You are on page 1of 8

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 56, NO.

5, MAY 2009

951

Accurate Characterization of TWTA Distortion in Multicarrier Operation by Means of a Correlation-Based Method


Marinella Aloisio, Member, IEEE, Piero Angeletti, Member, IEEE, Enrico Casini, Member, IEEE, Enrico Colzi, Salvatore DAddio, Member, IEEE, and Roger Oliva-Balague

AbstractComplex payload architectures, together with the use of high-order digital modulations and advanced coding schemes, call for an accurate evaluation of traveling-wave tube amplier (TWTA) nonlinearity impairments. To assess in a simple yet accurate manner the level of signal distortion generated by the TWTA nonlinearity, we adopt a correlation-based methodology. The proposed method provides an accurate evaluation of the TWTA intermodulation, particularly in the presence of multicarrier operation with nonconstant envelope modulations. Index TermsCommunications satellites, high-order modulations, intermodulation (IM) distortion, noise power ratio (NPR), output back-off (OBO), traveling-wave tube (TWT).

I. INTRODUCTION FFICIENT payload exploitation is a primary objective in satellite-communication systems in order to cope with limited power and spectral resources. In particular, travelingwave tube amplier (TWTA) parameters [e.g., saturated power, operating input/output back-off (IBO/OBO)] must be optimized according to the actual operational conditions and considering the severe impact that, in such circumstances, the amplier nonlinearity can have on the communication link performance. In addition, the bandwidth availability at Ka/Q-band frequencies, together with multibeam satellite system architectures, favors multicarrier (MC) operation of the onboard TWTAs, which reduces the number of ampliers while preserving overall high throughput. The sensitivity to real operational conditions increases with the adoption of modern satellite-communication standards, which foresee a wide set of different coding and modulations schemes [1], [2]. TWTA optimization must balance two contrasting effects: On the one hand, a drive level close to saturation is desirable to maximize dc-power efciency, and on the other hand,

Manuscript received October 14, 2008. Current version published April 22, 2009. The review of this paper was arranged by Editor W. Menninger. M. Aloisio, P. Angeletti, E. Colzi, S. DAddio, and R. Oliva-Balague are with the European Space Agency, European Space Research and Technology Centre (ESA/ESTEC), 2201 AZ Noordwijk, The Netherlands (e-mail: marinella.aloisio@esa.int; piero.angeletti@esa.int; enrico.colzi@esa. int; salvatore.daddio@esa.int; roger.oliva.balague@esa.int). E. Casini is with NII Communications Infrastructure Services, NATO C3 Agency, 2597 AK Den Haag, The Netherlands (e-mail: enrico.casini@nc3a. nato.int). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TED.2009.2016128

operating the TWTA close to saturation produces a high level of intermodulation (IM) distortions, particularly in MC operation. IM distortions can be associated with the TWTA amplitude-toamplitude (AM/AM) and amplitude-to-phase (AM/PM) transfer characteristics and cause signicant degradation of the end-to-end link performance with respect to the ideal case of a linear additive-white-Gaussian-noise (AWGN) channel. Back-off operation improves linearity but at a cost of reduced power efciency and higher onboard power consumption. Therefore, for a target link performance, the optimum TWTA operating point shall be identied by trading off the onboard power consumption with the TWTA nonlinearity effects, taking into account possible MC operation and different high-order modulations. A very detailed modulator, payload, and demodulator simulation can be realized to assess the end-to-end link performance of a satellite payload, but this would certainly be very complex and time-consuming, particularly due to the latest innovations in physical-layer techniques [3]. Accurate and efcient simulation techniques have proven to allow the assessing of different operational conditions and the performing of linearity versus efciency tradeoffs in a reasonable amount of time [4][6]. With these objectives in mind, we will describe in detail the simulation methodology adopted in [6], which is based on a time-domain correlation method. Such a methodology allows the analyzing of the impact of a nonideal payload on the end-to-end link performance. In Section II, we briey review techniques for assessing nonlinearity impairments, and in Section III, we describe the cross-correlation technique used to evaluate the carrier-to-interference ratio, which is the general parameter that will be used to evaluate the payload/TWTA nonlinearity performance. The modeling methodology has been applied to the TWTA characterization and optimization, and the implemented simulator is described in Section IV. Examples illustrating the application to various second-generation digitalvideo-broadcasting (DVB-S2) modulation modes [1], and different MC scenarios are provided in Section V. A conclusion follows. II. MC MODELING ASPECTS The TWTA nonlinearity can be described in several manners, each with different complexity and level of accuracy. The simplest nonlinearity parameters are dened by measurements performed with a single unmodulated tone [i.e., continuous

0018-9383/$25.00 2009 IEEE

952

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 56, NO. 5, MAY 2009

wave (CW)] as the input test signal. The TWTA-input power is varied, and the output power and phase variation are measured to provide the classical single-tone AM/AM and AM/PM transfer characteristics. In principle, these parameters do not directly model IM among signals fed into the same amplier. However, an MC signal may be modeled in terms of the instantaneous amplitude and phase of a single tone while the effects of the amplier may be modeled as time-varying gain and phase blocks. An equivalent MC AM/AM curve may then be computed based on average input and output powers. Complex input signalenvelope statistics may also be used to predict MC AM/AM. However, the validity of this approach for the study of MCsignal distortions is limited to the knowledge of the overall MC-signal power loss while little insight can be gained on the distortions produced on each component of the MC signal. A simple, albeit effective, linearity parameter directly related to the TWTA IM products is the carrier-to-third-order-IM ratio (C/IM3), obtained by feeding the amplier with two unmodulated tones and measuring, at its output, the amplitude difference between the useful carrier and the corresponding IM3 product [7], [8]. Limitations of the applicability of this simplied two-tone measurement method and the corresponding parameter are evident, considering the nature and the number of signals present at the TWTA input in a typical MC scenario with high-order modulations. On the other extreme, the noise-power-ratio (NPR) linearity parameter corresponds to the carrier-to-IM ratio (C/IM) when the TWTA is operated with WGN [9], [10]. The NPR is measured by using a notch lter to remove a single frequency slot, known as a quit channel, from the WGN at the TWTA input. At the TWTA output, a spectrum analyzer measures the level of noise generated by the nonlinearity in the quit channel, which represents the IM power. The notch position can be changed in order to sweep the entire MC band. Both C/IM3 and NPR are dened at the device level and with given standard test signals (two tone or WGN) and can therefore be easily measured by the amplier manufacturer regardless of the specic operational scenario of the unit under test. On the other hand, the use of standard test signals implies that the IM-product estimate is generally not very representative of actual end-system performance. This is particularly true when the amplier is operated in an end system with one or more digitally modulated carriers as is the case for the latest satellitecommunication systems implementing advanced payload architectures (e.g., multiport power ampliers, exible TWTAs, etc.) and exploiting enhanced bandwidth-efcient modulation schemes. Furthermore, the estimate of the IM power provided by C/IM3 does not allow the direct evaluation of the in-band interference while that based on NPR is too pessimistic [4]. To overcome this limitation, Aloisio et al. [4] propose a more accurate method to evaluate the level of IM products in a realistic scenario which is based on the so-called NPRn , where the subscript n is used in order to distinguish this parameter from the standard NPR, calculated with WGN. In fact, in NPRn calculations, the standard WGN test signal is replaced by the actual signal waveform comprised of n modulated carriers. A band-reject impulse response lter creates a deep notch at the center frequency of one of the modulated carriers, and the

resulting signal is fed into the TWTA. The depth of the notch observed in the signal at the TWTA output is called NPRn , and for a high number of input modulated carriers (n > 8), it coincides with the standard NPR. As discussed in [4], the NPRn provides a much more accurate estimate of the IM products, when compared to standard NPR, due to the presence at the TWTA input of the actual modulated MC signal. However, for NPRn simulations, the TWTA-input signal is slightly modied by the deep notch in one of the modulated carriers. Although this perturbation of the input signal has only a second-order effect on the accuracy of the calculated IM level, in the next section, we propose a more precise way of calculating the carrier-to-interference ratio based on a correlation method. III. CORRELATION METHOD For current or advanced payload architectures, due to the complex operations usually carried out, all the assumptions valid for C/IM3 and NPR, i.e., measurements are performed at the device level with only the test signals present at the TWTA input, are no longer valid. In particular, MC operation has become the usual practice, and therefore, the payload and amplier nonlinearity distortions must be measured in test conditions as close as possible to reality in order to accurately predict the end-system performance. The objective of the correlation method is to dene a nonperturbative way to estimate the impact of the amplier nonlinearity on the end-to-end performance of a satellite link. While, in this paper, we limit the analysis to the case of a single TWTA operating with several input signals multiplexed in frequency, the extension of the method to more complex payload architectures is cumbersome but trivial [11]. The proposed measurement method exploits the properties of the correlation operation between two signals. If a correlator is driven with two equal sequences, its output shows a maximum value when the two sequences are aligned. Contrarily, if the two sequences are not the same and/or not aligned, the correlator output shows a very low value; the longer the correlation period, the lower such a value. This basic principle is employed to estimate the useful signal power, as well as the interference, which in the following analyses is considered to be limited to the IM distortion caused by the TWTA nonlinearity. When a single-carrier or MC signal is amplied by a nonlinear amplier, the signal at the output shows a signicant component that is represented by the amplied version of the input signal as well as an additional extra distortion/interference component produced by the amplier nonlinearity. By correlating the amplier output signal with a unitary power replica of the input signal, it is possible to estimate the useful power of the signal at the amplier output without directly measuring the IM products. In other words, only the amplier output signal component that is correlated to the transmitted one contributes to the correct detection of the carried information in the maximum-likelihood sense [12]. This thesis was initially proven in [13] for linear systems and extended in [14] and [15] to nonlinear systems and payload analyses [11]. In turn, the interfering power due to the IM distortion can be easily derived by subtracting the power

ALOISIO et al.: ACCURATE CHARACTERIZATION OF TWTA DISTORTION IN MULTICARRIER OPERATION

953

Fig. 1.

Simulation-implementation scheme.

estimated with the correlation method (useful power) from the total received power in the band of interest. The main advantage of the proposed method when compared to standard C/IM3 or NPR calculations is that it is capable of estimating the power of IM products when the TWTA is operated with actual payload signals, hence, accounting for all possible nonlinear effects. In addition, this method is capable of retrieving every single signal component from the whole amplied signal without affecting or modifying the signal operational conguration, as done, for example, for NPRn calculations. In other words, with the correlation method, the input signal is not affected by any artifacts/perturbations due to the measurement conditions. This allows the operating of the amplier with an exact replica of the representative driving signal, therefore generating the actual impairment on the signal itself. Finally, the accuracy of the proposed method can be properly tuned by dening the correlation time length. The proposed method is comparable to the error-vectormagnitude (EVM) technique, which is based on the instantaneous difference between the measured modulated signal and its ideal noise-free reference version [16], [17]. The major difference stands in the fact that the EVM is incapable of distinguishing cofrequential modulated signals while the proposed method provides this feature through correlation, making it particularly attractive for complex payload measurements (e.g., in far-eld analyses of performance of active antennas generating multiple beams reusing the same frequency [11]). By dening the signal-to-noise-plus-interference ratio (SNIR) as the ratio of the useful signal power (C ) to total in-band noise and distortion (I ) and assuming that the useful signal is not correlated to the noise and distortion terms, according to Blachmans expression [18], we derive C/I = C/ (P C ), where P is the total in-band power. By indicating with x and s the output and the input modulated signals, respectively, the SNIR estimator becomes
xs C C Ps = = 2 ( ) I P C Px Rxs Ps

input signals, respectively. The longer the correlation interval, the higher the rejection of the interference contribution and, consequently, the higher the accuracy of the signal power estimator. Before illustrating the set of simulation results, there is the need to dene the accuracy that such correlation-based estimation provides (a detailed theoretical elaboration of the achievable estimator accuracy in terms of the standard deviation of the C/I measure can be found in [11], [13][15]). We simply recall that the standard deviation of the C/I measure decreases as 1/ L, where L is the number of correlation samples. Similarly, the C/I standard deviation increases when the interference level is lower (i.e., C/I is higher) as expected due to the more difcult estimation of the interfering power. For the scope of this paper, it is sufcient to say that the estimated values have an accuracy of 0.5 dB. IV. SIMULATOR IMPLEMENTATION For the architecture shown in Fig. 1, we have performed simulations in the time domain with baseband complex representation. The signal generator module generates N uncorrelated binary data sequences and modulates them by applying DVB-S2-like modulation schemes, e.g., quadrature phaseshift keying (QPSK), 8-symbol PSK (8-PSK), and 16- and 32-symbol amplitude and PSK (16- and 32-APSK) [1]. No time delay or phase shift is applied among the modulated signals. Each modulated signal is then frequency-converted and ltered with a square-root-raised-cosine (SRRC) lter to avoid intersymbol interference. All the signals are up-converted so as to be uniformly spaced in the frequency domain, separated by a typical guard-band. The resulting N signals are multiplexed and passed through the nonlinear amplier model described by the AM/AM and AM/PM complex transfer characteristics. The signal bandwidth is assumed to be always smaller than the transponder bandwidth and, therefore, a memory-less TWTA model can be adopted. This simplifying assumption does not limit the validity of the described methodology. The distorted output signal is then passed through the SRRC lter matched, for instance, to the k th signal. Therefore, the output signal is composed of a distorted replica of the k th input signal plus the additional in-band component of adjacent channel interference and in-band IM products. In order to evaluate the ratio of the useful power to the interfering power at the demodulator,

R2 ( )

(1)

) represents the empirical crosswhere the term Rxs ( correlation function among the output signal x and the input signal s evaluated at , which is the estimated delay of the system. Px and Ps represent the power of the ltered output and

954

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 56, NO. 5, MAY 2009

Fig. 2. Single-tone LTWTA normalized amplitude and phase transfer characteristics.

the output signal is cross-correlated with the expected ideal replica of the k th signal. Computational time and complexity are reduced if the generated sequence is split into shorter subsequences on which the cross correlation and the power estimation are performed. The nal power estimation is carried out by averaging the power values obtained from the single subsequences. V. MC DVB-S2 RESULTS In this section, we present an accurate TWTA nonlinearity characterization. The results are obtained by applying the correlation-based methodology to a typical linearized TWTA (LTWTA). Several operating conditions are analyzed, from single-carrier to MC operation, and with DVB-S2-compliant modulation schemes (QPSK, 8-PSK, 16-APSK, and 32-APSK). At rst, the amplier nonlinearity is characterized by calculating the output power loss of the TWTA when driven with single-carrier or MC nonconstant-envelope modulated signals, with respect to nominal CW operation. High-order modulations, as dened for the DVB-S2 standard, are characterized by a nonconstant instantaneous signal envelope, and the overall input signal envelope depends on the actual number of carriers, the modulation scheme, and the shaping-lter characteristics. Due to the typical compression shape of the TWTA AM/AM and AM/PM curves, the overall envelope time-varying amplitude causes a reduction of the average amplier output power with respect to single-carrier CW operation. In the following, we will use input and output power quantities normalized CW SAT and to CW input and output power at saturation (Pin CW SAT , respectively). By generalizing the CW back-off dePout nition [19], the IBO is dened as the ratio between the CW input power at saturation and the actual operating input power, expressed in decibels as
CW SAT IBO[dB] = Pin [dB] Pin [dB].

Fig. 3. QPSK OBO mod curves for different numbers of carriers as compared to the standard CW OBO curve.

(2)

Similarly, the OBO can be dened as


CW SAT OBO[dB] = Pout [dB] Pout [dB].

(3)

In order to distinguish it from the standard CW OBO, we introduce the term OBO mod (modulated OBO in [4]) which

refers to the OBO, as dened in (3), when the TWTA is driven by one or more digitally modulated signals at a given normalized input power [IBO as dened in (2)]. Clearly, OBO mod depends on the modulation scheme, the number of carriers, the shaping lter, and the IBO. Numerical simulations have been performed in order to analyze the impact of nonconstant envelope modulations on the OBO mod for different modulations and numbers of carriers, considering a typical communications LTWTA modeled by means of its single-tone (CW) amplitude and phase transfer characteristic (Fig. 2). Figs. 36 show a comparison of the standard CW IBOOBO curve with the IBOOBO mod characteristics in MC operation for one to eight carriers and, respectively, QPSK, 8-PSK, 16APSK, and 32-APSK modulation schemes. All the modulated carriers have a symbol rate of 45 Msps. A rolloff factor of 0.25 has been used for the SRRC lters. All the signals are equally spaced in frequency with a typical guard-band of 6.25 MHz. Each OBO mod curve corresponds to the total RF power, i.e., useful plus interference power, measured at the TWTA output. The OBO mod curves shown in Fig. 3 for 1- to 8-QPSK modulated carriers conrm that, even in the case of low-order modulation schemes and in single-carrier operation, degradation in output power of about 0.5 dB can be observed at typical operational IBOs. This is explained by the presence of the shaping lters along the end-to-end chain, which modify the

ALOISIO et al.: ACCURATE CHARACTERIZATION OF TWTA DISTORTION IN MULTICARRIER OPERATION

955

Fig. 4. 8-PSK OBO mod curves for different numbers of carriers as compared to the standard CW OBO curve.

Fig. 7. Estimated C/I curves as a function of IBO for QPSK modulated carriers, along with the standard NPR and two-tone C/IM3 curves.

2) The OBO mod curves of the different modulation schemes converge to the same value as the number of input carriers increases, i.e., the modulation has a lesser impact on the OBO mod when the TWTA is driven with MCs (this is further highlighted in Fig. 11). This is expected because of the following reasons: 1) the complex envelopes of single-carrier higher order modulated signals are characterized by a signicantly wider probability density function (pdf) than QPSK and 8-PSK, and 2) when many signals are summed together, no matter the modulation, the pdfs of the resulting MC complex envelope signal tend to assume the same shape, approximating that of a Rayleigh distribution. By employing the correlation method, we have also assessed the level of IM distortion generated by the TWTA in MC operation and with realistic modulation schemes (QPSK, 8-PSK, 16-APSK, and 32-APSK). The results are shown Fig. 710, in terms of C/IM power (C/I ) curves as a function of IBO. In the case of MC operation with more than two carriers, a higher level of IM is expected on the carriers internally positioned as compared to the external ones; the C/I values we report are those corresponding to the worst case (i.e., the most internal carriers). Fig. 7 shows the analysis for MC operation with QPSK modulation and shows a comparison of it with standard NPR and (two tone) C/IM3 calculations. As expected, a signicant C/I degradation occurs when the number of carriers increases from one to eight. As an example, at saturation, the NPR value is about 8 dB lower than the two-carrier C/I and about 2 dB lower than the eight-carrier C/I . Therefore, as expected, the NPR is proven to overestimate the level of IM of an LTWTA operating under realistic conditions (not with WGN). An interesting result shown in Fig. 7 is related to the comparison of the calculated C/I with C/IM3: The C/IM3 curve is lower than both the calculated two- and one-carrier C/I , i.e., C/IM3 actually overestimates the level of IM products in these two cases while, as expected, it underestimates the IM with 4- and 8-QPSK modulated carriers. Similar results are shown in Fig. 8 for 8-PSK modulated carriers. When the number of carriers is increased from one to eight, the C/I curves derived for

Fig. 5. 16-APSK OBO mod curves for different numbers of carriers as compared to the standard CW OBO curve.

Fig. 6. 32-APSK OBO mod curves for different numbers of carriers as compared to the standard CW OBO curve.

signal envelope (see Fig. 1). In addition, Fig. 3 shows that, as the number of input carriers increases, the OBO mod curves tend to coincide. The MC operation produces an additional degradation of up to 1 dB with respect to a 1-QPSK modulated carrier and a degradation of over 1.5 dB as compared to the standard CW OBO. Similar results are shown in Fig. 4 for the 8-PSK modulation scheme. A comparison of Figs. 36 shows the following features. 1) The difference between the single-carrier and the MC OBO mod curves decreases with the modulation order.

956

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 56, NO. 5, MAY 2009

Fig. 8. Estimated C/I curves as a function of IBO for 8-PSK modulated carriers, along with the standard NPR and two-tone C/IM3 curves.

Fig. 11.

OBO mod as a function of the number of carriers at IBO = 4 dB.

Fig. 9. Estimated C/I curves as a function of IBO for 16-PSK modulated carriers, along with the standard NPR and two-tone C/IM3 curves.

Fig. 10. Estimated C/I curves as a function of IBO for 32-PSK modulated carriers, along with the standard NPR and two-tone C/IM3 curves.

high-order modulations (16- and 32-APSK) show a different behavior when compared to those for QPSK and 8-PSK modulations. In fact, as shown in Fig. 9 for 16-APSK modulation, at saturation, C/I decreases by 5 dB when the number of carriers is increased from one to eight against the 8 dB in the QPSK and 8-PSK cases (Figs. 7 and 8, respectively). Fig. 10 shows that 32-APSK has the lowest relative C/I degradation

due to MC operation (e.g., 2 dB at saturation). This can be explained by the complexity of the signal envelope for highorder modulated signals. For the same reason, in this case, the standard C/IM3 curve provides a more optimistic estimate (lower values) of the IM products. The predicted C/I can be used to estimate the link degradation of the end-to-end system without resorting to extensive simulations. Let us consider a typical modulation and coding scheme for broadcast applications such as the 8-PSK with a low-parity density-check code rate 2/3 currently employed in the transmission of HD-TV via DVB-S2 satellite broadcast systems. Such a scheme achieves a BER of 105 when C/N is around 6.7 dB, and the channel is linear (see [1], [3]). When a nonlinear channel is considered, the actual IM interference must be taken into account in the computation of the overall link budget. This will produce a degradation on the end-to-end SNIR which is more signicant for higher levels of interference (i.e., lower C/I ). In order to achieve the same BER performance, the additional interference must be compensated by a higher C/N ratio. This required C/N ratio can be predicted either on the basis of the C/I estimations (Fig. 8) or by means of direct BER evaluation [4]. As an example, when the LTWTA is operated at 4-dB IBO, for a two- and four-carrier scenario, our method predicts C/N values of 6.8 and 7.2 dB, respectively; while C/N values of 6.8 and 7.3 dB would be required, respectively, to obtain the target BER by direct error counting. Fig. 11 shows a comparison of the OBO mod calculated for QPSK, 8-PSK, 16-APSK, and 32-APSK modulations, for a number of carriers varying from one to eight and at a xed IBO of 4 dB. In single-carrier operation, quasi-constant envelope modulation schemes such as QPSK and 8-PSK present identical OBO mod values (1.4 dB), while an evident output power degradation appears for high-order modulation schemes (16and 32-APSK), particularly in single-carrier operation. For a higher number of carriers, the low- and high-order modulation OBO mod values converge. In addition, Fig. 11 shows the discrepancy that results from calculating the OBO with the standard single-tone procedure instead of the proposed correlation-based methodology: At 4-dB IBO, differences range from 0.5 dB in the case of 1-QPSK/8-PSK modulated carrier to 1.4 dB for eight carriers (regardless of the specic modulation scheme).

ALOISIO et al.: ACCURATE CHARACTERIZATION OF TWTA DISTORTION IN MULTICARRIER OPERATION

957

Fig. 12. C/I as a function of the number of carriers at IBO = 4 dB.

We have also analyzed the variation of C/I with the number of carriers for different modulations and, again, at a xed IBO of 4 dB; the results are shown in Fig. 12. A C/I degradation of almost 6 dB is observed for QPSK/8-PSK modulations against the 2 dB corresponding to 32-APSK. With eight carriers, the difference among the C/I values corresponding to the various modulations tends to be signicantly reduced (0.5 dB maximum). The C/I curves are also compared with the standard NPR, which is calculated with WGN and, therefore, does not depend on the number of carriers. Fig. 12 shows that the standard NPR overestimates the IM level of eight modulated carriers by 1.7 dB while, for four modulated carriers, the overestimation can be between 2 and 3 dB, depending on the actual modulation scheme. A signicant discrepancy of about 8 dB is observed between NPR and the correlation-based C/I value in the case of 2-QPSK/8-PSK modulated carriers, due to the quasi-constant envelope of these modulations; a difference of 3.5 dB is still observed with two 32-APSK modulated carriers. The latter results prove the importance of relying on an accurate evaluation of the IM, as provided by the C/I calculated with the correlation method. VI. CONCLUSION In this paper, we have presented an accurate analysis of typical LTWTA performance under realistic MC operation. First, we discussed possible ways of characterizing the TWTA nonlinearity, each with different complexity and level of accuracy; then, we described the methodology proposed, which is based on the correlation method. This methodology can be applied to a number of practical problems in satellitecommunication systems and has the major advantage of not introducing any perturbation/manipulation on the TWTA-input signal and, therefore, represents a nonperturbative way to accurately estimate the impact of the amplier nonlinearity on the end-to-end performance of a satellite link. At rst, we characterized the TWTA with respect to the total output power loss due to nonconstant envelope input signal(s) for different DVB-S2-compliant modulation schemes and numbers of carriers. The analysis has shown that if the TWTA is driven by a single carrier, then different modulations

lead to different output power degradations. However, when the number of input carriers increases, the output power loss becomes less dependent on the specic modulation scheme and on the actual number of carriers. This can be explained by the tendency of an MC signal to resemble a Rayleigh-distributed complex envelope signal, independent of the type of modulation, when the number of carriers is increased. The modulated OBO has also been compared with the standard OBO. In addition, we performed a precise evaluation of the TWTA C/IM by adopting power estimation with the correlation method. The aggregate impact of high-order modulation schemes, number of carriers, and shaping lters has been analyzed. The C/I curves have been calculated for QPSK, 8-PSK, 16-APSK, and 32-APSK modulation schemes and considering MC operations up to eight carriers. The analysis conrms that, for a low number of input carriers, high-order modulations, such as 16- and 32-APSK, result in C/I values which signicantly differ from those of more conventional modulations (e.g., QPSK). In MC operation, the difference is considerably reduced. The most noteworthy differences between C/I and the standard NPR have also been shown, highlighting how C/I compares reasonably well to NPR in MC operation but NPR signicantly overestimates the IM level when operating with one or two carriers, for any modulations. When the overall (end-to-end) satellite link is considered, the C/I curves calculated by means of the correlation-based methodology can be used to accurately evaluate the carrierto-noise-plus-IM ratio, C/(N + I), which allows for tradeoffs between TWTA output power and linearity and, hence, for the determination of the optimal TWTA RF operating point (IBO). The results presented in this paper conrm the need to account for the actual number of carriers, modulations, and coding schemes when comparing TWTA performance within the endto-end satellite link. ACKNOWLEDGMENT The authors would like to thank R. De Gaudenzi for the initial idea on the cochannel measurement methodology and for the continuous encouragement on the subject. They would also like to thank F. Coromina for his continuous support. R EFERENCES
[1] Digital Video Broadcasting; Second Generation Framing Structure, Channel Coding and Modulation Systems for Broadcasting, Interactive Services, News Gathering and Other Broadband Satellite Applications, v1.1.1 (2005-03). ETSI EN 302 307, 2004. [2] NATO standardization agreement, Super High Frequency (SHF) Military Satellite Communications (MILSATCOM) Frequency Division Multiple Access (FDMA) Non-EPM Modem for Services Conforming to CLASS-B of STANAG 4484. STANAG 4486 Ed 3, DRAFT C. [3] E. Casini, A. Ginesi, and R. De Gaudenzi, DVB-S2 modem algorithms design and performance over typical satellite channels, Int. J. Satell. Commun. Netw., vol. 22, no. 3, pp. 281318, 2004. [4] M. Aloisio, E. Casini, and A. Ginesi, Evolution of space traveling-wave tube amplier requirements and specications for modern communication satellites, IEEE Trans. Electron Devices, vol. 54, no. 7, pp. 15871596, Jul. 2007. [5] M. Aloisio, P. Angeletti, E. Colzi, S. DAddio, R. Oliva-Balague, E. Casini, and F. Coromina, End-to-end performance evaluation methodology for TWTA-based satellite exible payloads, in Proc. IVEC, 2008, pp. 175176.

958

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 56, NO. 5, MAY 2009

[6] M. Aloisio, E. Casini, P. Angeletti, R. Oliva-Balague, E. Colzi, and S. DAddio, Accurate characterization of TWTA distortion in multicarrier operation by means of a correlation-based method, in Proc. IVEC, 2008, pp. 409410. [7] A. S. Gilmour, Jr., Principles of Travelling Wave Tubes. Norwood, MA: Artech House. [8] D. M. Goebel, R. R. Liou, W. L. Menninger, X. Zhai, and E. A. Adler, Development of linear traveling wave tubes for telecommunications applications, IEEE Trans. Electron Devices, vol. 48, no. 1, pp. 7481, Jan. 2001. [9] G. Maral and M. Bousquet, Satellite Communication Systems. New York: Wiley, 2002. [10] J. G. Proakis, Digital Communications, 4th ed. New York: McGraw-Hill. [11] E. Braunschwig, E. Casini, and P. Angeletti, Co-channel signal power measurement methodology in a communication and payload joint simulator, presented at the 24th AIAA International Communications Satellite Systems Conf., San Diego, CA, 2006, AIAA 2006-5351. [12] R. M. Fano, Transmission of Information: A Statistical Theory of Communications. Cambridge, MA: MIT Press, 1961. [13] W. H. Tranter and M. D. Turner, Signal-to-noise ratio estimation in digital computer simulation of low-pass and bandpass systems with applications to analog and digital communications, Univ. MissouriRolla, Rolla, MO, Tech. Rep. CSR-77-6, Jul. 1977. Final Report, Contract NAS 9-14848. [14] M. C. Jeruchim and R. J. Wolfe, Estimation of the Signal-to-Noise Ratio (SNR) in communication simulation, in Proc. GLOBECOM, Nov. 2730, 1989, vol. 2, pp. 12741278. [15] M. S. Rae, J. L. Fernandez, and K. S. Shanmugan, Simulation-based estimation of intermodulation distortion and C/IM, in Proc. IEEE GLOBECOM, Dec. 69, 1992, vol. 2, pp. 700706. [16] K. Voelker, Apply error vector measurements in communications design, Microw. RF , vol. 34, no. 17, pp. 143152, Dec. 1995. [17] R. Hassum, M. Flaherty, R. Matreci, and M. Taylor, Effective evaluation of link quality using error vector magnitude technique, in Proc. IEEE Wireless Commun. Conf., 1997, pp. 8994. [18] N. M. Blachman, Noise and Its Effect on Communication. New York: McGraw-Hill, 1966. [19] ITU, Handbook on Satellite Communications, 3rd ed., Wiley, 2002.

Enrico Casini (M09) received the Laurea degree (summa cum laude) from the University of Florence, Florence, Italy, in 2002. In 2001, he was with the European Space Research and Technology Centre, European Space Agency, Noordwijk, The Netherlands. From 2001 to 2003, he was a Trainee investigating advanced synchronization and interference mitigation techniques for space communications. From 2003 to 2007, he was Communication System Engineer with the RF Payload and System Division, where he focused his work on the characterization of nonlinear satellite channel, advanced modem design, and communication systems simulation. Since January 2008, he has been with an area team of the NII Communications Infrastructure Services, NATO C3 Agency, The Hague, The Netherlands, where he is responsible for the design and prototype of deployable communication systems.

Enrico Colzi received the Laurea degree (summa cum laude) in electronic engineering from the University of Florence, Florence, Italy, in 1992 and the Ph.D. degree in technical sciences from the Technical University of Delft, Delft, The Netherlands, in 1999. From 1992 to 2000, he was a Microwave Systems Engineer with the European Space Agency (ESA), where he worked on advanced communication and Earth-observation system concepts and techniques. From 2000 to 2001, he was a Project Manager for a commercial company (TV Files) providing multimedia distribution services to business users. From 2001 to 2006, he was with the Italian Space Agency, where he was rst a Platform Manager of a satellite mission (AGILE) and then a Project Manager for different telecom projects. Since 2006, he has been with ESA, where he is working for Vega Group PLC as a Consultant Senior RF Payload Engineer in the RF Payload Systems Division of the European Space Research and Technology Centre, Noordwijk, The Netherlands. His current research interests include study and design of advanced telecommunication and navigation-satellite payloads.

Marinella Aloisio (M07) received the M.S. degree (summa cum laude) in electronic engineering and the Ph.D. degree in physical engineering from the University of Catania, Catania, Italy, in 2002 and 2007, respectively. In 2001, she was an Internal Student with the European Space Agency (ESA), European Space Research and Technology Centre (ESTEC), Noordwijk, The Netherlands, where her work involved the analysis and computational modeling of traveling-wavetube (TWTs) collectors. Upon graduation, she joined ESA/ESTEC as a Microwave Engineer with the RF Payload Systems Division, where she is currently in charge of the development and operation of vacuum electronics for space applications. Her current research interests include the analysis and design of slow-wave structures and the characterization of satcom TWTs performance. Dr. Aloisio is a member of the IEEE Electron Devices Society Technical Committee on Vacuum Devices.

Salvatore DAddio (M07) received the Laurea degree (summa cum laude) in telecommunications engineering from the University of Naples, Naples, Italy, in 2005 and the M.S. degree in satellites and orbiting platforms from the University of Rome La Sapienza, Rome, Italy, in 2006. He is currently a Member with the technical staff of the European Space Agency, European Space Research and Technology Centre (ESA/ESTEC), RF Payload Systems Division, Noordwijk, The Netherlands. From 2004 to 2006, he was involved in the radio-frequency testing of 3G UMTS mobile terminals and in optoelectronics sensors design and development. His current research interests include analysis, modeling, and design of advanced telecommunication and navigationsatellite payloads and, in addition, Earth-observation radio-frequency active and passive instruments.

Piero Angeletti (M07) received the Laurea degree (summa cum laude) in electronic engineering from the University of Ancona, Ancona, Italy, in 1996. From 1997 to 2004, he was involved in aerospace systems engineering activities, joining in succession the following companies in Italy: Augusta Helicopters, Alenia Spazio, Elettronica, and Space Engineering. He is currently a Member with the Technical Staff of the European Space Agency (ESA), European Space Research and Technology Centre (ESTEC), Noordwijk, The Netherlands. He published more than 100 papers in international journals and conferences. His current research interests are in the eld of analysis, modeling, and design of telecommunication and navigation-satellite systems and payloads.

Roger Oliva-Balague received the M.Sc. degree in telecommunications engineering from the Catalonian Polytechnic University, Barcelona, Spain, in 2004 and the D.E.A. degree in astronomical science from Barcelona University, Barcelona, in 2006. Since 2006, he has been with the European Space Agency, European Space Research and Technology Centre (ESA/ESTEC), RF Payload Systems Division, Noordwijk, The Netherlands, where he worked, under sponsorship by the Spanish Trainee Grant, on the analysis of advanced satellite telecom payloads and on the development of specic simulation tools for this purpose and, currently, is working in the Earth Observation Directorate on the SMOS mission.

You might also like