Measurement 45 (2012) 1001–1014

Contents lists available at SciVerse ScienceDirect

Measurement

journal homepage: www.elsevier .com/ locate/measurement

Measurement and functional validation of detect and avoid ultrawideband devices

Gianmarco Baldini a,⇑, Detlef Fuehrer a, Janie Banos b, Manuel García Fuertes b,Xiaochen Chen c, Di Wu c

a European Commission Joint Research Centre, IPSC, Ispra, Italyb AT4 Wireless, SA, Málaga, Spainc Telecommunication Metrology Center of MIIT, Beijing, China

a r t i c l e i n f o

Article history:Received 14 February 2011Received in revised form 21 July 2011Accepted 31 January 2012Available online 11 February 2012

Keywords:Ultra Wideband (UWB)Detect and Avoid (DAA)MeasurementsTestingWireless communicationsRadio frequency spectrumTestingElectro-magnetic radiated measurementsElectro-magnetic conducted measurements

0263-2241/$ - see front matter � 2012 Elsevier Ltddoi:10.1016/j.measurement.2012.01.043

⇑ Corresponding author.E-mail address: gianmarco.baldini@jrc.ec.europa

a b s t r a c t

Ultra Wideband (UWB) radio regulations have been discussed in recent years because ofthe potential interference of UWB radios to other radio systems with which they sharethe spectrum. A mitigation technique for UWB devices named Detect and Avoid (DAA)has been proposed as a potential solution to reduce the risk of interference. DAA testingpresents specific challenges related to the low emission power of UWB devices, the valida-tion of spectrum sensing algorithms and timing considerations. This paper presents thedesign of a test bed to validate the correct implementation of DAA mitigation techniquesby UWB devices. It describes the testing challenges and the related test bed design solu-tions for both radiated and conducted test methods. A number of measurement campaignson UWB devices with detect-only capabilities were separately conducted at three differenttest beds facilities. The measurements results are analyzed and compared.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Ultra Wideband (UWB) is a very promising technologyfor the broadband transmission of data using spectrum-efficient and flexible radio techniques. UWB is a rather va-gue term to describe the nature of a Radio Frequency (RF)signal, which occupies a large instantaneous bandwidth.Definitions of the minimum bandwidth of an UWB signalvary; while the European Telecommunication StandardsInstitute (ETSI) defines in [1] a minimum bandwidth of50 MHz, the Federal Communications Commission (FCC)specifies a minimum instantaneous bandwidth of500 MHz [2] in a frequency range from 3.1 to 10.6 GHz.Because of this large frequency span, UWB devices canoperate in the same RF spectrum as many other wireless

. All rights reserved.

.eu (G. Baldini).

communication technologies including WiMAX™, Wi–Fi,satellite communications and others (see [3]). Fig. 1 pro-vides an overview of the primary wireless services orincumbents operating in the same frequency range ofUWB, with the allocated frequency bands, bandwidth,range and data throughput.

To address the risk of interference from UWB to othersystems and services, regulatory bodies around the worldhave defined stringent limits for the emission power ofUWB devices. In most cases, the limit is given as an Equiv-alent Isotropically Radiated Power (EIRP) emission mask.EIRP emission mask was defined by the FCC in 2002, theEuropean Union in 2006, China in 2008, Japan in 2006and Korea in 2006. The disadvantage of the EIRP maskis that UWB transmission power is limited even in theabsence of WiFi or WiMAX communication.

A more flexible approach is to allow higher emissionpower for UWB devices when no other wireless system is

Fig. 1. Primary wireless services in the UWB frequency bands.

1002 G. Baldini et al. / Measurement 45 (2012) 1001–1014

transmitting within the same coverage area [4]. In this casean opportunistic approach could be used, where secondaryusers (e.g., UWB devices) are required to detect the trans-mission of primary users in specific spectrum bands andconsequently refrain from transmitting in those bands orreduce their emission power. In the case of UWB, this ap-proach is also named Detect and Avoid (DAA). Radio regu-lators have proposed the adoption of DAA in certainregulatory domains and for specific radio frequency bands.An important task for the deployment of UWB DAA en-abled devices is to verify that the DAA mechanism isimplemented according to the specifications defined bythe spectrum regulators.

In this paper, we present a test bed designed to evaluateUWB DAA devices operating co-channel with WiMAX. Inthis paper, we limit our scope to WiMAX operating at3.5 GHz because it is the frequency band with higher riskof interference from UWB devices. We also limit the scopeto Multi Band Orthogonal Frequency Division Modulation(MB-OFDM) UWB techniques because they are used inthe PHY layer specification of the WiMedia Alliance, whichis the largest industry association to promote the adoptionof UWB technology worldwide. The definition of the testbed and related measurement campaigns were the mainfocus of the FP7 project WALTER (Wireless Alliance forTesting Experiment and Research). Project WALTER (see[5]) was a project co-financed under the 7th framework

program of the European Commission, whose main objec-tive was to develop a networked test bed to validate UWBinterference mitigation techniques and coexistence mech-anisms including DAA. The networked test bed was basedon three testing facilities based in China (TMC-MIIT), Spain(AT4 wireless) and Italy (Joint Research Project of the Euro-pean Commission).

The paper is structured in the following sections: Sec-tion 2 provides a global overview of the regulatory andstandardization requirements for UWB DAA enabled de-vices. It also provides a description of the MB-OFDM UWBDAA system. Section 3 describes the methodology to spec-ify requirements, define test procedures and identify themain validation and testing challenges. Section 4 describesthe design and implementation of the test beds (conductedand radiated) and it provides an analysis of the main factorscontributing to the quality of the measurements. Section 5describes the measurement campaigns and the results.Finally Section 6 concludes the paper.

2. DAA UWB standards and regulations

2.1. DAA UWB standards – ECMA-368

ECMA-368 standard [6] and ETSI EN 302 065 [1] definethe physical and MAC layer of the MB-OFDM UWB

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1003

implementation. ETSI TS 102 754 ([7] includes the Detect-and-Avoid specifications (DAA) for UWB in 3.1–4.8 GHzand 8.5–9 GHz for the protection of Broadband WirelessAccess (BWA) and radars.

At the PHY layer, ECMA-368 standard [6] divides thespectrum range 3100–10,600 MHz into 14 bands, eachwith a bandwidth of 528 MHz (see Fig. 2). The first 12bands are then grouped into four band groups consistingof three bands. The last two bands are grouped into a fifthband group. A sixth band group is also defined within thespectrum of the first four, consistent with usage withinworldwide regulatory regulations. Depending on the localregulations, some of these bands can use DAA mitigationtechniques to reduce the risk of interference withincumbents.

A MB-OFDM scheme is used to transmit information. Atotal of 110 sub-carriers (100 data carriers and 10 guardcarriers) are used per band. In addition, 12 pilot subcarriersallow for coherent detection. Frequency-domain spread-ing, time-domain spreading, and Forward Error Correction(FEC) coding are provided for optimum performance undera variety of channel conditions.

2.2. Regulations for DAA UWB technology

In USA, FCC [8] has opened the 3.1–10.6 GHz frequencyband for the operation of UWB devices provided that theEIRP power spectral density of the emission is lower thanor equal to �41.3 dBm. FCC regulations do not specify theuse of mitigation techniques for UWB devices operatingin the mentioned frequency range.

In China mainland, in the 4.2–4.8 GHz band, the maxi-mum EIRP is restricted to �41.3 dBm by the date of 31stDecember, 2010. After that, the UWB devices adopt anInterference Relief Technology, such as DAA. There are nospecific parameters or limit values for DAA in the currentChinese UWB regulation specification [9].

In Japan, in the 3.4–4.8 GHz frequency range, UWB de-vices without interference avoidance techniques such asDAA may not transmit at a level higher than �70 dBm. Inthe 3.4–4.2 GHz band, UWB devices may transmit at or be-low the limit of �41.3 dBm, under the condition that theyare equipped with interference avoidance techniques such

Fig. 2. UWB Bands in th

as DAA. In the 4.2–4.8 GHz band, UWB devices adopt aninterference avoidance technique after 31st December,2010.

In Korea, the UWB emission limit mask requires theimplementation of an interference avoidance techniquesuch as DAA in the 3.1–4.2 GHz and 4.2–4.8 GHz bands toprovide protection for IMT Advanced systems and broad-casting services. The requirements in the 4.2–4.8 GHz bandare implemented after 31st December, 2010.

In Europe, the regulation for generic UWB devices (i.e.,not specifically DAA enabled) is composed of two ECC Deci-sions: the baseline Decision ECC/DEC/(06)04 [10], whichdefines the European spectrum mask for generic UWB de-vices without the requirement for additional mitigationand Decision ECC/DEC/(06)12 [11], recently amended by[12], which provides supplementary provisions such asLow Duty Cycle (LDC) or DAA. These regulations are de-scribed in detail in the next section.

2.3. Technical requirements for UWB DAA technology

DAA is based on the concept of defining coexistencezones, which correspond to a minimum isolation distancebetween an UWB device and the victim system. For eachDAA zone, in conjunction with the given minimum isola-tion distance, the detection threshold and the associatedmaximum UWB transmission level are defined based onthe protection zone the UWB device is operating within.

Three zones are defined on the basis of the detected up-link power of the victim signal:

1. Zone 1 with a detection threshold for the uplink victimsignal of �38 dBm. In this zone, the UWB device isrequired to reduce its emission level in the victim bandsto a maximum of �80 dBm. As an alternative, the UWBdevice is allowed to move to a non-interfering channel.

2. Zone 2 with an uplink detection threshold of �61 dBm.In this zone, the UWB device is required to reduce itsemission level to a maximum of �65 dBm. As an alter-native, the UWB device is allowed to move to a non-interfering channel.

3. Zone 3 where the UWB device does not detect any vic-tim signal transmitting with a power greater than

e Ecma standard.

1004 G. Baldini et al. / Measurement 45 (2012) 1001–1014

�61 dBm. In this case, the UWB device is allowed tocontinue transmitting at maximum emission level of�41.3 dBm.

Ref. [3] provides flowcharts for the implementation ofthe DAA algorithm as depicted in Fig. 3.

The flowcharts and detection algorithms are imple-mented on the basis of the following parameters:

� Minimum initial channel availability check time. It is theminimum time the UWB device spends searching forvictim signals after power-on.� Signal detection threshold. It is the detected incumbent

power threshold (uplink signal) that triggers UWBdevice to transition between protection zones.� Avoidance level. It is the maximum UWB transmit power

in a given protection zone.� Minimum avoidance bandwidth. It is the minimum

portion of the victim service bandwidth requiringprotection.� Maximum detect and avoid time. It is the maximum time

duration between a change of the external RF environ-mental conditions and adaptation of the correspondingUWB operational parameters.� Detection probability. It is the probability for the DAA

enabled UWB device to make a correct decision todetect the presence of a victim signal before startingtransmission.

The technical requirements for the application of DAAin Europe in bands 3.4–4.2 GHz are described in ECC Re-port 120 [3] and they are presented in Table 1.

The detection functionality in the DUTs validated dur-ing the test was implemented through energy detection:

Fig. 3. DAA UWB flow chart in the

the FFT output was used to compute the power spectraldensity of the victim signal. A discussion on the various op-tions for detect and avoid implementation is presented in[13]. A performance comparison of the various detectiontecniques for DAA UWB beyond simple energy detectionis also provided in [14].

In the rest of the paper, all mean power levels are mea-sured within a 1 MHz bandwidth.

3. Development of testing requirements and procedures

3.1. Methodology

The WALTER project brought together stakeholders andtechnical experts to develop requirements and test proce-dures. The process to develop the requirements (shownin Fig. 4) is composed of four phases: collection, analysis,classification and specification.

The objective of the collection phase is to collect all thenecessary information from various inputs in order tounderstand the needs to be covered by the technologyand define requirements. Inputs include radio frequencyregulatory decisions (e.g., EIRP emission masks) atinternational, european and national level, standards(e.g., ETSI, Ecma) and government, market and researchtrends, which provide insight on the evolution of theUWB technology. The latter category includes informa-tional (i.e., not regulatory decisions) documents, whichcan be used to prioritize the requirements. For example,a government ‘‘white paper’’ on future trends for innova-tive wireless technologies can indicate the urgency of aspecific need and related requirement. For the Europeancontext, the regulations from conference europeenne des

frequency band 3.4–4.2 GHz.

Table 1Technical requirements for DAA UWB in Europe in the frequency range 3.4–4.2 GHz.

Parameter Zone 1 Zone 2 Zone 3

Minimum initial channel availability check time 5.1 s 5.1 s 5.1 sMaximum detect and avoid time (VoIP) 2 s 2 s 2 sDetection probability (initial phase) 99% 99% 99%Detection probability (continous mode) 95% 95% 95%Avoidance level �80 dBm �65 dBm �41.3 dBmMinimum avoidance bandwidth 200 MHz 200 MHz 200 MHz

Fig. 4. Methodology for requirements collection and definition.

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1005

administrations des postes et des telecommunications(CEPT) have been used. The standards from Ecma andEuropean Telecommunications Standards Institute (ETSI)are the main inputs.

In the analysis phase, the requirements are comparedagainst operational scenarios or usage cases for UWB tech-nology like Personal Area Network for media entertain-ment at home or use of UWB devices in planes. Thisphase is useful to tailor the input defined in the collectionphase to realistic uses of UWB technology.

In the classification phase, a general taxonomy for theinputs is specified to establish a coherent classificationof requirements. In the WALTER project, six require-ments areas are defined: Regulatory, which includes in-put from applicable regulatory domains (EU, USA, EastAsia), Radiated Performance Testing, which includes theradiated performance of the device, Over the Air forcontrolling the device under test, Conformance toapplicable standards and regulations, Interoperability todemonstrate the ability of two systems to interoperateusing the same communication protocol and Coexistenceto demonstrate the coexistence of UWB implementa-tions with other wireless technologies. The test proce-dures and measurement campaigns described in this

paper belongs to the Regulatory category and the DAAsubcategory.

Finally, in the specification phase, the inputs analyzedand classified are used to define requirements for testingand validation of UWB devices.

The requirements are used to drive the definition of thetest procedures, which have the objective to validate theUWB devices for specific requirements. Requirements arealso used to define the test bed specifications, which in-clude the identification of test and measurement elementsand related test bed configurations.

An example of test procedure specification is ‘‘The testprocedure verifies the radio location detection and avoid-ance capability for the selected UWB operational frequencyin normal UWB operation using an increasing radio loca-tion test signal level’’. In this test the detect and avoid timeand the corresponding avoidance operation will be verifiedagainst the technical requirements defined in regulationsand standards. Fail criteria is a measured time, which islonger than required (e.g., 2 s in Zone 1).

Test procedures are then executed in measurementcampaigns at three test bed sites in Spain (AT4 wireless),China (Telecommunication Metrology Center) and Italy(Joint Research Centre of the European Commission).

1006 G. Baldini et al. / Measurement 45 (2012) 1001–1014

The methodology is also based on the concepts ofTraceability and Change Management. A RequirementTraceability Matrix is used to trace the requirements withthe test procedures and test specifications to guaranteecomplete coverage and coherence of the test activityacross the project. Change Management procedures andtools are used to support changes in the requirementsdue to modifications of the regulations or standards acrossall the deliverables. CM will be based on a tailored versionof the change request process defined in CMMI-DEV[15].

3.2. Challenges for testing and verification

During the process of test procedures definition, weidentified specific challenges for the testing and verifica-tion of the DAA mechanism implemented in UWBdevices:

1. Low power level: Due to the low power levels to bedetected, as well as the power levels to be measured(i.e., �80 dBm in Zone 1), a dynamic margin problemcan arise, as these levels are close to the noise floor limitof most conventional spectrum analyzers (i.e.,�70 dBm). This constraint is much worse in radiatedtest environments. Furthermore, the low power UWBsignal must be detected and discriminated in the pres-ence of the victim signal. As a consequence, test equip-ment must have high sensitivity to precisely detect theuplink victim signal as defined by regulators. An impor-tant parameter is the Noise Figure (NF), which is theincrease of noise power from the front end input noiseto the displayed noise floor of the measurement equip-ment that is greater than the signal gain.

2. Timing: Manual test methods make timing-related mea-surements inaccurate. As an example, the analyzers arerequired to start capturing the signal at the sameinstant the device is powered on, which is a hard taskwhen using manual test methods. Automatic methodscan use the UWB DAA power-on signal to trigger theanalyzers to start capturing the signal almost at thesame time. Although there might be a delay betweenthe capture start time and the real power-on time, itcan be considered as fixed compared with the manualmethod. This delay can be later compensated in thefinal timing calculations.

3. Detection probability: The criteria for detection probabil-ity is defined as 95% or 99% in ETSI TS 102 754 ([7]).Generally it takes considerable time to measure thedetection probability. It is important to define alternatemethods, which still validate the testing requirementbut reduces the testing time.

4. Radiated tests: The measurement of the power levels ofthe Device Under Test (DUT) in radiated test environ-ments is particularly complex since the positions ofthe test and interfering antennas relative to the DUTposition are different and because of the low power lev-els as in point (1), which requires short distancesbetween DUT and measurement antenna. The shortmeasurement distance may present the risk of

conducting near-field measurements, while far-fieldmeasurements are mandatory.

4. Design and implementation of the test bed

4.1. General setup

As described in the introduction, both a radiated and aconducted test bed were set up to execute the test proce-dures. This section describes the common elements andconfigurations used for both radiated and conducted testbeds.

In both test beds, a signal generator generates a WiMAXsignal, which coexists with a UWB link created by a UWBDAA enabled device or Device Under Test (DUT) and a gen-eric UWB device. The signal generator used in the test bedsetup is a Rohde&Schwarz (R&S) SMBV100A Vector SignalGenerator with a frequency range from 9 kHz to 6 GHzand low single-sideband (SSB) phase noise (<�122 dBc at1 GHz). R&S SMBV100A is capable of generating a widerange of signals including HSPA+, LTE, WLAN and WiMAX.All the three test beds of the WALTER project have thesame configuration but the second test bed has also useda real WiMAX base station and Customer Premise Equip-ment (CPE) for the DAA test.

The signals in the frequency domain are amplified andcollected by a spectrum analyzer. The spectrum analyzerused in the test beds is a Tektronix RSA3408A real-timespectrum analyzer. It has a frequency range from DC to8 GHz, 36 MHz bandwidth and a very low backgroundnoise floor (�150 dBm/Hz at 2 GHz and a frequency range).

The spectrum analyzer was configured with settings ofResolution BandWidth (RBW) of 1 MHz, Video BandWidthof 1 MHz and Detector Mode set to Root Mean Square(RMS). These settings are recommended in [7] as the besttrade-off for UWB to have a good value of sensitivity andhave a reasonable sweep time for the large spectrum bandof UWB signals. If the RBW is increased, the sweep time isreduced but the sensitivity is also decreased. If the RBW isdecreased, the sweep time may not be short enough tocomplete the test in the timing defined in the standards.

All test bed equipments are connected and controlledby a PC unit running LabVIEW to automate the test execu-tion. Various link distances between the DUT, the WiMAXsignal generator and the generic UWB device are simulatedwith a variable 0–81 dB variable attenuator.

The DUT will be able to maintain a stable link with apeer UWB device in all test scenarios. The performance sta-tistics from both devices are used to verify the stability ofthe link. This method is only valid if the interferer and vic-tim emissions can be differentiated in the spectrum ana-lyzer. The RF signal emitted from the DUT is capturedand recorded using the RSA3408 spectrum analyzer de-scribed before. The victim RF signal is used to activatethe DAA mechanism and evaluate the reaction of theDUT. In order to create a realistic scenario, different typesof traffic are generated, such as data (TCP/UDP), IPTV andVoIP, using either PC-based software traffic generatorssuch as Iperf, IxChariot or WatchIPTV, or a dedicated net-work tester. The spectrum analyzer will be able to detect

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1007

the DUT level and the victim level in all the required dy-namic range. This method is only valid if the interfererand victim emissions can be differentiated in the spectrumanalyzer. The emission power of the WiMAX signal is set to�38 dBm and �61 dBm, as defined in Section 2, for each ofthe three zones. For each WiMAX power level, the behaviorand emission power of the DUT are measured to verify thatrequirements are met. The measurements are collectedand analyzed offline.

To accurately measure the different time periods forDAA phases, signal generators and spectrum analyzersare connected to a timing reference (e.g., rubidium oscilla-tor), which is used to improve the time synchronizationamong the test bed elements.

To ensure that the appropriate power levels are cor-rectly detected, all the equipment used in the test bed iscalibrated. As a first step, the test bed is validated againsta golden reference unit to record the response in amplitudeand frequency. The response is then used during the mea-surement phase to correct the collected data. Because themeasurements were conducted in three separate test labo-ratories, it was important to validate the test sites against acommon reference. The test site validation was achievedby verifying that, in a shielded enclosure, the coupling be-tween a Comparison Noise Emitter (CNE) mounted on anon-conducting bench and a calibrated broadband mea-suring antenna set 3 m from it, is within certain tolerancesof the coupling which would be obtained if the same setupwas situated in a reference site over the frequency rangefrom 1 GHz to 7 GHz. The recommended tolerances are±4 dB. Calibration was done both for vertical and horizon-tal polarization.

In a similar way, in the radiated test, the radiation pat-tern and frequency response of the antennas were calcu-lated and used in the measurements phase. Thecorrections are stored in the spectrum analyzer as trans-ducer factors which include measurement antennae gain,propagation and cable loss and pre-amplifier gains.

The DUT used in the test did not have Avoid capability.The Avoid capability was simulated by controlling the testmode of the UWB device, to move the communication fre-quency to band 2, mode TFC 6 of WiMedia MB-OFDMAstandard to avoid interference to the WiMAX victim signal.The spectrum occupancy is shown in Fig. 5. The Error Vec-tor Magnitude (EVM) and Bit Error Rate (BER) of the Wi-MAX signal was measured. A BER of 10�6 after forwarderror correction coding is the threshold of minimum Wi-Max sensitivity. As expected, the calculated BER was below10�6 to prove that the DAA device transmitting in band 2/TFC 6 does not have an impact in term to the WiMAXsignal.

4.2. Conducted test bed

Conducted tests have the advantage of higher immunityto interference from other sources. To verify the compli-ance of a DAA enabled UWB device with a given emissionmask, the conducted approach is the most convenientone, provided that a conducted and impedance-matchedconnection between the test equipment and the DUT canbe established. Fig. 6 shows the schema of the DAA test

bed setup and Fig. 7 its implementation at the second testbed site.

In this configuration, the WiMAX signal generator ismodelling the victim device while the DUT represents theinterferer. The UWB device is operated in test mode or itis operating in regular mode and communicating withthe DUT. The UWB device (s) operate (s) in UWB bandgroup (BG) 1 (3.1–4.8 GHz), either in Time Frequency Code(TFC) 1 (band hopping) or in TFC 5 (static operation in the3.1–3.6 GHz range).

The low noise amplifier is needed to improve the noisefigure of the measurement system. Noise figure is the in-crease of noise power from the front end input noise tothe displayed noise floor of the measurement equipment(i.e., the spectrum analyzer) that is greater than the signalgain. It can also be described as the reduction of signal tonoise ratio of the measurement system.

The noise figure of a measurement system can be im-proved by the addition of a low noise pre-amplifier. Thehigher the gain of the pre-amplifier, the more the noise fig-ure of the measurement system is overcome.

The following equation is used to calculate the cascadednoise figure of the combined pre-amplifier and the mea-surement system. The units for the calculation are in linearterms and not in dBs.

For a two stage system:

Fn ¼FLNA þ ðFSA � 1Þ

GLNA

where FLNA is the noise figure of the amplifier, FSA is thenoise figure of the spectrum analyzer and GLNA is the gainof the amplifier. Low noise figure and amplifier gain is awell known trade-off in microwave amplifier design. Thehigher the gain, the more difficult is to obtain a low F. An-other aspect of the formula is that low gain amplifiersshould not be used, as the lower the gain the less effectit has on the overall improvement of the F of the measure-ment system. In the test bed, we used a low noise amplifierwith a gain of 39 dB and a noise figure of 1.2 at 3.5 GHz.

A number of test procedures have been defined to vali-date the DAA regulatory zone schema described in Section2. For example, the procedure for DAA detection of the Wi-MAX victim signal, when the DUT is set to ‘‘in-operation’’mode is the following: the level of the victim signal is grad-ually increased to the threshold of �61 dBm, when thethreshold is reached a trigger signal starts the measure-ment. The DUT Tx is monitored and when it falls belowthe avoidance level, a second trigger signal is generatedto stop the measurement. The procedure can continue totest the second threshold at �38 dBm. The victim signalis increased to �38 dBm and the DUT Tx power is collected.The testing time for this specific test is defined in [7] to thevalue of 150 s. Other tests can have different duration butthey must still be conformant to the requirements definedin the regulatory and standards documents.

For a precise detection of the Tx signal, it is importantthat its signal path, from DUT to the spectrum analyzer issufficiently isolated from both the WiMAX victim signaland the UWB device signal. To achieve a proper isolation,we can regulate the attenuator to ensure that the UWBdevice Tx power is reduced to an acceptable level at the

Fig. 5. 3.45 GHz WiMAX at �38 dBm, plus UWB (test mode) in band 2/TFC 6.

Fig. 6. Conducted DAA UWB test bed setup.

1008 G. Baldini et al. / Measurement 45 (2012) 1001–1014

spectrum analyzer. The overall attenuation from the UWBdevice to the input of the LNA is 33 dB with the attenuatorvalue set to 0 dB. If we assume that the UWB device trans-mits at a maximum power of �41 dBm, which is the limit

defined in [11], we have a resulting signal of �74 dBm atthe LNA. If the DUT is supposed to transmit at �80 dBm(Zone 1 from Section 2), its signal would be masked bythe UWB device at �74 dBm. In our test bed, we set the

Fig. 7. UWB DAA conducted test bed setup of the second site.

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1009

attenuation level at the variable attenuator to 35 dB, sothat the UWB device signal at the input of the LNA is equalto �109 dB, which does not affect the signal of the DUT.

This test setup is able to validate DAA UWB enabled de-vices in the majority of the cases.

Unfortunately, there are also many cases where con-ducted tests cannot be performed because DUTs may nothave detachable antennas. Furthermore, a complete testingand validation activity should consider a realistic scenario,where spurious emissions from the UWB receivers, multi-ple sources of interference and victim services are takeninto consideration. For those reasons, a radiated testingand measurement setup is also necessary.

4.3. Radiated test bed

The radiated test setup is designed in a similar way asthe conducted test setup. Fig. 8 shows a test configurationfor a radiated DAA setup.

In the radiated setup, a signal generator generates a vic-tim signal (e.g., WiMAX), which is transmitted through acalibrated horn antenna (i.e., test antenna) to the DUT,which acts as interferer. A radio link is maintained be-tween the DUT and the UWB device. The UWB signal trans-mitted by the DUT is captured by the spectrum analyzer byusing a calibrated test antenna. The test antenna is a Dou-ble Ridged Broadband Horn (BBHA 9120 D) from Schwarz-beck, with an isotropic gain between 11.18 dBi (at 3.2 GHz)and 11.63 dBi (at 4.7 GHz) at 1 m distance and between11.61 dBi (at 3.2 GHz) and 12.34 dBi (at 4.7 GHz) at 3 mdistance. The BBHA 9120 D was chosen because the gainwas relatively stable in the frequency range (3.2–4.7 GHz) for testing. The gain response in this frequencyrange was stored in the spectrum analyzer for automaticcalibration.

The positions of the victim signal generator and theUWB device are changed to test different spatial configura-tions. The spatial configurations are controlled through aMulti-Axis Positioning System (MAPS), which is a com-bined-axis system that mounts the phi-axis positioner onthe theta-axis positioner to rotate the DUT along these

two axes. The MAPS is needed to address the challenge de-scribed in Section 3.2 for measurements in radiated tests.The spatial configurations defined by MAPS are correlatedwith the antenna radiation pattern.

ETSI EN 302 065 [1] defines a standard measurementdistance of 3 m. Acknowledging the very low PSD levelsof UWB, measurements may also be done at a distance of1 m. It has to be observed, however, that measurementsare done in the antenna far-field. In the far-field of a radi-ating source, both electric (E) and magnetic (H) fields pos-sess single components only, both of which exhibit simple1/r dependencies. The single components of the E and Hfields will be orthogonal to each other and at right anglesto the direction of propagation. Depending on the distancethey have to travel the points of a wave emitted by thetransmit antenna arrive at the measurement antenna notsimultaneously but spread over time. The measurementerror introduced by this dispersion effect is considerednegligible if the difference between the maximum andthe minimum path length does not exceed k/16, which cor-responds to a phase lag of 22.5� at the extremities of theaperture of the measurement antenna.

In case of a radiated measurement setup consisting of aDevice Under Test (DUT) and a measurement antenna, theapertures sizes of both DUT antenna (d1) and measurementantenna (d2) have to be taken into account for determiningthe far-field distance DFF.

DFF P2kðd1 þ d2Þ2 �

k64

2" #

ð1Þ

For (d1 + d2)2� k2/64 this expression can be simplifiedas follows:

DFF P2ðd1 þ d2Þ2

kð2Þ

In addition, the following condition has to be fulfilledDFF P 5d1

Ref. [1] cites these formulas for calculating the far-fielddistance. The detailed derivation of the far-field distancecalculation is given in [16].

Fig. 8. Radiated DAA UWB test bed setup.

1010 G. Baldini et al. / Measurement 45 (2012) 1001–1014

The antenna of the DUT or d1 has a size of 1.5 cm, whilethe size of the measurement antenna or d2 (i.e., BBHA 9120D) has a size of 28.3 cm.

Inserting these physical aperture sizes into Eq. (2) re-sults in a far field distance that varies between 1.8 and6 m over the UWB frequency range. At the standard dis-tance of 3 m far-field measurements would be limited toa maximum frequency of 5 GHz. Reducing the measure-ment distance to 1 m as in principle permitted by the EN302 065 without violating the far-field conditions wouldbe impossible under these conditions. For broadbandantennas, however, not the physical aperture size but theeffective aperture has to be considered [17]. According to[18], the effective aperture De can be calculated from theantenna 3 dB-beamwidth (or half-power bandwidth =HPBW) values as follows:

LE ¼56k

HPBWE; LH ¼

67kHPBWH

; De ¼ffiffiffiffiffiffiffiffiffiL2

Eþq

L2H ð3Þ

where LE and LH are the effective aperture lengths in the E-and H-planes. As De is proportional to k, and HPBWE andHPBWH remain relatively constant over the UWB fre-quency range, the size of the effective aperture decreaseswith frequency. Applying the 3 dB-beamwidth figures pro-vided by the antenna manufacturer [19], the resultingeffective aperture De of the BBHA 9120D can be calculated.Over the UWB frequency range, De varies between 5 cmand 19 cm and is thus significantly smaller than the phys-ical aperture size of 28.3 cm.

By substituting d2 in (2) with De the effective minimumfar-field distance vs. frequency can be obtained. We findthat for the given measurement configuration the far-fielddistance is less than 1 m over the entire UWB frequencyrange. Resolving (2) and (3) for the maximum DUT antenna

dimension we find that the far-field distance does not ex-ceed 1 m for DUT antenna apertures of up to 3 cm.

Even if, the far-field condition is validated, we still needto ensure that it is possible to reliably detect the very lowUWB signal levels at a distance of 1 m.

For this purpose we examine the signal chain and thecorresponding link budget. The received power at the an-tenna output can be calculated as:

PRx ¼ PTx � GM=LFS

where LFS is the free-space path loss, PTx is the transmittedpower and GM is the isotropic measurement antenna gain,which is provided by the manufacturer [19].

The signal power at the input of the spectrum analyzercan be calculated as

Pln ¼ PRx � GLNA=ðGCables � GConnÞ

where GLNA is the gain of the low-noise amplifier, andGCables, GConn are the losses introduced by the cables andconnectors.

The noise-floor of the receiver system is determined bythe ambient noise level and the gains and noise figures ofthe individual system components. Here the receiver sys-tem consists of two microwave cables or connectors, alow-noise amplifier (LNA) and a spectrum analyzer.Assuming the measurements are done in a perfectlyshielded anechoic chamber (which is the case for all thethree test beds), the system noise floor is calculated as:

PNoise ¼ P0 � GLNA=ðGCables � GConn � FsystemÞ

where:

P0 ¼ KTB

is the background noise power via the Rayleigh–Jeansapproximation, k = 1.38 10�23 J/K is the Boltzman’s

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1011

constant, T is the ambient temperature in Kelvin, and B isthe bandwidth in Hertz. For T = 293 K, P0 is approximately�114 dBm. Fsystem is the system noise figure, which in-cludes the noise figure of the spectrum analyzer (FSA) aswell.

Table 2 shows the receiver system parameters for a spe-cific test bed, which provide an average system noise figureof FSystem = 1.7dB.

With the information compiled above we can nowdetermine the obtainable UWB signal-to-noise ratio(SNR) for the measurement system:

SNR ¼ PIn

PNoiseð4Þ

To test the system with the above settings, we gener-ated a signal which sweeps from 1 GHz to 11 GHz follow-ing the UWB power spectral density mask. The resultsare shown in Fig. 9

From Fig. 9, we can see that the Rx signal is above thenoise floor until 10.6 GHz. For measurement campaigns a6 dB margin is recommended. An option investigated inthe project, is to reduce path loss by decreasing measure-ment distance whilst still fulfilling the far-field conditionsthrough the use of small-size planar UWB broadbandantennas.

In summary, the challenges described in Section 3.2,were addressed by the following test bed design solutions:

1. Low power level of the UWB signal: a LNA was used toincrease the power of the signal emitted by DUT. Thespectrum analyzer was configured with 1 MHz RBW.Isolators and attenuators were configured to minimizethe interference on the DUT signal. The parameters ofLNA and equipment were calculated on the base ofthe noise figures.

2. Timing: the testing procedures were automated withLabVIEW and triggers. A frequency standard was usedto synchronize the elements of the test bed.

3. Radiated tests: MAPS is used to identify the optimumspatial configuration for the radiated test.

5. Measurement campaigns and results

This section describes a subset of the results of themeasurement campaigns conducted at the three test bedsduring the WALTER project. The results presented here arerelated to measurements executed in the conducted testbed as described in Section 4.2.

The most relevant parameters of the WiMAX victim sig-nal are described in Table 3.

Table 2Receiver system parameters.

Parameter Value Unit

GLNA 40 dBFLNA 1.6 dBL1 0.1 dBL2 1.4 dBFSA 20.93 dBT 293 K

The test of the DUT samples, is based on threeparameters:

� Time: which is the scanning time in seconds of the DAAUWB device.� Number of detections: it shows the number of detections

within the scanning time (according to ‘‘Time’’) in casethat detection is achieved. The maximum number ofdetections is 255.� Sensitivity threshold: it is the sensitivity threshold for

detection. Higher values will result in better sensitivity.Valid values for the DAA UWB devices used in the testare from 0 (minimum) to 30 (maximum).

The DUT can operate in a number of scanning modeswith different scanning times. In the test scenarios, weevaluated various modes: Full DAA, when the DUT use100% of the time for detection, partial 25%, when it uses25% of the time for detection, partial 18.75%, partial12.5% and partial 6.25%. Samples are supposed to detectsignal patterns according to [16] in a frequency range of3.1–3.6 GHz.

Measurement campaigns were executed, in all threetest beds, in the conducted test configuration describedin Section 4.3 (radiated test bed) for the combinations ofscanning modes, sensitivity thresholds and different valuesof the victim signal power (i.e., WiMAX). During the tests,the victim uplink signal power was progressively de-creased from a peak of �20 dBm to a minimum of�80 dBm. The pass/fail test criteria is the detection of theWiMAX victim signal at �61 dBm.

For each combination, the number of detections werecalculated in the measured scanning time. Examples ofthe collected results for measurements executed in thefirst test bed are provided in Fig. 10, for a scanning modeof 6.25% and in Fig. 11 for a scanning mode of 25% for asensitivity threshold of 0 � 30, which is the maximumvalue.

5.1. Analysis and discussion of the results

The measurement results were compared and analyzed.Fig. 12 shows the number of detections by the DUT with

a scanning mode of 25% for all the three test beds. The min-imum power levels of the WiMAX victim signal detected inthe various test-beds are different but in all cases theselevels allow for the detection of the �61 dBm limit (indi-cated with an arrow in the figure), which is the minimumpower level to be detected (from Ref. [16]) and determinesthe pass/fail criteria. Actually the measurements results, inthe first test bed, show that the DAA UWB device is capableof detecting WiMAX victim signals up to �80 dBm, evenwith partial scanning modes of 6.25% and 25%.

In Fig. 12, we note that the results of the second test bedare significantly lower then the other test beds. Even if theDAA UWB device still passes the test with detectionsrecorded up to �78 dBm, the performance is worst thanthe other test beds. As mentioned in Section 4, the secondtest bed used a real WiMAX device instead of a signal gen-erator like the other test beds. This could be the reason forthe difference in the results.

Fig. 9. Received signal at the spectrum analyzer input, compared to the system noise floor.

Table 3Receiver system parameters.

Parameter Value

Transmission mode Uplink only – Frequency-Division Duplexing (FDD)Size of fast Fourier transform 1024Bandwidth 7 MHzCenter frequency 3.41 GHz, 3.5 GHz, 3.459 GHz, 3.61 GHz, 3.7 GHz and 3.79 GHz.Pattern Bursts of 8 OFDM symbols in the uplink repeated every 20 ms, to simulate VoIP trafficCoding scheme with Channel Coding (CC) QPSK (CC) 16QAM (CC) and 64QAMNumber of slots occupied 14–84. Slot is the basic unit of allocation in the time–frequency grid of WiMAX

Fig. 10. Number of detections for partial 6.25% mode in the first test bed. Fig. 11. Number of detections for partial 25% mode in the first test bed.

1012 G. Baldini et al. / Measurement 45 (2012) 1001–1014

Fig. 13 compares the detection performance of the DUTfor different values of the scanning mode in the first testbed. We can see that the detection time in millisecondsdecreases with higher scanning modes, where a greaterpercentage of the time is dedicated to scanning. This is ex-pected, as the DUT will have more time to detect the Wi-MAX signal. These results can be used to identify the bestvalues for the scanning mode when the DAA UWB devicesare deployed in the market. The scanning mode represents

clearly a tradeoff between the detection and the communi-cation functions of the DAA UWB, which cannot allocate allits processing power only to detect the WiMAX signal.Fig. 13 also shows the detection performance in relationto the WiMAX signal power from �20 dBm to �70 dBm.The performance is better (i.e., decreased detection time)when the WiMAX signal power is higher. This is also ex-pected, because the detection function performs better inthe presence of a more powerful WiMAX signal in space.

-20 -30 -38 -40 -50 -60 -70 -72 -75 -76 -77 -78 -79 -80Second Test Bed

Third Test BedFirst Test Bed

0

20

40

60

80

100

120

140

Second Test Bed

Third Test Bed

First Test Bed

Num

ber o

f det

ectio

ns in

a s

econ

d

WiMAX signal power in dBm

Fig. 12. Number of detections for Partial 25% mode in the first test bed.

6.25 12.5 18.75 25 100

-20 dBm-30 dBm-50 dBm-70 dBm

0

10

20

30

40

50

60

Scanning Modes

Det

ectio

n tim

e in

mill

isec

onds

Fig. 13. Detection performance in relation to the scanning modes.

6.25 12.5 18.75 250

10

20

30

40

50

60

70

80

Det

ectio

n tim

e in

mill

isec

onds

Sensitivity threshold

Threshold = 30Threshold = 24Threshold = 20

Fig. 14. Detection time in milliseconds in relation to the sensitivitythreshold.

-20 -30 -50 -70

Victim Signal Power

Det

ectio

n Ti

me

in m

illis

econ

ds

0

2

4

6

8

10

12

14First test bedSecond Test BedThird Test Bed

Fig. 15. Detection time in milliseconds in relation to the victim signalpower.

G. Baldini et al. / Measurement 45 (2012) 1001–1014 1013

Fig. 14 describes the detection time in milliseconds inrelation to different values of the scanning mode and fordifferent values of the sensitivity threshold. This graphconfirms the previous results that the detection time de-creases with greater scanning modes. The graph showsthat the sensitivity threshold is not a meaningful parame-ter to decrease the detection time. Only for lower values ofthe scanning mode (i.e., 6.25%) the sensitivity threshold isrelevant: 50 ms of detection time for threshold of 30 and67 ms for threshold of 20.

Fig. 15 shows the performance of the DUT in relation tothe victim signal power for the scanning mode 100% for allthe three test beds. As expected, lower values of the Wi-MAX victim signal power increase the detection time. Asdescribed before the DUT had higher detection times inthe second test bed.

After carrying out the previous set of tests, the effect ofusing different centre frequencies, modulation/codingschemes and number of slots occupied was analyzed andthe conclusions are provided below:

1014 G. Baldini et al. / Measurement 45 (2012) 1001–1014

1. The number of slots occupied was increased to 84 keep-ing the rest of the parameters as before. No effect wasappreciated in the number of detections or the mini-mum victim signal power detected. We can concludethat number of slots occupied by the WiMAX victim sig-nal has no influence on the number of detections or theminimum victim signal power detected.

2. The modulation scheme of the WiMAX victim signalwas modified to 16QAM (CC) 1/2, 16QAM (CC) 3/4,64QAM (CC) 1/2 and 64QAM (CC) 3/4 keeping the restof the parameters as before. No effect was appreciatedin the number of detections or the minimum victim sig-nal power detected. We can conclude that the modula-tion and coding scheme used for the victim signal hasno influence on the number of detections or the mini-mum victim signal power detected.

3. The centre carrier frequency used for the victim signalhas influence on the minimum victim signal powerdetected. For frequencies around 3.5 GHz the minimumpower level detected is around �80 dBm. For higherfrequencies the minimum power level detectedincreases to �59 dBm.

6. Conclusions and future developments

A networked test bed based on three test bed sites wasdesigned and implemented as part of the FP7 projectWALTER. Prototypes of DAA enabled UWB devices werevalidated against regulation and standardization require-ments. The test bed design and measurement campaignscan provide an important feedback to the regulatory,industry and research communities for the evaluationand testing of innovative wireless technologies and spec-trum management approaches.

WALTER project recommended to regulatory and stan-dardization bodies the testing procedure for the detectiontime of the DAA UWB devices. The project also recom-mended a change in the definition of the detection proba-bility to improve the efficiency of the testing procedure.Low NF and stable frequency response in the UWB fre-quency range (3.2–4.7 GHz) are still desirable features,which should be improved in the next generation ofUWB test bed equipment.

As DAA is a basic form of a cognitive radio technique, atest bed to evaluate DAA UWB enabled devices could bea first step to define a more generic test bed for cognitiveradio devices and dynamic spectrum managementimplementations.

Once full Detect and Avoid (DAA) UWB devices will beavailable in the next future, new measurements campaignscan be run to validate the Avoid function as well.

Aknowledgment

This work was performed in the FP7 project WALTERwhich has received research funding from the European

Community’s Seventh Framework program. This paper re-flects only the authors’ views and the European Commis-sion is not liable for any use that may be made of theinformation contained therein.

References

[1] ETSI EN 302 065 v1.1.1, Electromagnetic Compatibility and RadioSpectrum Matters (ERM); Ultra Wideband (UWB) Technologies forCommunication Purposes; Harmonized EN Covering the EssentialRequirements of Article 3.2 of the R&TTE Directive, EuropeanStandard Telecommunication Institute (ETSI), February, 2008.

[2] FCC CFR Title 47 Part 15 Subpart C, Intentional Radiators, FederalCommunications Commission, October, 2007.

[3] ECC Report 120 (June 2008) on Technical Requirements for UWBDAA (Detect And Avoid) Devices to Ensure the Protection ofRadiolocation in the Bands 3.1–3.4 GHz and 8.5–9 GHz and BWATerminals in the Band 3.4–4.2 GHz.

[4] F. Berens, The European flexible DAA approach towards an openUWB regulation, in: ICUWB 2008, IEEE International Conference onUltra-Wideband, vol. 3, 2008, pp. 71–75.

[5] FP7 WALTER Project. <http://www.walter-uwb.eu/> (accessed10.01.11).

[6] ECMA-368 – High Rate Ultra Wideband PHY and MAC Standard,December, 2007.

[7] ETSI Technical Specification ETSI TS 102 754. v.1.2.1 (2008-10),Electromagnetic Compatibility and Radio Spectrum Matters (ERM);Short Range Devices (SRD); Technical Characteristics of Detect andAvoid (DAA) Mitigation Techniques for SRD Equipment Using UltraWideband (UWB) Technology.

[8] FCC CFR Title 47 Part 15 Subpart F, Ultra-Wideband Operation,Federal Communications Commission, October, 2007.

[9] Ministry of Industry and Information Technology of the People’sRepublic of China, Wireless Project Number 354, [2008] Notification‘‘to release UWB technology spectrum specification’’. <www.miit.gov.cn> (accessed 10.01.11).

[10] [ECC/DEC/(06)04] ECC Decision of 24 March 2006 on theHarmonised Conditions for Devices Using Ultra-Wideband (UWB)Technology in Bands Below 10.6 GHz.

[11] [ECC/DEC/(06)12] ECC Decision of 1 December 2006 on theHarmonised Conditions for Devices Using Ultra-Wideband (UWB)Technology with Low Duty Cycle (LDC) in the Frequency Band 3.4–4.8 GHz.

[12] ECC Decision of 1 December 2006 amended 31 October 2008 onSupplementary Regulatory Provisions to Decision ECC/DEC/(06)04for UWB Devices Using Mitigation Techniques.

[13] S.M. Mishra, R.W. Brodersen, S.T. Brink, R. Mahadevappa, Detect andavoid: an ultra-wideband/WiMAX coexistence mechanism [topics inradio communications], IEEE Communications Magazine 45 (6)(2007) 68–75.

[14] S. Kandeepan, G. Baldini, R. Piesiewicz, Experimentally detectingIEEE 802.11n Wi–Fi based on cyclostationary features for ultra-wideband cognitive radios, in: 2009 IEEE 20th International Symposiumon Personal, Indoor and Mobile Radio Communications, 2009. pp.2315–2319.

[15] CMMI for Development, Version 1.2V by Software EngineeringInstitute, Carnagie Mellon.

[16] ETSI TR 102 763 v1.1.1, Electromagnetic Compatibility and RadioSpectrum Matters (ERM); Short Range Devices (SRD); TechnicalCharacteristics of Detect-And-Avoid (DAA) Mitigation Techniquesfor SRD Equipment Using Ultra Wideband (UWB) Technology,European Standard Telecommunication Institute (ETSI), June,2009.

[17] Y. Qin, Measurement of 1–40 GHz radiated electric field spuriousemissions, in: 2001 IEEE International Symposium on ElectromagneticCompatibility, vol. 1, 2001, pp. 458–463.

[18] J.D. Kraus, Antennas, second ed., McGraw-Hill, Inc., 1988.[19] Schwarzbeck BBHA9120D Radiation Pattern. <http://www.sch

warzbeck.de/Datenblatt/ri9120d.pdf> (accessed 10.01.11).