802.11a PN9(15) As above. Test all other channels within the band, Non-adjacent on a 20MHz spacing from wanted signal.Channel rejection 1000 frames Interferer is unsynchronized with wanted signal.
18.4.8.4 802.11b: Multiple test conditions apply.
17.3.10.5 802.11a:1. With input signal -82dBm, >90% (within 4µs)
probability of Carrier Sense showing Signal Busy.2. Signal Busy shown for any signal above -62dBm.
Bit Errors, so the longer a packet is,the less likely it will be successfullyrecovered if the signal propagationpath is poor.
6.2.1Bit Errors and RF
Bit Errors are created when thesignal vector is not at the rightplace on the IQ plane when thereceiver reaches a decision point.There are many reasons why theconstellation becomes distorted,some of which are discussed in thetransmitter modulation tests insection 3.5. Figure 25 shows an IQconstellation for a 5% EVM signalwith Bit Errors occurring. Theactual BER depends on the type of distortion.
The correlation between modula-tion errors and bit errors becomesmore complex when multiple carri-ers (OFDM) are used. All WLANstandards discussed here use tech-niques which reduce the probabilityof bit errors caused by poor signal-to-noise ratios (energy per bit/noise,or Eb/No). In IEEE 802.11b, thisincludes spectrum spreading; in802.11a, forward error correctionis applied.
DSP algorithms, such as Viterbi,improve receiver performance byusing a short amount of data historyto predict what was most likely sent.The DSP then knows the differencebetween what it received as a rawbit and what it calculates as the cor-rect bit. This correction process alsoprovides signal quality informationas a by-product.
The result of these techniques is thatBit Error measurements are a verystrong function of RF signal level.Figure 26 shows how quickly BitError and Packet Error deterioratewith a reduction in Eb/No. The plotswere generated using AgilentAdvanced Design System software.
6.2.2 Bit Error vs. Packet Error
A Bit Error Rate measurement ispossible with a WLAN system andmay give an indication of perform-ance more quickly than PacketError Rate. The test configurationof Figure 27 (pg.21) is used. Atypical setup is with a PN 9 orPN15 sequence in a repeatingframe having a 1024-byte payload.Scrambling should be on. Vendor(proprietary) software would thenrecover the bit pattern and syn-chronize it to the PN sequence.
The Agilent ESG can run this BERtest itself, if decoded data, clock,and frame signals are availablefrom the Baseband circuit.
In practice, repeatable measure-ments can be made only if care istaken in setting the signal levelfor the test.
The performance of the modulatorin the test source may also affectthe PER reading. The usual practiceis to use a high-quality test sourceto remove this variable. Impair-ments may be deliberately added,if required.
6.3 Receiver EVM Measurements
Analog measurements of the outputof the receiver downconversionchain can provide much moreinformation than BER and PERabout the impairment suffered bya recovered signal. The same tech-niques described in transmittermodulation measurement in sec-tion 3.5 apply. Some options of theAgilent 89600 Vector Signal Ana-lyzer are specifically designed toaddress a situation where the RFsignal is downconverted to DC.
Figure 25: IQ constellation when bit errors occur
Figure 26: Plot showing variations on BER and PER vs. Input level, with 1024byte payload, 24Mbps
Section Six: Receiver Measurements 21
Figure 27 shows the configurationfor this test. Further details on thistype of measurement may beobtained by contacting your localAgilent sales representative.
6.4Frame Error Rate, Packet Error Rate
WLAN systems operate on a posi-tive acknowledgement-based tech-nique. When a frame is sent, itincorporates extra data to allowthe receiver to determine if any biterrors have occurred. This is therole of the Cyclic RedundancyCheck (CRC) and Frame CheckSequence (FCS). A receiver usesthe payload data it recovers to cal-culate a CRC, employing exactlythe same algorithm as the trans-mitting device. The two CRCs arethen compared. Any differencesbetween them mean that one ormore bit errors have occurred inthe payload data. If this is thecase, the ACKnowledge frame isnot sent. The CRCs are themselveserror-protected, so they will notsuffer bit errors unless the per-formance of the data content ofthe frame is very poor.
Frame Error Rate (802.11b) andPacket Error Rate (802.11a) use thesame measurement configuration.Figure 28 shows the necessary con-nections.
Actual test operation will depend onthe software used for the receivermeasurements.
The term for IEEE 802.11b testing isFrame Error Rate (FER). This meas-urement relies on the detection offailures in the Cyclic RedundancyChecks, which in 802.11b are usedin both header and payload.
FER is defined as:
In normal operation, an Acknowl-edge packet is sent from the receiv-ing station only if the CRCs arecorrect. It is possible that the sending station may not correctlyreceive the ACK signal itself. Forthis reason, the Retry field is setto 1 to mark re-transmissions andother changes with adjacent frames.
As described in section 1.5.3,Random Back-off intervals may beapplied in normal use if Acknowl-edgement packets are not deliv-ered. This is one of the functionswhich should be overridden by aproper test mode to avoid problemsin test sequencing.
Similarly, the effect of re-transmis-sion requests could cause problemsif not dealt with by the signal source.Again, available test modes shouldallow these functional requirementsto be disabled.
HiperLAN/2 and IEEE 802.11a alsouse a frame-based receiver perform-ance test called Packet Error Rate(PER). Unlike Frame Error Rate in802.11b, CRCs are not used, butthere is a frame check sequenceafter the user data, by which thereceiver determines if the data wascorrupted. If a frame is not detectedat all by the receiver, the FER/PERreading will be low. The test soft-ware used must have a way to deter-mine the number of frames sent.
Figure 27: Diagram of analog-only receiver measurement paths
Figure 28: Diagram of complete receiver measurement path
22 Section Six: Receiver Measurements
6.5Minimum Input Sensitivity, Maximum Input Level
This test is run according to thedescription and configuration inFigure 28. The test limits for thedifferent standards are shown inTable 7.
These tests verify that the receivercan deal with other WLAN signalswithin the same band. Figure 29shows the test configuration. The RFisolator prevents high-level signalsfrom one source from creating inter-modulation products in the other.If both generators have 40dB or
more of internal attenuation applied,the isolator may not be required.
IEEE 802.11 explicitly states thatthe interference source must not besynchronous with the wanted sig-nal. One reason for this is to testthe way a ZIF (Zero IntermediateFrequency) receiver deals with theeffect of an RF burst that appearsin the middle of a wanted signalframe. It is difficult to specify asingle timing/frequency relation-ship between the wanted and inter-fering signals. Choosing a shortidle period for the interferer is oneapproach. During design it is rec-ommended that the test configura-tion lock the reference frequenciesof the two signal sources and makeuse of external triggering for theinterferer. The trigger delay func-tion should be used to adjust frametiming. Stepping the delay throughthe frame period will highlight prob-lems within a design. Frequencyoffsets can also be entered for theinterference source within the oper-ational limits of the WLAN device.
The combination of these techniqueswill improve the reproducibility ofthe test.
6.7HiperLAN/2 Receiver BlockingPerformance
A blocking test is designed to checkthe performance of the receiverwhen signals originating outsidethe WLAN system are present. OnlyHiperLAN/2 includes such tests inthe specification. When a system isoperating in a license-exempt fre-quency band, it may be consideredappropriate to carry out some formof blocking test—at minimum, thismay identify issues with a particu-lar design.
The test configuration is similar tothat shown for the adjacent chan-nel rejection tests. The signalsource should be replaced withone having higher frequency cover-age if needed.
Note: Fc in Table 9 is the operating frequency of the test device. Testing in the 2.4GHz ISM bandwould have to be adapted to suit the operation of802.11b devices.
Frequency Range of Test LimitBlocking Signal (dBm)0.1 – >2500 MHz 0
2.5 – >4.5 GHz -104.5 – >5.15 GHz -30
5.15 GHz – >FC-50 MHz -30FC
+50 MHz – >5.35 GHz -305.35 – >5.47 GHz -30
5.47 GHz – FC-50 MHz -30FC
+50 MHz – >5.725 GHz -305.725 – >7 GHz -30
7 – >13 GHz -20
Section Seven: Power Supply Measurements 23
6.8Clear Channel Assessment, RSSI
Clear Channel Assessment detectiontimes are specified as < 4µs for802.11a and <25µs (5µs plus oneMAC slot length) for 802.11b. Accessto the carrier sense signal is needed.An oscilloscope triggered from thesignal generator can be used to per-form the test. Figure 30 indicates theappropriate connection points.
The Receive Signal Strength Indicator(RSSI) is measured during the pre-amble. The only performance stipula-tion is that it be monotonic. Theresult is reported only to the receiv-er’s MAC processor, not to the trans-mitter of the signal. RSSI is fre-quently found in end-user softwareto provide signal strength graphicsto help in system configuration.
Some receiver characteristics whichare important to the operation of aWLAN device, such as ClearChannel Assessment, are complexand difficult to measure without theappropriate software from the sup-plier of the device.
Additional receiver tests may berequired to address the EuropeanDynamic Frequency Selectionrequirements. A number ofRADAR systems operate in theupper 5GHz bands.
HiperLAN/2 also has an RSSImeasurement requirement. Thevalue, between 0 (-91dBm) and 62(> -20dBm) recorded by the receiv-er, is transmitted to the AccessPoint for system transmissionpower management.
7. POWER SUPPLY MEASUREMENTS
All equipment designs need to betested at extremes of supply voltage,even if specifications do not requireit. Operating limits will vary accord-ing to the conditions imposed by thehost device, whether a PC or a com-bination cell phone.
Other power supply measurementscan also be very informative. Theseinclude the current consumption asa function of the operational stateof the device. Receiver PowerManagement (RPM) is part of thespecification, because current con-sumption for listening is similar tothat for transmitting. Careful tim-ing of the receiver's active periodsis required. The longer oscillatorsand digital circuitry are turned off,the longer the battery life will be.
Monitoring power supply currentrelative to the timing of radiotransmission or reception helpsensure that firmware and hard-ware work together as expected.It is also straightforward to makebefore and after comparisons fol-lowing firmware updates to ensurethat no unwanted changes haveoccurred. Battery emulation allowsrepeatable testing of a DUT underrealistic conditions.
Agilent offers a complete line of DCpower supplies for these tests. Theseinclude general-purpose instrumentsas well as instruments specificallytailored to the demands of mobilecommunication. These DC voltagesupplies also offer low-current meas-uring capability, which is useful forevaluating battery consumption dur-ing standby operation.
Figure 30: Test configuration for clear channel assessment tests
24 APPENDIX A
APPENDIX A: Agilent Solutions for Wireless LAN
Agilent Equipment for Wireless LAN PHY Layer (RF) Testing
� Full measurement capability� Some measurement limitations
IEEE 802.11 IEEE 89600 SERIES PSA, ESA SERIES ESG-C SERIES EPM-P SERIESRF LAYER TESTS REFERENCE VECTOR SIGNAL SPECTRUM SIGNAL POWER METERS
ANALYZERS ANALYZERS GENERATORS
Transmitter Tests
Output Power 18.4.7.1.2� �1
� 217.3.9.1
Power Rise/Fall 18.4.7.6 �
Spectrum Mask 18.4.7.3� �
17.3.9.2
Carrier Suppression 18.4.7.7 � �
Center Frequency Leakage 17.3.9.6.1 �
Spectral Flatness 17.3.9.6.2 � �
Transmission Spurious 18.4.6.8 � �
Center Frequency Tolerance 17.3.9.4� �
18.4.7.4.5
Symbol Clock Frequency 17.3.9.5 � �
Tolerance
Constellation Error 17.3.9.6.3 �
Error Vector Magnitude 18.4.7.8 �
Transceiver Tests
Out-of-band Spurious Emission 17.3.8.4�
18.4.6.9
Receiver Tests
Sensitivity 18.4.8.1�3
17.3.10.1
Max Input Level 18.4.8.2�3
17.3.10.4
Adjacent Channel 18.4.8.3�3
Rejection 17.3.10.2
Non-adjacent Channel 17.3.10.3 �3
Rejection
Clear Channel Assessment 18.4.8.4�3
17.3.10.5
1 Channel power measurement indicates Average transmit power2 Thermal sensor gives true rms power reading. Peak detector under-reads peak-average result when OFDM/modulation is applied.3 Use second source as interferer. CW interference (for blocking tests) can be generated using Agilent E8241A Microwave Signal Generator.
APPENDIX A 25
Test Equipment with WLAN Capability
1. Vector Signal Analyzers, 89600 Series
Versatile and precise signal analysis,with 36MHz capture bandwidth for802.11a turbo modes. In-depthanalysis of IEEE 802.11 transmitterand receiver chains. Automaticdetection and demodulation of802.11a formats. Provides modula-tion quality analysis for IEEE802.11a OFDM signals, includingEVM versus time & EVM versussub-carrier.
Recommendation for 802.11a:
89641A, with dc-6GHz tuner
Opt. AYA/B7R: Vector SignalAnalysis and OFDM demodulation
Opt. 105: Dynamic links toEESof/ADS
2. Signal Generators, ESG-D Series (3-4GHz), E4438C with Signal Studio
Generate IEEE 802.11 signals fortransmitter and component tests,send formatted packets for receiverPER testing, generate Baseband signals for direct input to MAC orAnalog circuits.
Recommendation for 802.11a,b:
E4438C, with Signal Studio
Opt. 410: IEEE 802.11a
Opt. 405: IEEE 802.11b
Opt. 506: 6GHz operation
Opt. UNJ: Enhanced Phase noise
Opt. 002: Internal BasebandGenerator
Opt. 005: 6Gbyte hard-drive
Opt. UN7: Internal BER Tester(measurement capability dependenton vendor test configuration)
3. Spectrum Analyzer, PSA (6.7-50GHz)and ESA-A Series
Automated "one button" test execu-tion for Swept Spectrum transmittermeasurement. Performs a broadrange of spectrum measurements.
Recommendation:
E4440A PSA, 26.5GHz
Opt H70: 70MHz IF for use with89611A VSA
4. EPM-P Power Meter, 8482A ThermalSensor, E9327 Peak Power Sensor
Make accurate average power meas-urements with thermal sensor.Measure and inspect framed signalswith the Peak Power Sensors.
5. Simulation Software, ADS with E8874AWLAN Design Guides
Essential software tool for designand simulation of custom WLANsystems. Pre-defined WLAN compo-nent models speed the simulationprocess. Can be linked with the ESG-D and 89600 Series.
Other Test Equipment
1. RF Shielded Enclosure
Allows repeatable RF measurementsto be made, without interferencefrom external environment.
2. DC Sources, 66319,66321 B/D
Fast, programmable dynamic DC power sources with batteryemulation.
3. Logic Analyzers –1680/1690 Series
Comprehensive system-level debug-ging for digital hardware design andverification
4. Logic Analyzers –16700 Series
Provides comprehensive system-leveldebugging for multiple processor/bus designs. Use E5904B withEmulation Trace Macrocell port forARM processor triggering.
5. Mixed Signal Oscilloscopes – 54600 Series
Use for verification and debugging ofIEEE 802.11 baseband signals.
6. Network Analyzers – 8753E Series
Provides measurement of AntennaVSWR and performance of PA, LNAand RF switch.
7. Function Generator, 33250A, 80MHz Function/Arbitrary Waveforms
Generate clock signals and noise, or combine with an oscilloscope toreproduce baseband waveforms.
Accessories
1. Oscilloscope Probe–54006A
Passive probes with very low capaci-tance (0.25pF).
2. Close Field Probe–11940A
Measures magnetic field radiationup to 1GHz.
3. Splitter –11667A
Use for ratio measurements andequal power splitting.
4. Directional coupler–773D
Use for monitoring one RF wave-form (2-18GHz) while two IEEE802.11 devices are connected bycables.
5. Dual directional coupler –772D
Useful for monitoring both RF wave-forms (2-18GHz) while two IEEE802.11 devices are connected bycables.
1. The Design and Verification ofIEEE 802.11a 5GHz Wireless LANSystems, Agilent article, web onlyhttp://www.chipcenter.com/networking/technote019.html
Product Overviews:
1. Bluetooth & Wireless LAN TestProducts, Systems and Services,lit. number 5988-4438EN
2. Agilent EPM-P Series Single-and Dual-Channel Power MetersDemo Guide, lit. no. 5988-1605EN
Product Notes:
1. Agilent 89600 Series WideBandwidth Vector SignalAnalyzers,lit. no. 5980-0723E
2. Agilent 89640A dc to 2700MHzVector Signal Analyzer TechnicalSpecifications,lit. no. 5980-1258E
3. Using Vector ModulationAnalysis in the Integration,Troubleshooting and Design OfDigital RF CommunicationsSystems, Product Note 89400-8.,
lit. no.5091-8687E
4. Ten Steps to a Perfect DigitalDemodulation Measurement,Product Note 89400-14A, lit. no. 5966-0444E
5. 802.11a WLAN Signal StudioSoftware, Option 410 for theE4438C ESG Signal Generator,lit. no. 5988-5765EN.
6. 802.11b WLAN Signal StudioSoftware, Option 405 for theE4438C ESG Signal Generator,lit. no. 5988-5766EN.
7. Customizing Digital Modulationwith the Agilent ESG-D Series Real-Time I/Q Baseband Generator, Option UN8,lit. no. 5966-4096E
8. Generating and DownloadingData to the Agilent ESG-D RFSignal Generator for DigitalModulation, lit. no. 5966-1010E
9, Generating Digital Modulationwith the Agilent ESG-D SeriesDual Arbitrary WaveformGenerator, Option UND,lit. no. 5966-4097E
APPENDIX C, D, E 27
APPENDIX C: Glossary
Acknowledgement: The short frame sent by a receiverwhen it is able to correctly decodea frame.
Beacon: A regular RF transmission usedby stations to discover if thereare other devices within its oper-ating range. Beacons are alsoused for coarse timing alignment.
Carrier Sense MultipleAccess/Collision Avoidance: The term used to describe the way802.11 devices listen for other RFsignals before transmitting.
Hidden Node: A station that is not "heard"directly when another stationlistens as part of the CSMA/CAprocess.
Medium Access Control: The function of the software thatadapts wired LAN transmissions,making them suitable for sendingover an RF link.
Network Allocation Vector: A timing variable used to let eachstation know how long it has towait before transmitting.
Station: The Access Point or NetworkInterface card that interfacesbetween the host device andthe RF link
APPENDIX D: Symbols and Acronyms
ACK AcknowledgementADS Advanced Design SystemAP Access PointBER Bit Error RateBPS Bits Per SecondBPSK Binary Phase Shift KeyingBRAN Broadband Radio Access NetworkBSS (Infrastructure) Basic Service SetCCA Clear Channel AssessmentCCDF Complementary Cumulative
IBSS Independent Basic Service SetIF Intermediate FrequencyISM Industrial, Scientific, and MedicalLLC Logical Link ControlLO Local OscillatorMAC Medium Access ControlMT Mobile TerminalNIC Network Interface CardOFDM Orthogonal Frequency
PLL Phase Locked LoopPMD Physical Medium DependentPN9,15 Pseudo Random NumberPSDU PLCP Service Data UnitQAM Quadrature Amplitude ModulationRBW Resolution BandwidthRSS0 Receive Signal Strength 0 (Zero)RSSI Receive Signal Strength IndicationRTS Ready To SendRX ReceiverSAW Surface Acoustic Wave (Filter)SIFS Short Interframe SpaceSTA Station (in HiperLAN/2)TDD Time Division DuplexTPC Transmit Power ControlTX TransmitterUNII Unlicensed National Information
Infrastructure
VBW Video BandwidthVCO Voltage Control InterfaceVSA Vector Spectrum AnalyzerVSWR Voltage Standing Wave RatioWECA Wireless Ethernet
Compatibility Alliance
WLAN Wireless Local Area Network
APPENDIX E:References
1. Supplement to IEEE Standard for Information Technology, IEEE Std 802.11a–1999 (supple-ment to IEEE Std 802.11-1999)
2. Higher Speed Physical Layer in the 2.4GHz Band, IEEE Std802.11b/D8.0,Sept 2001 (draft sup-plement to IEEE Std 802.11-1999)
3. Broadband Radio AccessNetworks (BRAN); HiperLANType 2; Physical Layer ETSI TS101 475 V1.1.1 (2000-04)
Agilent Technologies’ Test and Measurement Support, Services, and Assistance
Agilent Technologies aims to maximize the value you receive, while minimizing your risk and problems.We strive to ensure that you get the test and measurement capabilities you paid for and obtain the support you need. Our extensive support resources and services can help you choose the right Agilentproducts for your applications and apply them successfully. Every instrument and system we sell hasa global warranty. Support is available for at least five years beyond the production life of the product.Two concepts underlie Agilent's overall support policy: "Our Promise" and "Your Advantage."
Our Promise
Our Promise means your Agilent test and measurement equipment will meet its advertised performanceand functionality. When you are choosing new equipment, we will help you with product information,including realistic performance specifications and practical recommendations from experienced testengineers. When you use Agilent equipment, we can verify that it works properly, help with productoperation, and provide basic measurement assistance for the use of specified capabilities, at no extracost upon request. Many self-help tools are available.
Your Advantage
Your Advantage means that Agilent offers a wide range of additional expert test and measurementservices, which you can purchase according to your unique technical and business needs. Solve problems efficiently and gain a competitive edge by contracting with us for calibration, extra-costupgrades, out-of-warranty repairs, and on-site education and training, as well as design, system integra-tion, project management, and other professional engineering services. Experienced Agilent engineersand technicians worldwide can help you maximize your productivity, optimize the return on invest-ment of your Agilent instruments and systems, and obtain dependable measurement accuracy for the life of those products.
By internet, phone, or fax, get assistance with all your test & measurement needs.
