The present invention relates to telecommunications, and more specifically to a method to support for compensation of time domain phase variations. [Background Art] [0002] Phase variations in the time domain can be caused by various phenomena. For example the presence of phase noise, frequency drifts due to Doppler shift or insufficient frequency synchronization can produce phase variations in the time domain. [Summary of Invention] [Technical Problem] [0003] Orthogonal Frequency Division Multiplexing (OFDM) systems appear to be quite sensitive to phase noise. For instance, it can be noticed that OFDM systems are more sensitive to phase noise than single-carrier systems. That is part of the reason that systems being subject to important phase noise environment avoid the use of OFDM waveforms. Satellite systems are an example of systems that are particularly subject to challenging phase noise environments. [0004] Time domain effects appear to be easier to monitor and compensate in the time domain rather than in the frequency domain. Time domain training sequences are recognized to be particularly effective for phase noise compensation and carrier offset compensation. [0005] Discrete Fourier Transformation spread Orthogonal Frequency Division Multiplexing (DFTsOFDM) waveform corresponds to blocks of bit interleaved and coded modulation symbols that are pre-coded by a Discrete Fourier Transformation (DFT). The number of modulation symbols in the block corresponds to the DFT size and to the number of active subcarriers. Then subcarrier mapping is performed so that an Inverse Discrete Fourier Transform (IDFT) can be performed as well. For example, blocks of M bit interleaved and coded modulation symbols are precoded by an M sized DFT, mapped to M out of N subcarriers and then passed through an N sized IDFT. A cyclic prefix (CP) can be optionally appended after the IDFT. In that case, it is referred as cyclic prefix Discrete Fourier Transformation spread Orthogonal Frequency Division Multiplexing (CP DFTsOFDM). CP DFTsOFDM has been used as the uplink (UL) waveform in Long-Term-Evolution/Long-Term-Evolution Advanced (LTE/LTE-A) systems. CP DFTsOFDM will be used as one of the UL waveforms in 5G New Radio (5G NR) systems. One of its fundamental properties is the low peak to average power ratio (PAPR) allowing a user equipment (UE) to use its high power amplifier (HDPA) in an energy efficient manner near its saturation point. [0006] Modulation symbols obtained from a single codeword may spread several DFTsOFDM blocks, generally contained within a scheduling unit. A scheduling unit that can be for example a slot, a minislot in NR 5G or a subframe in LTE-LTE-A is formed by blocks containing reference signal for demodulation and blocks not containing reference signal for demodulation. There is a need of sequences serving as support in order to compensate phase variations in the time domain for millimeter (mm) wave systems that are subject to relatively important phase noise environment. Such sequences can also be used for compensating residual Carrier Frequency Offset (CFO) or effects due to Doppler shifts. For example, such sequences can compensate high velocity in low dispersive channels. For OFDM systems, it is known to insert a sequence at the subcarrier level (in the frequency domain). The advantage of such sequence inserted in the frequency domain is that it provides support for relatively simple estimation of phase variations during frequency-domain processing at the receiver side. However, the inconvenient is that frequency domain observation of a time domain effect cannot be performed with a finer granularity than one OFDM symbol. [0008] Regarding DFTsOFDM, reference signal for phase tracking can also be inserted in the frequency domain at subcarrier level. In such cases, reference signal for phase tracking can either be inserted onto carriers non-occupied by data or can puncture occupied subcarriers. Both insertions types of reference signal for phase tracking for DFTsOFDM, in the frequency domain, lead to degraded PAPR. Another drawback of inserting reference signal for phase tracking in the frequency domain onto carriers non-occupied by data is that DFTs of different sizes need to be implemented. Another drawback of inserting reference signal for phase tracking in the frequency domain by puncturing occupied subcarriers, is that it can lead to a decrease of demodulation performance. [0009] As a consequence, there is a need to explore further reference signal for phase tracking insertion for DFTsOFDM. The present invention aims to improve the situation. [Solution to Problem] [0011] The invention relates to a method implemented by computer means of a communicating entity, for inserting a reference signal for phase tracking, said communicating entity using a discrete Fourier transformation spread orthogonal frequency division multiplexing modulator. In particular, the method comprises: - obtaining a succession of signal samples by inserting said reference signal for phase tracking within a succession of data samples, according to at least one insertion pattern chosen among pre-defmed patterns with respect to predetermined criteria of communication conditions, and - feeding said modulator with a succession of signal blocks obtained from said succession of signal samples, so as to apply the discrete Fourier transformation after the insertion of the reference signal for phase tracking. [0012] It is meant by "pattern" here above a predetermined combination of positions intended to be occupied by reference signal for phase tracking in the aforesaid succession of signal samples, and the insertion of the reference signal for phase tracking is then performed in the time domain, prior to the DFT application. [0013] It is meant by "data samples" a succession of samples in the time domain including the bit interleaved coded modulation symbols to be transferred during the transmission. For simplicity, it is also included here any other reference signal samples other than reference signals for phase tracking if included before the DFT, when configured (e.g. mobility reference signals, reference signals for fine time/frequency tracking of the channel, additional demodulation reference symbols, etc.). "Signal samples" are obtained after insertion of the reference signal for phase tracking within the data samples succession, still in the time domain. A succession of "Signal blocks" are then obtained from that signal sample succession and are applied to the DFTsOFDM modulator. Usually, the DFTsOFDM modulator is implemented in the frequency domain and it implements at least a discrete Fourier transformation, then a mapping, then an inverse discrete Fourier transformation. Equivalent time-domain implementations of the DFTsOFDM modulation also exist but are usually not used in practical implementations due to a lack of flexibility and a higher complexity. [0014] Therefore, support for time domain phase error compensation is provided thanks to reference signal for phase tracking insertion according to an insertion partem. Moreover, advantageously the choice of pattern among predefined patterns is performed according to predetermined criteria of communication conditions. Thus, the pattern that suits best the communication conditions can be selected. [0015] In an embodiment, one of the predetermined criteria of communication conditions is at least one of the following: - scheduled bandwidth; - modulation; - coding rate; - carrier frequency; - physical resource block bundling; or - other reference signal densities. Therefore, implicit or low signaling pattern insertion choice can be performed. [0016] In an embodiment, a reference signal for demodulation is further inserted within a transformed succession of signal blocks which results, in the frequency domain, at least from the application of the discrete Fourier transformation to the succession of signal blocks. This reference signal for demodulation is inserted in scheduled positions of the transformed succession of signal blocks, but however the applied pattern for reference signal for phase tracking insertion takes into account these scheduled positions. [0017] Therefore, since reference signal for demodulation can serve as support for phase error compensation, taking into account reference signal for demodulation position enables not to insert reference signal for phase tracking closely to reference signal for demodulation so it limits the penalty in maximum attainable throughput. [0018] In an embodiment, transformed succession of signal blocks includes Ndata transformed signal blocks, each transformed signal block being mapped to M active subcarriers. The chosen pattern can be defined therefore by Nk groups of Kj reference signal for phase tracking separated by D; data samples, such as: Nk-\ Nk ;=0 /=0 where Dj are non-null positive integers when i is different from 0 and Nk. [0019] In such an embodiment, controlling the values of K;, enables controlling, at the receiver side, the accuracy of the phase error estimate based on the group of Ki reference signal for phase tracking. Controlling the values of D; enables further to control the accuracy of the interpolation between groups of reference signal for phase tracking at the receiver side. In an embodiment: A - ®NK - 0. In an embodiment: Dx =D2 =.... = DNK_X . Therefore, the time domain distance between two groups of reference signal for phase tracking is equal and can be set according to a given phase noise variation degree. [0021] In an embodiment: KQ = KX = ....= KNK_1 . Therefore, estimation on each group of reference signal for phase tracking is equally reliable. [0022] In an alternative embodiment: Kx =.... = KNK_2 =K0+ KNK_] . [0023] In an embodiment: (D0 + £?=0(^i + Di+1)) mod M = D0 mod M , for each k from 0 to Nk_2. In this embodiment, reference signal for phase tracking group starts at the same relative position with respect to the beginning of the signal block in each signal block where the sequence is present and thus allows easier de-mapping implementation. [0024] In an embodiment, (Xf=0(£i + K{)) mod M = (Z)0 + K0) mod M for each k from 0 to NK-i- Therefore, the reference signal for phase tracking group ends at the same relative position with respect to the beginning of the signal block for easier de-mapping implementation. [0025] In an embodiment: Kt + Di+1 = KNk_t + DNk + D0 = A for each k from 0 to Nk„2, where either: - A = M ; or - A < M ; or In this embodiment, if AM , reference signal for phase tracking overhead Ktot/MNdata can be reduced by less frequent insertion of reference signal for phase tracking groups, where Ktot = Xjio ^i- FmaUy> if A = M, equally reliable phase estimation can be performed in each signal block. [0026] In an embodiment, a non-null cyclic prefix, CP, is inserted after the inverse discrete Fourier transformation which is carried out by the modulator after the discrete Fourier transformation, and here D; (with i=l...NK-i) can take one of two values: - a chosen integer D', if D; does not span across two signal blocks, or - D"=D'- E(NCp*M/N), if Dj spans across two signal blocks (where NCp corresponds to a number of CP samples appended after the inverse discrete Fourier transformation, and E(x) designates the closest integer to x, N corresponding to the number of subcarriers related to the inverse discrete Fourier transformation). In a variant, E(x) designates the closest integer inferior to x. In another variant, E(x) designates the closest integer superior to x. Therefore, regular insertion is obtained after CP insertion and it enables time domain processing before discrete Fourier transformation at the receiver side. [0027] In an embodiment, Dj can be chosen so as to avoid reference signal for phase tracking insertion in signal blocks having positions scheduled for reference signal for demodulation or for other reference signal allowing phase tracking (in the frequency domain, or possibly in the time domain according to the present invention). [0028] Another aspect of the invention relates to a telecommunication device, using a discrete Fourier transformation spread orthogonal frequency division multiplexing modulator, the device comprising a computer circuit for inserting a reference signal for phase tracking and more particularly for: - obtaining a succession of signal samples by inserting said reference signal for phase tracking within a succession of data samples, according to at least one insertion pattern chosen among pre-defined patterns with respect to predetermined criteria of communication conditions, and - feeding said modulator with a succession of signal blocks obtained from said succession of signal samples, so as to apply the discrete Fourier transformation after the insertion of the reference signal for phase tracking. [0029] A third aspect of the invention relates to a computer program product, comprising instructions for performing the method previously described, when run by a processor. [0030] Other features and advantages of the method and device disclosed herein will become apparent from the following description of non-limiting embodiments, with reference to the appended drawings. [0031] The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings, in which like reference numerals refer to similar elements. [Brief Description of the Drawings] [0032] [FIG. 1] FIG. 1 is a flow chart illustrating the different steps of the reference signal for phase tracking insertion method. [FIG. 2] FIG. 2 is an example of a representation of reference signal for phase tracking insertion. [FIG. 3] FIG. 3 is an example of an insertion pattern. [FIG. 4] FIG. 4 is an example of an insertion pattern. [FIG. 5] FIG. 5 is an example of an insertion pattern. [FIG. 6] FIG. 6 is an example of an insertion pattern. [FIG. 7] FIG. 7 is an example of an insertion pattern. [FIG. 8] FIG. 8 is an example of an insertion pattern. [FIG. 9] FIG. 9 is an example of an insertion pattern. [FIG. 10] FIG. 10 is a possible embodiment for a device that enables the method of the present invention. [FIG. 11] FIG. 11 shows examples of an alternative parameter set to describe a pattern sequence. [FIG. 12] FIG. 12 shows different embodiments where the pattern insertion is chosen by a base station while the insertion of the reference signal for phase tracking is performed by a terminal UE, in the context of an uplink communication. [FIG. 13] FIG. 13 shows different embodiments where the pattern insertion is chosen by a base station while the insertion of the reference signal for phase tracking is performed by a terminal UE, in the context of an uplink communication. [FIG. 14] FIG. 14 shows different embodiments where the pattern insertion is chosen by a base station while the insertion of the reference signal for phase tracking is performed by a terminal UE, in the context of an uplink communication. [Description of Embodiments] [0033] FIG. 1 is a flow chart illustrating the different steps of the reference signal for a phase tracking insertion method according to a possible embodiment of the present invention. [0034] A data samples succession that represents a set of data DATA_S 101 may be modified by inserting therein a reference signal for phase tracking. That insertion follows a chosen pattern selected among a collection of patterns (patterns and pattern combinations as presented below). The selection of a particular pattern can be performed according to communication criteria COMCRIT 102. Communication criteria COM_CRIT 102 may enable thus the choice of an insertion pattern INS_PAT 103 for reference signal for phase tracking. Therefore, data sample may be modified into signal sample SIG_S 104 when reference signal for phase tracking is inserted according to the chosen insertion pattern PAT. Signal blocks can be obtained from signal sample 104 in order to feed a discrete Fourier transformation spread orthogonal frequency division multiplexing modulator FEED_MOD 105 with said signal blocks, the modulation staring with a discrete Fourier transformation DFT 106. [0035] The choice of insertion pattern PAT is based on communication criteria COM_CRIT 102 such as the allocation size M (or equivalentfy the scheduled telecommunication bandwidth), the modulation type, the coding rate, the carrier frequency, the PRB (physical resource block) bundling, and/or other reference signal densities. Furthermore, a combination of preexisting patterns can be chosen as well. [0036] FIG. 2 is an example of a representation of reference signal for phase tracking insertion according to a possible embodiment of the present invention. More particularly, FIG. 2 represents an insertion of K reference signal for phase tracking symbols in one discrete Fourier transformation spread orthogonal frequency division multiplexing symbol carrying reference signal for phase tracking. [0037] In FIG. 2, reference signal for phase tracking RS_Phase Track 201 and data samples Data 202 are fed to a reference signal for phase tracking insertion module RS_PhaseTrack PAT 203, which is inserting said reference signal for phase tracking within the succession of data samples, according to at least one insertion pattern. [0038] A scheduling unit such as a minislot, slot or other unit is considered. The scheduling unit contains Nsym blocks among which Ndata signal blocks do not contain any reference signal for demodulation. Each of the Ndata signal blocks will be mapped to M active carriers. In module 204, Ndata signal blocks of M signal samples can be created. Each block is therefore M sized hereafter. Module 205 can apply then a discrete Fourier transformation to the Ndata signal blocks of M signal samples. Subcarrier mapping of Ndataand NRSDemod blocks can be performed in module 206 before application of an inverse discrete Fourier transformation IDFT 207. NRSj3emod blocks correspond to blocks comprising reference signals for demodulation. [0039] In FIG. 2, the reference signal for demodulation RS_Demod positions are represented by full pilots in module 206. However hybrid data/pilot symbols is another possible alternative, with reference signal for demodulation insertion before or after application of a discrete Fourier transformation 205 (DFT). Cyclic prefix (CP) may be appended after the inverse discrete Fourier transformation 207. [0040] Reference signal for phase tracking insertion can offer support for time-domain phase error compensation. However, inserting reference signal for phase tracking may bring a penalty in terms of maximum attainable throughput. [0041] For this reason, reference signal for phase tracking density should not exceed the minimum density necessary for attaining a performance target. Since other reference signals such as reference signal for demodulation, or other reference signals, when configured (e.g. mobility reference signals, reference signals for fine time/frequency tracking of the channel, additional demodulation reference symbols, etc.) can be used as support for phase error compensation, there may not always be a need to insert reference signal for phase tracking. [0042] FIG. 3 is an example of an insertion pattern according to a possible embodiment of the present invention. The pattern is composed of Nk groups of Ki reference signal for phase tracking separated by Di data sample, such as: Nt-\ Nt where Di are non-null positive integers when i is different from 0 and Nk. [0043] The number of inserted reference signal for phase tracking is Ktot = ft T -J Hi Jo ^i for a total reference signal for phase tracking overhead of Ktot/ MNdata-[0044] When applying a discrete Fourier transformation, data samples and reference signal for phase tracking are spread together. Due to multipath channel, interference may exist between data sample and reference signal for phase tracking at the receiver side. [0045] For a given reference signal for phase tracking overhead, small values of K; may allow finer granularity of reference signal for phase tracking insertion (small Dj), profitable at the receiver side to interpolation among reference signal for phase tracking groups. [0046] However, phase estimation on each reference signal for phase tracking group may be degraded due to interference with data samples. In that case, the minimum value of Ki may be superior to a threshold so that reliable averaging is possible. [0047] Large reference signal for phase tracking groups may allow reliable phase estimation within a group with limited noise/interference. However, if the reference signal for phase tracking overhead may be kept reasonable then groups of reference signal for phase tracking may be rather spaced and it may result in large Dj values. In that case, interpolation between reference signal for phase tracking groups may be less reliable. [0048] A compromise between minimum K; and Maximum Dj, where i goes from 0 to Nk-1 under constraint of overhead Ktot/ MNdata can be implemented. [0049] Other insertion pattern can be defined, for example with K0 =KX =.... = KNK_X. This pattern may allow equally reliable estimation on each group of reference signal for phase tracking. [0050] Another possible insertion pattern can be defined with *, =.... = KNK_2 =K0+ KNK_, and A = DNK = 0 • A variant of insertion pattern is a cyclic shift of above insertion patterns. [0051] Another possible insertion pattern can be defined with the following formula: (D0 + £f=0(/Q + A+i)) m°d M = D0 mod M where k goes from 0 to Nk„2 or any cyclic shift of this pattern. Therefore, the reference signal for phase tracking group starts at the same relative position with respect to the beginning of the signal block in each signal block where the reference signal for phase tracking is present. Easier de-mapping implementation is thus allowed. This pattern can be combined with other features, for example: - NK=Ndata, thus one reference signal for phase tracking group can be present in every data block, - Kj can depend on the distance to the closest position containing other reference signal that may be used for phase estimation, - K, =.... = KNK_2 =K0+ KNK_I or K0=K^.... = KNK_, . (£iLo(Di + Kj)) mod M = (D0 + K0) mod M, where k goes from 0 to Nk_i can define another possible insertion pattern. Any cyclic shift of this pattern is also a possible insertion pattern. Therefore, the reference signal for phase tracking group ends at the same relative position with respect to the beginning of the signal block for easier de-mapping implementation. This pattern can be combined with other features, for example: - Nx^Ndata^ one reference signal for phase tracking group can be present in every data block, - Ki can depend on the distance to the closest position containing other reference signal that can serve for phase estimation, and - Kx =.... = KNK_2 =K0+ KNK_X or K0=KX =.... = KNK_X . [0053] Equally spaced insertion patterns can be defined with the following formula D, = D2 =.... = DNK_X =D. Any cyclic shift of this insertion pattern may also be a possibility. Therefore, the time domain distance between two reference signal for phase tracking groups may be equal and may be set according to the phase noise strength. [0054] Regular insertion pattern may be defined with the following formula Ki + Di+i = KNk--t + DNk + DQ = A where k goes from 0 to Nk.2. A variant of this pattern may be determined with K0=KX =.... = KNK_X. Different cases may be distinguished depending on the value of A. FIG. 4 is an example of an insertion pattern according to a possible embodiment of the present invention. In FIG. 4, A = M, therefore equally reliable phase estimation may be performed within each data symbol. [0055] FIG. 6 is an example of an insertion pattern according to a possible embodiment of the present invention. In FIG. 6, &M, and reference signal for phase tracking overhead can be reduced. [0057] Irregular insertion pattern can be implemented also when a non-null cyclic prefix is to be inserted after the inverse discrete Fourier transformation. The insertion pattern can be adapted to that case as well and Dj, i=l...NK-i can take one of two values: - a chosen integer D', if D; does not span across two signal blocks, or - D"=D'- E(Ncp*M/N), if D; spans across two signal blocks, where NCp corresponds to a number of CP samples appended after the inverse discrete Fourier transformation, and E(x) designates the closest integer to x, N corresponding to the number of subcarriers related to the inverse discrete Fourier transformation. In a variant, E(x) designates the closest integer inferior to x. In another variant, E(x) designates the closest integer superior to x. [0058] FIG. 7 shows an example of irregular insertion pattern since reference signal for phase tracking are not contained here within each M sized signal block. Thus, overhead is reduced. [0059] FIG. 5 is an example of an insertion pattern according to a possible embodiment of the present invention. Irregular insertion patterns may also be defined where Di is chosen so as to avoid reference signal for phase tracking insertion near reference signal for demodulation positions as it may be seen on FIG. 5. Therefore, reference signal for phase tracking density may be reduced by avoiding insertion where phase error may be corrected with low error by other means such as: - Signal blocks near reference signal for demodulation positions may not carry reference signal for phase tracking; - X signal blocks near reference signal for demodulation positions may not carry reference signal for phase tracking; - X signal blocks following reference signal for demodulation positions may not carry reference signal for phase tracking; - Y samples (Y