Table Of Content

This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 workshop proceedings. A Cognitive Radio Receiver Supporting Wide-Band Sensing Volker Blaschke, Tobias Renk, Friedrich K. Jondral Universita¨t Karlsruhe (TH), Institut fu¨r Nachrichtentechnik {blaschke|renk|fj}@int.uni-karlsruhe.de Abstract—IEEE 802.22 defines the world-wide first cogni- presented. Within the IEEE 802.22 frequency range several tive radio (CR) standard. In the range between 41MHz and types of channel utilization can be identified. This a-priori 910MHz CR overlay-systems can be installed besides licensed knowledge combined with appropriate information processing radio services such as radio and TV broadcasting. In order lead to an optimized CR front-end structure supporting wide- to fulfill the regulative guidelines for interference limitations, adequatespectralsensinganduserdetectionhastobesupported band spectral sensing. by the CR terminals. The wide frequency range specified in The paper is structured as follows: In the next section a IEEE 802.22 and the high dynamic range of signals in this brief introduction to ADCs is presented. In section III a band lead to high demands on the CR receiver’s front-end. detailed description of sensing algorithms in CR terminals Especially the performance requirements on analog-to-digital is given. Furthermore, expected signal characteristics within converters (ADCs) increase significantly compared to current wirelesssystems.Basedonmeasurementstakeninthefrequency the IEEE 802.22 frequency bands are identified. Based on range between 41 MHz and 910 MHz requirements to CR’s these considerations, two general wide-band sensing methods ADCs are figured out. Furthermore, the measurement results includingacomparisonoftheirdemandsonADCperformance are analyzed regarding expectable allocation scenarios and their are described in section IV. In section V a CR receiver impacts on spectral sensing. Derived from these results and a structure is presented which includes adapted spectral sensing comparisonofgeneralspectralsensingmechanisms,anapproach for a CR receiver enabling wide-band sensing is presented. By and convenient information processing. Finally, a conclusion combining a-priori information resulting from scenario analysis is given. with adapted information processing in the CR terminal, the ADC’s performance requirements can be reduced. II. ANALOGTODIGITALCONVERSIONINCRRECEIVERS A suitable sampling and quantization of the analog input I. INTRODUCTION signal is a precondition for digital signal processing. Most Spectrum Sensing becomes more and more important, es- CR concepts described in literature only assume an appro- pecially in the context of cognitive radio (CR). Due to the priate ADC as precondition without detailed specification. increased demand on wireless transmission resources and the But supporting the performance constraints assumed in these ongoing installation of new radio access technologies for approachescontainseveralchallengesinADCdesign.Thekey broadbandaccess,enhancedresearchinthefieldofmobileCR parametersforsummarizingtheADC’sperformancearestated receivers is necessary. Resulting form spectral measurements resolution, signal-to-noise ratio, spurious free dynamic range, [1], [2] a low utilization over wide frequency ranges was and power dissipation [8]. Furthermore, aperture jitter as well identified. In order to overcome this waste of resources the as the two-tone intermodulation distortion are important for development of intelligent allocation mechanisms becomes characterizing ADCs. A detailed description and performance necessary. Dynamic spectrum allocation (DSA) of unused analyzes can be found in [8] and [9]. In [8] the performance radio resources is required for increasing the overall utiliza- of on-the-market ADCs is analyzed in order to describe the tion. This approach is supported by the CR concept [3], [4]. evolution and trends in ADC technology. This evaluation was For providing DSA, sufficient information about the spectral continued in [9] including also present-day trends. environment has to be collected using different acquisition In this article ADC parameters which restrict an implemen- methods.Ononeside,allinformationiscollectedbyacentral tation of current ADCs in mobile CR terminals are figured control unit and distributed to the mobile terminals. In this out. These are sampling frequency fs, impacting the effective casetrafficloadinformationcouldbeexchangedbetweenjoint resolutionbandwidth,effectivenumberofbitsNeff,describing networks via backbone [5]. Other approaches are based on the dynamic range that is supported by the ADC, and power distributed spectral sensing where all mobile terminals of a dissemination Pdiss influencing battery running time. radio network are used. This requires sensing capabilities in In order to fulfill the Nyquist criterion the converter sampling eachmobileentityandanefficientalgorithmforconsolidation frequencyfshastobemorethantwotimestheefficientanalog oftheresults.Especiallyinsuchscenarios,swarm-intelligence bandwidth[10].Furthermore,theeffectivenumberofbitsNeff algorithms could offer additional benefits [6]. Nevertheless, is lower than the stated number of resolution bits specified by appropriate spectral sensing and information extraction forms the vendor. Due to hardware imperfections and quantization a precondition for DSA. In order to avoid unacceptable inter- noise Neff is given by Neff = D6−.012.76, where D describes the ferences reliable user detection has to be supported by CRs. effective dynamic range of the converter in dB [8]. Especially This leads to additional demands on its front-end. indetectionandsensingapplications,ahighdynamicrangeis In this paper the performance demands on analog-to-digital important. If the dynamic range of the expected input signals converters(ADCs)inmobileCRreceiverssupportingspectral is higher than the ADC dynamic range, weak signals may not sensingarediscussed.Basedonthefrequencyrangesspecified be detected due to resolution limitations. in IEEE 802.22 [7] general aspects of wide-band sensing as Havingalooktotheresultsdepictedin[8]and[9]threemain well as specific demands on the structure of mobile CRs are hardware architecture concepts become potential candidates 978-1-4244-2052-0/08/$25.00 ©2008 IEEE 499 Authorized licensed use limited to: CNUDST. Downloaded on January 4, 2009 at 17:23 from IEEE Xplore. Restrictions apply. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 workshop proceedings. for implementation in mobile CR terminals. These are: Flash −40 Converters, Pipelined ADCs, and Σ∆ Converters. −45 Flashconvertersofferveryhighsamplingratesduetoaparal- −50 lel comparator structure. This enables sampling rates of about −55 1higGhsppsowbuetr rceoqnusiurmesphtiiognhmhaarkdewsatrheemcomunpaltetxraitcyt.ivFeufrothremrmoobrilee, wer (dBm)−−6650 ParepisppoellilucinatietoidonnAos.fDDoCunselyNtoNetffehffecah=nig6hb..eh.8airndbcwriteasarseceacdnomubpeplsteouxpi1tpy5oarbtneidtes.ffUaetcstitinhvgee avgeraged po−−−877050 expenseofareducedsamplingfrequencyfs whichislessthan −85 500 Msps. The specific design of Pipelined ADCs requires −90 theimplementationofanalogtrack-and-holdblocks.Thethird −95 40 ch1 ch2 ch3 910 group of potential candidates offers an effective resolution of frequency (MHz) up to 20 bits but the achievable sampling rate is less than 100 Msps.DuetotheirlowpowerconsumptionΣ∆Convertersare Fig.1. Averagepowervs.frequencyinIEEE802.22frequencyrange. veryinterestingforanimplementationinmobileCRterminals. FurtherinformationaboutthedifferentADCstructurescanbe 1 ch. 1 found in [10], [11]. 0.9 cchh.. 23 aunadtDiIouIcnIrhi.naanngSdnPetEuhlCseaeTlrRllaorUsectMqatuyiSioerEneaNmrsaSedInIsaNMtepsGvtPeewAAdraCeNlrTtDeoSapRptuhpEberLoliAacsTcuhhEreredDes.nHTtfoAhsrepRseDeDcWtScrAouAnmRcE[e1spi2tts]- averaged channel utilization0000000.......2345678 combine available transmission resources of several systems 0.1 optimizingthespectrumutilizationofallparticipatingsystems. 0 In order to provide traffic load information of each system to time all other systems a global control entity and dedicated control channels via the wired backbone network are necessary. This Fig.2. AveragechannelutilizationOtn ofchannel1,2and3. means that each system has to be connected to this dedicated control channel. Therefore, this architecture requires coopera- utilization, the full frequency range has to be supported by tion between all radio access networks participating in DSA, the mobile terminals. This also includes the ability to fast and which again results in an unflexible system configuration. efficient sensing of wide ranges in order to adapt the radio Anotherapproachforincreasingspectralutilizationareoverlay transmission parameters to the licensed user’s allocation. The systems [13]. Based on the observation that the frequency widesystembandwidthresultsinincreasedrequirementstothe range of most established radio systems is under-utilized, this hardwarearchitecture.Especiallytheanalogsignalprocessing approach describes the allocation of local wireless networks and the ADC performance limit the supported bandwidth. As exploitingunoccupiedradiotransmissionchannels.Duetothe it is described in section II, there is a trade-off between the basic precondition that established, licensed systems should bandwidth and the dynamic range of the converter. not be changed for overlay usage, the rental user must A. Basics on Spectral Analysis observe the communication channel in order to provide a reliable detection of the licensed user’s channel allocation. Analyzing the receiving signals within the IEEE 802.22 Furthermore, the rental user’s signal has to be adapted to frequency range, several allocation characteristics can be ob- the licensed system’s transmission parameters with respect served. In Fig. 1 the received signal power per frequency to channel bandwidth, maximum channel allocation dura- averaged over a sensing period of 3hours is depicted. The tion, transmission power, etc. Static system parameters, e.g., measurements presented here are taken in an urban area. As channel bandwidth, can be defined in a database available it can easily be seen, wide frequency ranges are character- for each rental user. For dynamic parameters, e.g., current ized by a very low average signal energy. But also highly channel allocation, licensed user’s allocation statistics have to utilized bands, e.g., the European FM radio broadcast service be observed and analyzed. In order to gain information about between 88MHz and 108MHz (cf. ch 1), can be figured the current allocation of frequency bands, energy detection out. Furthermore, there are some TV broadcast signals as can be used in overlay systems. If the received signal power wellastemporarilyallocatedchannels.Describingthechannel inacommunicationchannelishigherthanthemeasurednoise allocation during sensing time a binary spectrogram can be level, this channel is already occupied. This gives general defined: (cid:2) (cid:1) isTnihgfoenrraemlfsoatrthieoa,ntaaanrbaeolyucztliontshgeethcoiugnrhroeeirnsteourldteivelierzlasettianotenirs.gtiBycsudteoitfnecactiaossinegncoaaflnwmfeaaailyk. O{x(t)}(cid:1)(cid:1)f=fm = 10 ffoorrSS{{xx((tt))}}(cid:1)(cid:1)(cid:1)ff==ffmm >≤PPtthh , (1) become necessary [14]. (cid:1) IcnomIEmEuEnic8a0ti2o.n22chaanCneRl wteirtmhianablanisdwaildlothweBdchto=u6s.e..a8rMadHioz twhheesrepexct(rto)griasmthoefrxec(et)ivaetdfsreigqnuaeln,cSy{fxm(t,)a}n(cid:1)df=Pftmh ddeessccrriibbeess inthefrequencyrangebetween41MHzand910MHz[7].So, thedetectionthreshold.Theresultingbinarydescriptionofthe the overall system bandwidth is BS =869MHz. This is more spectralallocationcanbeaveragedoverthesensingtimeusing than 100 times the signal bandwidth Bch. In order to provide a window length Nw. The average channel utilization can be a flexible CR system which is able to optimize the spectral written as 500 Authorized licensed use limited to: CNUDST. Downloaded on January 4, 2009 at 17:23 from IEEE Xplore. Restrictions apply. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 workshop proceedings. Fig.3. SuperheterodynereceiverwithtwoIFchainsandlowfrequencyADCfortheIEEE802.22specification. frequency of 970MHz mixes the 1000MHz signal down to Otn(cid:1)(cid:1)f=fm = N1w tn+(cid:3)Nw−1Otn(cid:1)(cid:1)f=fm, (2) fcaIFn2ea=sil3y0bMeHfizl.teIrmedagoeuftrebqyuetnhceyIiFs294fi0ltMer.HzBowthhicLhOasgaairne t=tn controlledbyPLLsthatareconnectedtoareferenceoscillator where tn is the time index. The resulting characteristics for to increase frequency accuracy. The automatic gain control threedifferentfrequenciesaredepictedinFig.2.Thecurveof (AGC) block is followed by a tunable bandpass filter that channel1describesatypicalutilizationofbroadcastchannels. determines the resolution bandwidth. The ideal case for the Channel 2 offers a varying average occupation between 0.01 resolution filter is a rectangular filter with bandwidth BR. In and 0.92. In channel 3 some time periods with an average ordertoachieveshortmeasurementtimes,however,optimized utilization of 0 can be noticed. During this period the channel Gaussian filters are used that are temperature stable and is not allocated by the licensed user and is thus an interesting possess a higher bandwidth accuracy as well. Nevertheless, candidate for CR usage. In order to detect such periods that BR influences the sweep time Tsw that is necessary to scan are also known as white spaces [4] a suitable observation of the whole frequency range BS. If measurement time falls wide frequency ranges has to be provided. As can be noticed below Tsw, amplitude losses and signal distortions occur in Fig. 1, the depicted signal possesses a high dynamic range. that eventually lead to frequency offsets. After filtering there Duringmeasurementsamaximumsignalpowerof−41.8dBm are the envelope detector and ADC. The video filter is a atchannel1wasobserved.Thegeneralnoiselevelmeasuredin lowpass with the bandwidth BV that suppresses noise and unallocated sub-bands is −92dBm. So, the overall dynamic helps to smooth the signal spectrum. The micro-processing range compared to the bandwidth of more than 850 MHz block (µP) includes among others averaging and threshold marks the main challenge in finding suitable solutions for decision making. mobile CR receivers supporting this wide frequency range. For the definition of the required sweep time, two cases must be taken into consideration. One where the video bandwidth is higher than the resolution bandwidth and vice versa [15]: IV. SPECTRUMSENSINGMETHODS  Due to the fact that currently available ADCs possess a  k· BB2S forBV >BR limitedbandwidth,itisnotpossibletodigitizelargefrequency Tsw = R , (3) s(FpaFnTs) aatnaolynzceer.sTahreereofnolrye,ussaob-cleallfeodr lfoawst fFroequurieenrcytrasnigsfnoarlms, k· BRB·SBV forBV <BR because the input signal is lowpass filtered and converted where the parameter k denotes a proportional factor that into the digital domain directly [15]. In order to investigate is usually in the range of 1 to 3. The first case of high frequency signals, superheterodyne receivers must be (3) is illustrated in Fig. 4. System parameters were cho- applied. Here, the overall input frequency span is mixed to sen according to the IEEE 802.22 specification. The verti- a common intermediate frequency (IF) by a tunable local cal dashed lines define the channel bandwidth of Bch = oscillator [15]. Spectral resolution is directly determined by 6...8MHz [7]. For one single scan through the whole the IF filter. The smaller the resolution bandwidth BR, the frequency range of 869MHz approximately 3 · 10−4s higher the spectral resolution. A well-known problem occurs are necessary if we consider a resolution bandwidth of iftheinputfrequencyrangeismorethantwotimeshigherthan BR = 2.5MHz which means we take two to four frequency theIF,becausethensuppressionoftheimagefrequencyisnot bins per channel. As can be seen in Fig. 3, the ADC is possible without affecting the input signal. Hence, a tunable bandpass is necessary which would be very complicated for a frequency range of several hundred MHz. This problem can 102 be overcome if several IF stages are used, where the first one 101 transforms the input signal to a higher frequency. It is then possible to suppress the image frequency without affecting 100 tFfhoierg.itnh3peusIhtEosEwigEsnaa8l0.s2u.p2e2rhsepteecriofidcyantieonre.cTeihveerRwFitihnptuwtosiIgFnaclhafiirnsst weep time (s)1100−−21 Bch s passes RF attenuation that helps to prevent overload and 10−3 distortion. Afterwards, a preselector lowpass filters out higher frequency signals. In order to mix the RF signal up to 10−4 k = 1 i1n00a0MfreHqzue(fnIcFy1)r,atnhgeefirfsrotmloca1l0o4s1cMillaHtozrt(oLO1911)0mMusHtzo.peTrahties 10 4 kk == 23 Bs = 86190 5MHz 106 107 resolution bandwidth (Hz) leads to an image frequency that ranges from 2041MHz to 2910MHz which can easily be suppressed by the following Fig.4. Tsw inrelationtoBR withparameterk=1,2,3 IF1 filter. After this, a second local oscillator (LO2) with a located after the resolution filter and the envelope detector. 501 Authorized licensed use limited to: CNUDST. Downloaded on January 4, 2009 at 17:23 from IEEE Xplore. Restrictions apply. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 workshop proceedings. This means that only a bandwidth of BR has to be digitized for a successful operation of flexible overlay systems. On the whichleadstoahigheramplituderesolution.Moreover,aquite other side, each concept has significant drawbacks that can simpleADCissufficientinthatcase.Drawbacksare,however, not be solved within the next couple of years and preclude that pretty much analog hardware is necessary and that the an implementation in mobile terminals. So, a combination of measurement time is increased (cf. section V). Additionally, both gives the opportunity to combine the advantages. This no feature detection is possible since only energy detection is approach is discussed in section V. performed. V. WIDE-BANDCRRECEIVERFORIEEE802.22 A. Wide-Band Sensing A. Sub-band Spectrum Sensing Another possibility for energy detection is the down- As we presented above, a digitization of the complete conversion of the wide-band radio signal to an intermediate frequency used for ADC. This reduces the number of inter- system bandwidth BS is not useful with respect to technical andeconomicalconstraints.ButenhancementsinCRspectrum mediate frequency stages required for sweeping the detection sensing are necessary to increase spectral utilization. In order windowasitisusedinwide-bandspectrumanalyzers.Butthe to simplify spectrum sensing and to reduce the hardware performance requirements to the ADC increase significantly. requirements mentioned above we analyze expectable sig- Ideally, the incoming analog signal is filtered by a bandpass withsystembandwidthBS.Afterwards,thesignalisamplified nal characteristics resulting from measurements in the IEEE and down converted from the radio frequency fRF to an 802.22 frequency range. Having a closer look at the results intermediate frequency fIF. After a second filter stage and depictedinFig.1andFig.2,thechannelutilizationinthesub- bandallocatedbybroadcastservices(cf.ch1)doesnotchange AGC there is the analog to digital conversion. The following duringthemeasurementtime.Furthermore,thereceivedsignal dataprocessingincludesanFFTinordertoextractthecurrent power of broadcast transmitters averaged over the sensing powerallocationoverthefrequency.Afterdigitizationthesig- time is significantly higher than the noise level. In general, nal contains information of the complete observed frequency the average power level of the broadcast signals is higher range. Of course, the information depth is characterized by than−70dBmandevenhigherthan−50dBmconsideringthe the resolution performance of the ADC, which is a main strongest signal. Therefore, a continuous sensing of these fre- drawback of this approach. As it was shown in the section quencyrangesdoesnotofferanyadditionalinformationforan before,theobservedsignalischaracterizedbyahighvariation operatingCRsystem.Thisleadstotheopportunitytoexclude of the spectral power density. The overall dynamic range is more than 50dB. Furthermore, under-utilized small-band such quasi-static frequency ranges considering its information entropy during the spectrum sensing phase. Additionally, the signalswithweaksignalamplitudescanbenoticed.Inorderto resulting decreased dynamic range of the input signal reduces provide a reliable detection of the licensed user, these signals the ADC hardware requirements. Without loss of generality still have to be noticeable after the ADC, otherwise signal statistical independence of observed communication channels detection fails. Therefore, the dynamic range of the ADC is animportantparameterforCRterminals.BasedonNeff given canbeassumed.Thus,thespectrumcanbedividedintoseveral sub-bands in order to detect the average channel allocation. above the presented measurements would require a minimum resolution of Neff =9bits. Besides a high bit resolution, a These sub-bands do not need to be observed at the same time but can be sensed sequentially as long as the sensing high sampling rate is required for wide band digitization. is repeated periodically and the sensing interval as well as Following the example of IEEE 802.22, the sampling rate is fsamp = 1.82Gsps meeting the Nyquist criterion. Due to the sensing period is suitable to the licensed user’s signal. the overall frequency range from 41MHz up to 910MHz, Based on this knowledge and assuming a suitable sensing a bandpass sub-sampling cannot be used for reduction of sequence the system bandwidth BS can be split into M sub- fsamp. As stated in [9], the maximum sampling rate for bandsresultinginBsub =BS/M.Eachsub-bandisseparately an effective resolution of Neff = 9bits is about fsamp = digitized reducing the sampling rate of the ADC. Due to the 500Msps...1Gsps.Theanalysesin[8]and[9]alsodiscover resultingdecouplingofcommunicationchannelswithdifferent utilizations, the dynamic range of the ADC’s input signal that sampling rates significantly higher than 100 Msps can can be optimized for sensing weak signals. Additionally, the onlybehandledbyFlashorPipelinedconverters.Asdescribed suppression of strong signals using analog notch filter is not in section II, these two architectures are characterized by a necessary, because the reduced sampling rate enables higher high power consumption and, therefore, not preferable for an implementation in mobile terminals. The group of Σ∆- bit resolution. converters offer lower power consumption but cannot provide B. Receiver Structure thehighsamplingrateavailablewithFlashconverters.Inorder to suppress the undesired signals analog notch filters can be The system architecture described in section III-A defines applied, too. This requires a first scan for identification of the the basis for our wide-band CR receiver supporting sub-band strongest signals. After tuning the notch filter to these fre- sensing. In contrast to the structure depicted in Fig. 3 the quencies a second scan is used for detection of weak signals. ADC is placed directly after the IF 2 filter. This position Besidesdoublingthescantimealsotheanaloghardwareeffort is marked with the letter ‘A’. The adapted signal processing increases significantly. structure is depicted in Fig. 5. Up to marker ‘A’ the analog Comparedtotheenergydetectiondescribedabove,wideband signalprocessingisthesameasdescribedinsubsectionIII-A. digitizationoffersabettertimeresolutioninawidefrequency In order to support the IEEE 802.22 specifications the sub- band. Thus, a detailed extraction of temporal features is pos- band bandwidth is defined to Bsub = 50MHz including six sible. A high time resolution becomes important for detection sub-channels with a bandwidth of 8MHz up to eight sub- of very short channel allocations or for detailed analysis of channels with a bandwidth of 6MHz, respectively. Similar to the licensed user’s allocation statistics. the number of sweep points defined in analyzer detection the Both concepts offer different advantages which are required LO1 can be tuned to 20 predefined frequency steps resulting 502 Authorized licensed use limited to: CNUDST. Downloaded on January 4, 2009 at 17:23 from IEEE Xplore. Restrictions apply. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the ICC 2008 workshop proceedings. the sub-band offering the lowest utilization is selected for the upcoming sub-band switch of all overlay users. In Fig. 6 the described sensing process is depicted. The data transmission including sensing in the currently used sub-band is named S1, the sensing of the sub-band(s) offering the next lowest utilization is named S2, and the full range sensing is named S3. With this sensing process the amount of irrelevant data that have to be processed in the CR terminal can be reduced. Fig.5. Structureofwide-bandCRreceiversupportingsub-bandsensing. VI. CONCLUSION In order to increase spectral utilization future CR receivers have to provide enhanced spectrum sensing capabilities for Fig.6. Sensingprocessforwide-bandCRsupportingsub-bandprocessing identification of the spectral environment. In IEEE 802.22 a frequency range from 41 to 910MHz is specified for CR in a small sub-band overlap. For a more flexible sub-band overlay-system usage. In order to provide reliable licensed configuration, a continuous oscillator tuning could also be usersdetection,theCRterminalfront-endhastofulfillseveral implemented. Generally, digitization of a 50MHz sub-band preconditions. In this paper demands on ADCs in wide- requiresasamplingrateofabout100Msps.Afterwards,digital band scenarios are discussed. As presented in section II, the sensing processing like energy detection or feature detection generalperformanceofADCsischaracterizedbythetrade-off algorithms could be applied. In Fig. 5 the block structure for between supported bandwidth and dynamic range defined by energy detection is depicted. Due to the digital processing the the effective number of bits. Today’s ADCs support sampling singlescantimecanbereducedbyafactorof1/20compared rates up to several Gsps at the expense of low dynamic to analog processing using BR =2.5MHz. range and high power consumption. But, in order to meet Besides decreasing scan time also the proposed sub-band the opposite wish for high dynamic ranges at low power digitization offers the possibility of sensing adjacent channels consumption,theinputbandwidthhastobereduced.Insection ofthecurrentlyallocatedcommunicationchannel.Asdepicted III the different radio services allocated in the considered in Fig. 5 the received signal is used for communications frequency range are analyzed. Due to significantly varying and sensing processing in parallel. This means, all spectral utilization, different demands for sensing in the sub-bands information within the observed sub-band can be collected are observable. Especially sub-bands allocated by broadcast along with current data transmission. Furthermore, the extra servicesdonotofferadditionalradioresourcesbutincreasethe time released due to parallel sensing and communication demands on ADC’s performance significantly. Based on the processing can be used for additional sensing of other sub- sensingalgorithmsdescribedinsectionIV,areceiverforwide- bands. The gained information increases the CR’s knowledge band mobile CRs based on the superheterodyne principle is about its spectral environment. But besides efficient sensing, presentedinsectionV.Duetoadistributioninto20sub-bands also suitable information processing and knowledge storage highly utilized sub-bands can be masked while the resolution has to be applied to CR. In the next subsection an algorithm in under-utilized sub-bands can be increased. is presented that uses the proposed sub-band sensing. REFERENCES C. Information Processing [1] FCC,“Spectrumpolicytaskforcereport,ETDocketNo.02-155,”Tech. Rep.,November2002. Another question in the context of CRs deals with pro- [2] V. Blaschke et al., “Occupation measurements supporting dynamic spectrumallocationforcognitiveradiodesign,”inCrownCom,Orlando cessing of data gained from spectrum sensing. As it was (FL),USA,1.-3.August2007. described by Mitola [3], one important enhancement of CRs [3] J. Mitola, “Cognitive radio - An integrated agent architecture for soft- is the implementation of reasoning algorithms. Reasoning can ware defined radio,” Ph.D. dissertation, Royal Institute of Technology (KTH),Kista,Sweden,2000. be applied for user centric applications such as user inter- [4] S. Haykin, “Cognitive radio: brain-empowered wireless communica- face adaptation or providing user-specific local information. tions,” IEEE Journal on Selected Areas in Communications, vol. 23, no.2,pp.201–220,February2005. Moreover, processing of experiences can also be used for [5] P. Cordier et al., “E2R cognitive pilot channel concept,” IST Mobile optimizing the spectral sensing procedure. As long as the Summit,Mykonos,Greece,June2006. requested service is fulfilled, it is not necessary to change [6] T.Renketal.,“Bio-inspiredalgorithmsfordynamicresourceallocation incognitivewirelessnetworks,”Orlando(FL),USA,August2007. the terminal’s configuration and spectrum allocation. Thus, [7] IEEE,“P802.22:Cognitiveradio,wideregionalareanetwork,”Technical sensing the whole system bandwidth BS is not necessary for Specifications,May2005. [8] R.H. Walden, “Analog-to-digital converter survey and analysis,” IEEE mostofthetime.Scanningthefullbandonlyatalowiteration JournalonSelectedAreasinCommunications,vol.17,no.4,pp.539– rateprovidesageneraloverviewoftheutilizationineachsub- 550,April1999. [9] B. Lee et al., “Analog-to-digital converters,” IEEE Signal Processing band. As long as the currently used sub-band offers enough Magazine,vol.22,no.6,pp.69–77,November2005. unallocatedradioresources,othersub-bandsdonotneedtobe [10] R. Plassche, CMOS Integrated Analog-to-digital and Digital-to-analog observedcontinuously.Iftheutilizationoftheactualsub-band Converters,2ndEd. Boston:KluwerAcademicPublishers,2003. [11] B.Brannon,SoftwareDefinedRadio-EnablingTechnology. London: increasesdetailedinformationaboutavailableothersub-bands JohnWileyandSons,W.Tuttlebee,Ed.,2002,ch.DataConversionin is required. Due to the possibility of the selective sub-band SoftwareDefinedRadios,pp.99–126. [12] P. Leaves et al., “A summary of dynamic spectrum allocation results sensing the scanning procedure can be optimized to bands fromdrive,”inISTMobileSummit,June2002,pp.245–250. that have been offered low utilization in the past. In order to [13] T.A.Weiss,F.Jondral,“Spectrumpooling:aninnovativestrategyforthe get a first overview of a sub-band, the averaged utilization enhancementofspectrumefficiency,”IEEECommunicationsMagazine, vol.42,no.3,pp.8–14,March2004. (cf. (2)) of a single communication channel or a sub-band [14] D.Cabricetal.,“Implementationissuesinspectrumsensingforcogni- could be considered. Due to the digitization of a full sub- tive radios,” in Conference Record of the 38. Asilomar Conference on band Bsub, all communication channels in this band can be [15] CSi.gnRaalus,scShyesrt,emGsruannddlaCgoemnpduetrerSs,pevkotlr.u1m,aNnoavlyesme.berR20o0h4d,ep&p.7S7c2h–w7a7r6z,. observedsimultaneously.Basedonthedetailedsensingresults 2004. 503 Authorized licensed use limited to: CNUDST. Downloaded on January 4, 2009 at 17:23 from IEEE Xplore. Restrictions apply.