Analysis of the use of CSK for future GNSS signals (797922), страница 7
Текст из файла (страница 7)
Indeed, once thereceiver is synchronized with the incoming signal, it isimportant that all the correlator outputs are synchronizedwith actual PRN code shifts. This means that the samplesof the correlation function fall exactly on multiples of achip. This means that the sampling frequency needs to bea multiple of the chipping rate, which is known to be nonoptimal for synchronization purpose [24], especially whenthe Doppler frequency is close to 0 Hz. For reducing thecomputational burden, the same constraint actuallyapplies to traditional correlation computation since typicalhardware receiver generate local replicas of the PRN codebased on shifts that are multiples of the sampling time.However, it seems easier to release this constraint onhardware receiver using traditional correlators than forhardware receiver using FFT-based correlators.VII.D.
Impact of CSK on AcquisitionIn this subsection, a comparison between the jointdata/pilot acquisition method (only applicable to BPSKsignals) and the pilot-only acquisition method (applicableto both modulations) is made in order to inspect thedegradation on the acquisition sensitivity introduced by aCSK modulation.
Both methods have the same frequencyand code delay uncertainty [2], and use the sameacquisition detector: the standard single dwell acquisitiontechnique described in [25].Erreur ! Source du renvoi introuvable. shows thesignal characteristics and acquisition parameters of thisanalysis. Table IX and Table X show the acquisitionthreshold (total signal C/N0) for the joint data/pilotacquisition method and for the pilot-only acquisitionmethod with different data/pilot power share. From thesetables, it can be see that the pilot-only acquisition methodneeds to have 75% of the power allocated to the pilotcomponent in order to have the same acquisitionperformance as the joint data/pilot acquisition method.
Ifonly 50% of the power is allocated, there is a degradationof acquisition sensitivity of 2 dB.Table IX: Acquisition Thresholds (Total Data+Pilot C/N0 of theSignal) for Different Data/Pilot Power Share, Coherent IntegrationTime and Non-Coherent Summations for a Targeted False AlarmProbability equal to the Inverse of the PRN Length (1023 Chips)and a Targeted Detection Probability of 90%Acquisition Technique (dB-Hz)DwellCoherentTime on 1Pilot- orPilot- orIntegrationacquisitionData-onlyData-onlyData+Pilottime (ms)bin (ms)with 75%with 50%34.6536.434.41030.253230.250128.4530.228.510024.626.324.750032.7534.532.21027.6529.427.550525.7527.525.610021.4523.221.550032.0533.831.51026.7528.526.4501024.6526.424.410020.252220.2500Table X: Acquisition Thresholds (Total Data+Pilot C/N0 of theSignal) for Different Data/Pilot Power Share, Coherent IntegrationTime and Non-Coherent Summations for a Targeted False AlarmProbability equal to the Inverse of the PRN Length (10230 Chips)and a Targeted Detection Probability of 90%Acquisition Technique (dB-Hz)DwellCoherentTime on 1Pilot- orPilot- orIntegrationacquisitionData-onlyData-onlyData+Pilottime (ms)bin (ms)with 75%with 50%35.253735.21030.9532.730.950129.1530.929.110025.1526.925.350033.6535.433.11028.4530.228.250526.4528.226.310022.1523.922.250033.0534.832.41027.6529.427.2501025.4527.225.210020.9522.720.9500Therefore, it can be concluded that for acquisitionpurposes, the BPSK data demodulation seems moreappropriate as it enables the possibility of a jointdata/pilot acquisition.
However, in case a 75%/25%pilot/data power split is used, BPSK and CSK signalswould have similar performances.carrier phase tracking, the implemented discriminators arethe Product discriminator (P) and the arctangentdiscriminator (Atan), which compared to the pilotdiscriminators, have the following disadvantages:narrower linearity region (more sensitive to large trackingerrors) and stability points every π (demodulated datacould be sign-reversed).Therefore, [26] show that although the joint data/pilotmethod can use all the available signal power [27] and forthe pilot-only method the data component power is lost,the pilot-only method provides better carrier phasetracking performance in poor environments: indeed thedata component discriminators start providing erroneousoutputs due to its reduced linear region while the pilotcomponent's discriminator actually performs correctly.
Inthis situation, the combination of the discriminatoroutputs will provide erroneous values. This can lead tocycle slips or loss of lock, while the tracking based on thepilot component only would have performed correctly.Therefore, it can be concluded that there is nodegradation of carrier phase tracking performancebetween a CSK signal and a BPSK signal since the BPSKwill probably track the signal carrier phase using a pilotonly method. Finally, it has to be reminded that CSKmodulation allows a non-coherent demodulation whereasa BPSK demodulation does not.VII.F.Impact of CSK on carrier frequency trackingCarrier frequency tracking is typically done using aFrequency Lock Loop (FLL).
Typical FLL discriminatorsare based on the measure of the carrier phase variationduring one coherent correlation. As a consequence, itsystematically uses correlator outputs of the current andprevious integration intervals. As a consequence, FLLdiscriminators are always sensitive to data bit transitions.It is possible to create FLL discriminators that areinsensitive to data bit transition, but this is in generaldetrimental to the performance of the FLL (lowersensitivity, more susceptibility to high frequency errors)[26].For the carrier phase tracking process, the maindifference between the joint pilot/data carrier phasetracking method and the pilot-only carrier phase methodis the type of discriminators of the PLL which can beimplemented [2].Another important feature of the FLL discriminator isits linearity region that will define the FLL pull-in range.This pull-in range is inversely proportional to the coherentintegration time.
It is thus in general desirable to have thepossibility to use short coherent integration for thetransition from acquisition to tracking when the frequencyuncertainty is high, and then to use long coherentintegration in order to have more accurate frequencytracking [26].For a pilot-only carrier phase tracking method, theimplemented discriminators are generally either the fourquadrant arctangent (Atan2) or the quadrature-phasecorrelator output (Q) [26].
The advantages of thesediscriminators are the appearance of stable points every2π and a wider linear region. For the data componentTherefore, it can be concluded that it is better to useonly the pilot component to track the carrier frequency ofthe signal since this components provides a largerintegration time that can be conducted over more than onerelatively short PRN code, and since this component doesnot carry data.VII.E. Impact of CSK on carrier phase trackingIn conclusion, there is no degradation of the carrierfrequency tracking performance when a CSK modulatedsignal is used instead of a BPSK modulated signal.VII.G.
Impact of CSK on code delay trackingAs opposite to a joint pilot/data carrier phasetracking method, a joint pilot/data code delay trackingmethod can implement the same discriminators on thedata component as on the pilot component. The reason isthat the mostly used discriminators are non-coherent, thusremoving the BPSK data [26].However, the use of the same discriminator onboth components implies that the correlator outputs of thedata and pilot components have to be output at the sametime, and thus that the coherent correlation duration is thesame on both components. Therefore, although thediscriminators are non-coherent, the coherent integrationmust still be restricted over a data symbol.Moreover, the code delay tracking processadmits longer coherent integration than the carrier phasetracking process. And it is well known that long coherentintegrations improve the code tracking jitter, filter moreslowly varying multipath and interference and potentiallyenables the use of the secondary code properties [26].Therefore, although a joint/data code delaytracking method can use the same discriminator for bothcomponents and can use the entire signal power, an onlypilot code delay tracking method will provide betterperformance [26] since it could implement longercoherent integrations.VIII.CONCLUSIONSThis paper has further analyzed the introduction of aCSK modulation on a GNSS signal than pervious works[13][14] and has clearly highlighted the advantages ofsuch modulation:the non-coherent demodulation,the possibility to increase the bit rate withoutmodifying the symbol rate, chip rate or PRNcodes properties andthe flexibility of dynamically changing the signalbit rate.Moreover, the drawbacks have been identified: thenecessity of introducing a pilot channel to synchronize thesignal since the synchronization process cannot beconducted on the data channel and the increase of thecomplexity of the receiver demodulator block.Afterwards, this paper has proposed different optionsof pair channel codes – decoding methods and hasevaluated them with a proposed methodology fordesigning CSK modulated signals which pursue twoopposite objectives: keeping the same useful bit rate as areference BPSK signal and increasing the useful bit ratewith respect to a reference BPSK signal while keeping thesame symbol rate.
This paper has shown the trade-offbetween demodulation performance, receiver complexityand codeword duration of the different pairs.Moreover, the analysis of the two previous CSKdesigned signals and all the determined particular signalscharacteristics could be combined in a dynamic signalwhich could change its useful bit rate from low(ephemeris, clock error corrections) to high (newservices) and thus be better adapted to the nature of itsbroadcasted information.Finally, the analysis of the impact of a CSK modulatedsignal on a GNSS receiver has shown that although aFFT-based demodulator (radix-2) reduces the complexityof the receiver, this kind of receiver raises an issue withthe required sampling frequency: a FFT-demodulatorrequires a sampling frequency proportional to thechipping rate whereas this kind of sampling frequency hasshown to be detrimental for the tracking block.Moreover, this paper has shown that there is noacquisition performance degradation if at least 75% of thepower is allocated to the pilot channel, and that thereshould not be carrier or code delay tracking performancedegradation since pilot-only tracking methods outperformjoint pilot/data methods.IX.FUTURE WORKOn-going work is centered on determining thedemodulation performance of a CSK signal in a urbanenvironment (mobile channel).
The demodulationperformance is being analyzed for the four pairs “channelcodes –decoding methods” presented in this paper and fora non-coherent demodulation.Besides, future work will analyze the limitations on thenumber of bits mapped by a CSK symbol, U, of a CSKmodulation and will inspect the best CSK configurations(U, N) for a urban environment since the long symbolscan be undesired due to the fading of these types ofenvironments.Finally, the concern about the frequency sampling for aFFT-based type of demodulator will be addressed.ACKNOWLEDGMENTSThis work was developed on the framework of aresearch project financed by CNES and ENAC and wasunder the management of TéSA (Telecommunications forSpace and Aeronautics). Special thanks to PhilippePaimblanc and Marie-Laure Boucheret for theircontribution to this work.REFERENCES[1] B.W.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
