6G NTN in C and Q/V-Bands: Link Budget Analysis and Waveforms Performance

6G NTN in C and Q/V-Bands: Link Budget Analysis and Waveforms Performance Carla Amatetti∗, Juan Bucheli†, Dorin Panaitopol‡, Sorya Tong‡, Xavier Leturc‡, Marius Caus§, Musbah Shaat§, Estefania Recayte¶, Mehmet Gurelli∥, Miguel A. Vazquez§, Nathan Borios∗∗, Alessandro Vanelli-Coralli∗ ∗UniBo Italy, †Qualcomm France, ‡TH-SIX France, §CTTC Spain, ¶DLR Germany, ∥Qualcomm USA, ∗∗ TASF France Abstract—Significant progress in the development, standardization, and enhancement of communication systems for 5G Advanced and 6G is set to deliver unprecedented connectivity and performance, unlocking a wide spectrum of vertical services. The full integration of non-terrestrial components into 6G is instrumental in driving this paradigm shift, paving the way for ubiquitous communication and truly global coverage. However, the native unification of terrestrial and non-terrestrial components into 6G brings forth a set of challenges, particularly in radio access technologies (RATs). Notably, among the RAT challenges, defining a waveform optimized for terrestrial and non-terrestrial networks (NTN) is crucial. To this end, in this paper, we present an initial selection of candidate waveforms for the NTN components of 6G and evaluate their performance in terms of Block Error Rate (BLER), Peak to Average Power Ratio (PAPR), and Power Spectral Density (PSD), in both the Q/V and C-bands, accounting for hardware impairments such as phase noise and the nonlinear effects of high-power amplifiers. Additionally, we conduct a comprehensive link budget analysis for both the C and Q/V-bands to assess overall system performance. Index Terms—6G, NTN, waveforms, link budget I. INTRODUCTION With the development of the upcoming generation of communication systems, namely 6G, a unified network is anticipated, through a 3D multilayered and multi-frequency architecture, illustrated in Figure 1, wherein both terrestrial and non-terrestrial network (NTN) elements will seamlessly provide access to users, providing benefits such as an unprecedented level of connectivity and performance, enabling a diverse range of vertical services [1]. Although the NTN inclusion offers various advantages, it also entails several challenges, particularly in radio access technologies (RAT), due to the inherent peculiarities of the satellite channel. Notably, among the RAT challenges, defining a waveform optimized for terrestrial networks (TN) and NTN is crucial [2]. To enable efficient and reliable transmission, extensive research was undertaken to identify the most appropriate waveform among various candidates during the initial standardization efforts for New Radio (NR) communication systems via NTN [3]–[6]. 3GPP Release 17 adopted both the Cyclic Prefix Orthogonal Frequency division Modulation (CP-OFDM) and the Discrete Fourier Transform spread OFDM (DFT-s-OFDM) as waveforms for NR via NTN since they incorporate certain flexibility aspects, which have also been retained in NTN. Looking forward, it is expected that the waveform design for 6G standards will be developed with even greater flexibility to meet the requirements of the different use cases [4] and Fig. 1. 6G-NTN 3D Network Concept [8]. will mitigate some of the drawbacks of CP-OFDM, such as the high Peak to Average Power Ratio (PAPR) [2]. To achieve such flexibility, one of the enablers is the utilization of the whole radio frequency spectrum also in NTN, ranging from frequencies sub-6 GHz, up to terahertz frequency, passing through to lower millimeter wave (mmWave) bands (24-52 GHz) [7]. Waveform design at these frequencies is influenced by hardware and system limitations. Therefore, in this paper, we present an initial selection of candidate waveforms for the NTN components of 6G and evaluate their performance in terms of Block Error Rate (BLER), PAPR, and Power Spectral Density (PSD), in both the C and Q/V-bands, accounting for hardware impairments such as phase noise and the nonlinear effects of high-power amplifiers. Additionally, we conduct a comprehensive link budget analysis for both the C and Q/V-bands to assess overall system performance. The remainder of this paper is structured as follows: Section II delineates the system model and the link budget analysis. Section III introduces the candidate waveforms. Section IV presents the numerical results. Finally, Section V concludes this work. II. SYSTEM MODEL The system model is represented in Figure 1, where we can identify: •The ground segment, where Gon-ground gateways (GWs) provide NTN access to the TNs. Specifically, the GWs establish connectivity between satellites within the constellation, the gNBs, and the Core network. •The non-terrestrial access segment, which includes the NTN nodes in the constellation. The nodes host a regenerative payload. Therefore, the user access link between the satellite and the on-ground UEs implements the traditional Uu air interface. Inter-Satellite Links (ISL) can 2024 IEEE Globecom Workshops (GC Wkshps) | 979-8-3315-0567-7/24/$31.00 ©2024 IEEE | DOI: 10.1109/GCWkshp64532.2024.11101279 Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply. TABLE I Q/V-BAND NTN UE PARAMETERS [12] Parameters NTN VSAT Transmit power 2 W (33 dBm) Antenna type 15 cm eq. aperture diam. (circ. polarization) Antenna gain GT x = 35.2dBi, GRx = 32.9dBi Noise figure 2 dB Output loss 1.5 dB EIRP 36.7 dBW TABLE II Q/V-BAND NTN SAN PARAMETERS [12] Parameters GEO LEO Equivalent antenna aperture DL [m] 2.7 0.27 Equivalent antenna aperture UL [m] 2.1 0.21 Maximum antenna gain (Tx/Rx) [dBi] 60.4 40.4 EIRP density [dBW/MHz] 45 15 3 dB beamwidth [deg] 0.1884 1.884 Beam diameter at nadir [km] 117.7 19.7 Repeater noise figure [dB] 4 4 be employed to guarantee a connection with the satellite and the gateway, in case not all the satellites can establish a feeder link. Regarding the feeder link between the satellite and the GW, it implements the Next Generation (NG) air interface. In terms of coverage, the satellite is able to serve a portion of the on-ground coverage area through a multi-beam antenna. Moreover, the payload is equipped to steer the on-board antenna to always cover the same on-ground area, meaning that the beams are fixed. •User Segment represented by User Equipments (UEs) with handheld or Very Small Antenna Aperture (VSAT) directly connected to an NTN node. A. Link budget Based on the presented system model, a comprehensive user link budget analysis is conducted. The link budget analysis is performed using the NTN channel model (6.6-1) from [9] that takes into account the free space path loss, the shadowing, the atmospheric, and the scintillation loss. The shadowing standard deviation was taken from Table 6.6.2-2 in [9] and its value is thus set to 4 dB. Antenna gains at both the transmitter and receiver side are also taken into account, please see [10] for the detailed equations. The simulation analysis is conducted for both DownLink (DL) and UpLink (UL) scenarios, where DL refers to the transmission from the satellite to the UE, and UL denotes the transmission from the UE to the satellite. Simulations are carried out at a 90° elevation angle for both Q/V-band and C-band frequencies, examining two types of satellite constellations: Geostationary Earth Orbit (GEO) at 35.786 km and Low Earth Orbit (LEO) at 600 km. For the Q/V-band, the DL and UL carrier frequencies are 37 and 47 GHz, respectively. Table I and Table II provide the reference parameters for the NTN UE and the Satellite Access Node (SAN), respectively. The antenna patterns for both the NTN VSAT UE and the SAN are modeled using a Bessel function, as detailed in [11]. For C-band, the DL and UL carrier frequency is considered at 3.5 GHz. Table I and Table II provide the reference parameters for the NTN UE and SAN, respectively. The antenna of the NTN UE is omnidirectional whereas the antenna pattern TABLE III C-BAND NTN UE PARAMETERS [12] Parameters NTN UE Transmit power 200 mW (23 dBm) Antenna type Omnidirectional Antenna gain GT x = 0 dBi, GRx = 0 dBi Noise figure 9 dB TABLE IV C-BAND NTN SAN PARAMETERS [12] Parameters GEO LEO Equivalent antenna aperture DL [m] 12 2 Equivalent antenna aperture UL [m] 12 2 Maximum antenna gain (Tx/Rx) [dBi] 50.5 35 EIRP density [dBW/MHz] 59 34 3 dB beamwidth [deg] 0.4207 2.5247 Beam diameter at nadir [km] 263 26.5 Repeater noise figure [dB] 3 3 of the SAN is modeled with a Bessel function (similar as for Q/V-band) adapted to the wavelength. The link budget is represented in terms of the Cumulative Density Function (CDF) of the Signal to Noise Ratio (SNR). More precisely, 104different topologies have been simulated for each band, each one corresponding to different positions of the NTN UE in the NTN beam, and then the associated SNR is computed for both DL and UL before plotting the resulting CDF. The total bandwidth used in DL is 200 MHz for Q/V-band and 20 MHz for C-band, whereas only 1/10 of the bandwidth is used for UL Q/V-band, and 1/50 for UL Cband. Figure 2 and Figure 3 (resp. Figure 4 and Figure 5) provide the CDF of the SNR for the DL and the UL for the GEO (resp. LEO) orbit of both the Q/V and C-bands. The obtained results indicate that for DL LEO less than 10% of UEs are expecting SNR lower than 4 dB for both Q/V and C-bands, while for DL GEO less than 10% of UEs are expecting SNR lower than -2 dB in Q/V-band and -7 dB in C-band respectively. For UL Q/V-band the results are generally better than UL C-band, indicating a higher throughput in Q/Vband as a result of higher antenna gain and transmission power for the Q/V-band VSAT UE. III. CONSIDERED WAVEFORMS This section provides a concise overview of the selected candidate waveforms, which include CP-OFDM, DFTs-OFDM, Weighted Overlap and Add OFDM (WOLAOFDM), Block-Filtered OFDM, Universal Filtered MultiCarrier (UFMC), Filtered OFDM (F-OFDM), and Orthogonal Time Frequency Space (OTFS). For brevity, the descriptions of CP-OFDM and DFT-s-OFDM are omitted, as these waveforms are well-established. A. WOLA-OFDM CP-OFDM uses a rectangular pulse, which causes signal discontinuities at the boundaries of OFDM symbols. This results in a slow decay of the OFDM spectrum in the frequency domain. WOLA-OFDM addresses this issue by introducing a transmitter windowing procedure that replaces the edges of the rectangular pulse with a smooth function. The OFDM symbol (included the CP) is cyclically extended and windowed by an odd-symmetric window, such as a raised-cosine window, Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply. Fig. 2. CDF of the SNR for the DL of Q/V and C-bands for GEO. Fig. 3. CDF of the SNR for the UL of Q/V and C-bands for GEO. of duration TT X,W OLA. The extended parts of the OFDM symbols overlap among adjacent OFDM symbols, thus reducing the supportable delay spread by the CP. At the receiver side, the CP is removed and the receiving filter, with duration TRX,W OLA, is applied to the Fast Fourier Transform (FFT) window only. Similar to the transmitting filter, the receiving one also reduces the delay spread supported by the CP. B. BF-OFDM BF-OFDM is part of the Filter Bank Multi Carrier (FBMC) family. The transmitter consists of a pre-distortion stage followed by pre-coding and filter bank. Let Mrepresent the number of carriers and Nthe number of subcarriers. Each carrier is allocated Nsubcarriers, but to preserve orthogonality, only N/2subcarriers within each carrier are used to carry data. The precoding scheme is performed through an Inverse FFT (IFFT) of size Nperformed for each carrier, followed by the addition of a CP to maintain the circularity of the received signal. The outputs from all Mstages are then processed through a filter bank, which is parameterized by a prototype filter with an overlap factor K. A predistortion process is applied to each subcarrier to ensure that the receiver can rely on a single FFT, and to compensate for distortions caused by the filter bank, including phase and amplitude distortions. Thanks to the predistortion stage, no filtering is required at the receiver side which can be reduced to a MN/2-points FFT preceded by a CP removal. C. UFMC UFMC divides the transmission band into smaller subbands and applies filtering to each of them. The full band is thus divided into sub-bands, e.g., resource blocks, each one having a fixed number of subcarriers. At each sub-band, an N-points IFFT spanning the full band is relied upon, with Fig. 4. CDF of the SNR for the DL of Q/V and C-bands for LEO. Fig. 5. CDF of the SNR for the UL of Q/V and C-bands for LEO. only the corresponding sub-band allocation present and zero power at all other subcarriers. Following the IFFT stage, an all-zero guard is attached in the time domain, and the resulting sequence is ultimately filtered. Different filters per sub-band can be applied. Sub-band filters may be up-converted versions of the same prototype filter, i.e., shifted versions to place the frequency response at the sub-band centre. At the receiver side, the samples corresponding to the n-th symbol are extracted using a window larger that the FFT size to capture the filterrelated tails. Then, zeros are appended to make the length equal twice the original FFT window size. A 2N-points FFT is applied but only even-indexed outputs are kept. Thus, at the FFT output, the effect of sub-band filtering and multipath appears as a multiplicative channel in the frequency domain which is then estimated and equalized using the classical single tap equalizer. The remainder process is the same as in CPOFDM. D. F-OFDM The F-OFDM transmitter and receiver structures are similar to those of CP-OFDM except for the use of full-band filtering to reduce out-of-band emissions. Compared to UFMC, the length of the filter can be designed longer than the CP, and resulting Inter-Symbol Interference (ISI) further reduced by soft windowing. At the receiver, the signal is matched filtered by the transmit filter, and then processed similarly to CPOFDM. E. OTFS The main characteristic of OTFS stems from encoding the information in the Delay-Doppler (DD) domain rather than in the Time-Frequency (TF) plane. Interestingly, if the channel has a small number of multipath components, then it follows that the channel impulse response on the DD domain Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply. TABLE V SIMULATION PARAMETERS Parameters Scenario Q/V-band C-band Carrier frequency 37 GHz 3.5 GHz Subcarrier Spacing 120kHz 30 kHz # Physical Resource Blocks 32 133 Modulation QPSK Code rate 679/1024 Rv 0 # OFDM symbols 12 # DMRS 1 IBO 2 and 5 [dB] is sparse. The OTFS modulation is divided into two stages. First, a set of symbols in the DD domain are converted to the TF domain through an Inverse Symplectic Finite Fourier Transform (ISFFT). In the second stage, the transmitted signal is obtained from the TF modulated sequence using the Heisenberg transform. Remarkably, a form of Heisenberg transform can be implemented by a multicarrier modulator, which generates the time domain signal that is transmitted over the channel. At the other end of the link, the stages are reversed. Hence, the OTFS demodulator maps the received symbols into the DD domain. First, the received samples are fed into the Wigner transform, and next, the SFFT is applied. The Wigner transform is a generalization of the demodulator that maps the received signal into the modulated symbols on the TF plane and it can be implemented in the form of a multicarrier demodulator. IV. WAVEFORMS SIMULATION RESULTS This section focuses on the performance evaluation of the proposed waveforms. Specifically, simulation results are analyzed to assess the effectiveness of these waveforms based on Key Performance Indicators (KPIs) such as BLER, PAPR, and PSD. It is important to note that this analysis is intended for preliminary comparative purposes only, and thus, receiver optimization is not included in this evaluation. A. Simulation set-up All the waveforms occupy the same resources. The transport block is generated in the same way for all the waveforms: the bits are randomly generated through a bit generator and enter the chain of baseband signal processing blocks including segmentation, cyclic redundancy check (CRC), and scrambling. The Forward Error Correction (FEC) scheme is the Low-Density Parity Check (LDPC) code. After that, the bits are mapped into the constellation symbols, which are then mapped on the different resource elements. Following, the chosen modulation is performed and the generated signal enters the channel block, which consists of the High Power Amplifier (HPA), Tapped Delay Line (TDL), Phase Noise (PN), and additive white Gaussian noise (AWGN). At the receiver side, the reverse path is followed. The simulation parameters are listed in Table V. The following details apply for the implementation of each single waveform: •WOLA-OFDM: a raised cosine with roll-off values equal to 0.03 and 0.008 has been applied at the transmitter and receiver, respectively. -35 -30 -25 -20 -15 -10 Pin [dBm] 15 20 25 30 35 Pout [dBm] -25 -20 -15 -10 -5 0 5 Phase [deg] C band AM/AM AM/PM -20 -15 -10 -5 0 5 10 Pin [dBm] 0 5 10 15 20 25 Pout [dBm] -50 -40 -30 -20 -10 0 Phase [deg] Q/V band AM/AM AM/PM Fig. 6. HPA AM/AM and AM/PM in C-band (above) and in Q/V-band (below). •UFMC: a sub-band size corresponding to the number of subcarriers in a resource block is assumed, along with a Chebyshev filter with 40 dB of side-lobe attenuation. A predistortion stage is added to compensate for in-band distortion induced by the passband filter. •BF-OFDM: a sub-band with an effective usable size corresponding to the number of subcarriers in a resource block is adopted, along with a Gaussian pulse shape of length KM (K= 4, the filter-bank overlapping factor, Mthe number of allocated sub-bands/resource blocks (RBs)). A predistortion stage is added to compensate the distortion induced by the transmit filter. •F-OFDM: a pass-band filter is applied to match the full spectrum allocation. The filter design is based on a truncated sinc of length equal to half the size of the FFT using a Hanning window. An offset of two subcarriers at each extreme is considered so that the filter has a sufficiently flat response over the entire transmission bandwidth. •OTFS: in the C-band, a rectangular pulse shape is utilized, whereas in the Q-band, a Square Root Raised Cosine (SRRC) pulse shape with a roll-off factor of 0.1 is employed. In both cases, the CP is applied on a perblock basis rather than per-symbol. On the receiver side, a Message Passing Algorithm (MPA) detection scheme is implemented. •DFT-s-OFDM: the DFT size corresponds to bandwidth allocation. Before presenting the results, we thoroughly describe the considered impairments. 1) HPA implementation: For the HPA in the Q/V-band and C-band, we used the Amplitude Modulation (AM)/AM and AM/Phase Modulation (PM) characteristics represented in Figure 6, obtained through measurements [13]. Both of them are Solid-State Power Amplifiers (SSPA). To apply the HPA model to the transmitted signal, the input and output of the HPA-specific AM/AM and AM/PM models are scaled so that two conditions are met: i) the resulting model saturates at an output power equal to the average power of the input signal leaving a gap equal to the input backoff (IBO), i.e., Psat = 10 log10(12·NP RB)+IBO, where NP RB corresponds to the number of sub-carriers which, in turn, provides the average power of the signal. ii) The gain of the resulting model Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply. Fig. 7. CCDF of PAPR in Q/V-band (left) and PSD in Q/V-band (right), blue lines: before amplification, black lines: after amplification. corresponds to 0 dB at the linear region of amplification. 2) Phase noise: The adopted phase noise profile (shown in Table VI) is the one proposed in the ETSI TR 103.886 and it is representative of the total aggregated phase noise contribution generated by the main contributors (i.e., the satellite and the terminal) on the forward link. The phase noise is modeled using the filtered Gaussian phase noise approach. This method aims to represent the equivalent power spectral density of the phase noise. Consequently, the actual time-domain model used in simulations is derived from an inverse Fourier transform of the spectral model, assuming the spectrum is conjugate symmetric. TABLE VI PHASE NOISE PROFILE Offset from Carrier frequency PSD [dBc/Hz] 10 Hz NA 100 Hz -25 1000 Hz -50 10 kHz -73 100 kHz -92 1 MHz -102 10 MHz -113 100 MHz -116 B. Numerical results This subsection presents the results of the numerical simulations. 1) PAPR: Extensive simulations were conducted to thoroughly investigate and analyze the distinct characteristics of the PAPR for various candidate waveforms. It is worth mentioning that only one user is considered. The results are graphically illustrated in Figure 7 (left), showing the Complementary Cumulative Distribution Function (CCDF) of PAPR for the different waveforms under investigation. The results are obtained in the Q/V-band since the same behavior is valid in the C-band. Remarkably, the findings reveal that DFTs-OFDM has a reduction of 2.5 - 3 dB across the waveform distribution with respect to the baseline CP-OFDM. The rest of the waveforms exhibit similar values for PAPR with deviations among one another of less than 1 dB. The PAPR of the OTFS is approximately like that of the OFDM-based waveforms, due to the high number of OFDM symbols. It can be observed that the OTFS waveform achieves a PAPR improvement of approximately 0.5 dB over UFMC, at a probability of 10−3. 2) PSD: The PSD is depicted in Figure 7 (right). It is obtained with a IBO of 5 dB in Q/V-band as the behavior is the same in C-band. The plot demonstrates the expected characteristics for the waveforms analyzed prior to amplification (indicated by the blue line). Specifically, the out-ofband emissions for CP-OFDM and DFT-s-OFDM are identical Fig. 8. BLER in C-band. TDL-A with IBO = 2 dB. Fig. 9. BLER in Q/V-band. AWGN with IBO = 2 dB and 5 dB. across the spectrum. For waveforms that are filtered at the resource block level, UFMC and BF-OFDM exhibit similarly low emission levels, approximately 40 dB below CP-OFDM outside the allocated spectrum. In contrast, WOLA and FOFDM display superior out-of-band emission performance, attributable to their inherently long fundamental pulses as implemented in their filters and windowing functions. OTFS (like CP-OFDM) introduces interference to the neighboring bands and, hence, pulse shaping is needed for the OTFS to reduce the out-of-band interference. Conversely, after amplification (represented by the black line), it is observed that the distortion introduced by the power amplifier establishes a lower bound on emissions, beyond which any improvements from waveform design become negligible. 3) BLER: The BLER performance has been evaluated for both the C-band and Q/V-band. For the C-band analysis, a TDL channel model is utilized considering the nonlinearities introduced by the HPA, while for the Q/V-band analysis, the AWGN channel and the hardware impairments have been considered. More precisely, the NTN-TDL-A channel model from [9], offering insights into the observed impact under Non-Line of Sight (NLoS) conditions, has been applied with a delay spread of 50 ns. Referring to Figure 8, OTFS shows superiority relative to other waveforms. The good performance of the OTFS is explained by the fact that the integer part of the delay is considered and the channel is sparse allowing the MPA receiver to properly work. However, this good performance comes at the expense of the high complexity of the receiver, which depends on the number of iterations of the MPA (set to 20), the number of delay and Doppler bins (equal to the number of sub-carriers and OFDM symbols, respectively), the number of paths, and the cardinality of the constellation (i.e., 4) [14]. Regarding the DFT-s-OFDM, its reduced PAPR causes the waveform to suffer from in-band distortion to a lower Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply. Fig. 10. BLER in Q/V-band. AWGN with phase noise. extent in comparison to other waveforms. Conversely, due to the absence of a CP, UFMC is more susceptible to ISI and Inter-Carrier Interference (ICI) caused by channel dispersion and non-linearities introduced by the HPA. The rest of the waveforms line up as they can be observed fundamentally as variants of CP-OFDM filtered at different scales (i.e., at a RB, at the full band, or reducing symbol transitions) for out-ofband emission reduction purposes. It is further demonstrated that the filtering-related reductions in the effective CP do not adversely affect performance up to the delay spread of the evaluated channel model. Figure 9 illustrates the BLER obtained in the Q/V-band in AWGN, considering the effects of the HPA. The dashed lines indicate results with an IBO of 2 dB, while the solid lines represent results with an IBO of 5 dB. Notably, due to its enhanced resistance to power amplifier non-linearities, DFT-sOFDM outperforms the other waveforms. The OTFS performs slightly better than the CP-OFDM, WOLA, BF-OFDM, and F-OFDM. As further observed, such a trend is maintained under different PA backoff values despite the various levels of PA-related compression (which varies depending on the waveform PAPR). Finally, Figure 10 presents the BLER in the Q/V-band, considering both AWGN and phase noise. Phase noise has a detrimental effect on the performance of all waveforms, due to an increased inter-carrier interference and rotation of the constellation, resulting in a 0.5 dB reduction for a value of the BLER equal to 0.1 in the case of DFTs-OFDM, compared to the results shown in Figure 9. OTFS demonstrates greater resistance to phase noise effects because the transformation in the time domain distributes and mixes symbols in time, breaking the correlated nature of the phase noise [15]. Regarding the other waveforms, they perform similarly under the influence of such an impairment. As in the previous case, only UFMC degrades slightly w.r.t other variants, effectively observing a detriment of less than 1 dB. V. CONCLUSION This paper investigated a 6G NTN system operating in the C-band and Q/V-band. We conducted a comprehensive UL and DL link budget analysis for both frequency bands and identified and evaluated a preliminary list of candidate waveforms for the system. Their performance was assessed under conditions of multipath and hardware impairments, including phase noise and the nonlinear effects of the HPA. The preliminary results indicate that OFDM-based waveforms and OTFS demonstrate nearly identical performance in terms of BLER in AWGN conditions. Filtered OFDM variants, including WOLA, UFMC, BF-OFDM, and F-OFDM, exhibit comparable out-of-band emissions following the amplification stage. In contrast, CP-OFDM and DFT-s-OFDM alternatives produce higher emissions due to their dependence on sharp rectangular pulses. When considering the effects of the HPA with varying IBO, all the waveforms show similar out-of-band emissions. In terms of BLER, DFT-s-OFDM shows slightly less degradation in performance related to HPA effects, owing to its lower PAPR. Additionally, the increased computational complexity of UFMC does not appear to be justified by the improvements observed in the KPIs, compared to the lower complexity of BF-OFDM. The OTFS excels in scenarios involving multipath and phase noise. This enhanced performance comes with the trade-off of the increased complexity of the receiver. Future studies foresee the assessment of additional KPIs to provide a more comprehensive comparison of the waveforms, particularly in multi-user scenarios. ACKNOWLEDGEMENTS This work has been funded by the 6G-NTN project, which received funding from the SNS JU under the European Union’s Horizon Europe research and innovation program under Grant Agreement No 101096479. The views expressed are those of the authors and do not necessarily represent the project. The Commission is not liable for any use that may be made of any of the information contained therein. REFERENCES [1] X. Lin et al., “On the path to 6G: Embracing the next wave of low Earth orbit satellite access,” IEEE Communications Magazine, vol. 59, no. 12, pp. 36–42, 2021. [2] H. Shahid et al., “Emerging Advancements in 6G NTN Radio Access Technologies: An Overview,” in 2024 EuCNC/6G Summit, 2024, pp. 593–598. [3] N. Cassiau and B. Chamaillard, “Assessment of multicarrier waveforms for 5G satellite,” in 2017 IEEE PIMRC. IEEE, 2017, pp. 1–5. [4] A. Jayaprakash et al., “New radio numerology and waveform evaluation for satellite integration into 5G terrestrial network,” in ICC 2020. IEEE, 2020, pp. 1–7. [5] ——, “Analysis of candidate waveforms for integrated satelliteterrestrial 5G systems,” in 2019 IEEE 2nd 5G World Forum (5GWF). IEEE, 2019, pp. 636–641. [6] P. Kim et al., “Performance Analysis for 5G/6G Satellite Communication under Nonlinear HPA Channel,” in 2022 27th Asia Pacific Conference on Communications (APCC). IEEE, 2022, pp. 641–642. [7] M. Sarajli´ cet al., “Waveforms for sub-THz 6G: Design guidelines,” in 2023 EuCNC/6G Summit. IEEE, 2023, pp. 168–173. [8] 6G-NTN project, “Report on 3D multi layered NTN architecture D3.5 V1.0.” [9] 3GPP, “TR 38.811 “Study on New Radio (NR) to support non-terrestrial networks (Release 15)”,” September 2020. [10] M. Conti, et al., “NB-IoT over Non-Terrestrial Networks: Link Budget Analysis,” in GLOBECOM 2020, 2020, pp. 1–6. [11] 3GPP, “TR 38.821 “Solutions for NR to support non-terrestrial networks (NTN) (Release 16)”,” May 2021. [12] 6G-NTN project, “Spectrum regulation analysis for 6G NTN scenarios D4.7 V1.0.” [13] ——, “Report on unified and data-driven air-interface for 6G-NTN, deliverable D4.1 V1.0.” [14] P. Raviteja et al., “Interference cancellation and iterative detection for orthogonal time frequency space modulation,” IEEE transactions on wireless communications, vol. 17, no. 10, pp. 6501–6515, 2018. [15] Y. Bello et al., “OTFS waveform with phase noise in sub-THz,” in 2022 IEEE VTC2022-Fall. IEEE, 2022, pp. 1–5. Authorized licensed use limited to: Politecnico di Milano. Downloaded on December 08,2025 at 18:08:07 UTC from IEEE Xplore. Restrictions apply.