FR2908254A1 - METHOD OF MODULATION-DEMODULATION HIGH SPEED AND LOW LATENCY - Google Patents

Abstract

The invention relates to a method for demodulating a modulated signal based on an OFDM modulation transmitted in a W-bandwidth channel, the signal being represented in a time-frequency space by a set of elementary boxes TF, characterized in that it comprises at least the following steps: during the modulation of the signal, for each elementary box TF assign a part M of the sub-carriers of this box as reference subcarriers ,. for the demodulation of the signal after crossing the W channel ,. determining for each elementary box at least one of the parameters of the vector mean of the reference symbols associated with the box ,. use the average parameter value associated with an elementary box TF to demodulate the received OFDM symbol for the elementary box TF ,. repeat the demodulation step for all elementary boxes TF.

Description

METHOD OF MODULATION-DEMODULATION HIGH SPEED AND LOW LATENCY

  The invention relates to a high-speed, low latency modulation-demodulation method, based in particular on the use of a modulation in which the high-speed bit stream is distributed over a series of low-speed modulated sub-carriers and it There exists an orthogonality of these sub-carriers, this modulation being usually designated by a person skilled in the art by the term Orthogonal Frequency Digital Multiplexing (OFDM) in a wideband channel with a large fast Fourier transform (FFT). (at least equal to 256 points, but typically rather equal to 1024 points), time slot compatible or TDMA slots short and operating in very high frequency bands or VHF Anglo-Saxon (very high frequency) tactics (30_88 MHz) even ultra high frequency UHF in ultra-high frequency ultra high frequency 225-400 MHz. The method applies, for example, in the field of tactical land-based communications with movable pins equipped with azimuth omnidirectional antennas and whose phase center is located at a maximum of 3 meters from the ground, in the area of the tactical Internet. and ad hoc networks (self-configuring mobile network). The method can constitute the base of the ad hoc waveform, operating on a single-channel radio in the alternating mode (Rx / Tx management in TDD (Time Division Duplex), meaning that the transmission-reception phases are alternating, and not simultaneous) and implementing a dynamic TDMA One of the problems posed in the present invention is to reconcile the range of radio communications with the speed and reliability / stability of radio links, for example in the context of the future tactical Internet, 1 Mbps point-to-point demand, the radio range operational requirements are greater than 10 km 2908254 2 This requires efficient modulation / demodulation schemes that are resistant to both multi-path usage of omnidirectional low antennas and doppler effects related to the mobility of the pions.The coherent demodulation and differential demodulation schemes are known from the prior art. Broadband modulations are relatively vulnerable to narrow-band interference, because of the very principle of the Fast Fourier Transform or FFT (Fast Fourier Transformation) which breaks down the scrambling power over all of the subcarriers of the OFDM composite signal. on the one hand, and because of the limited number of frequency steps per slot on the other hand. The other point concerns time slot cutting. The prior art uses a non-ad hoc configuration-fixed waveform based on a static multiple access method (static TDMA) implementing slots or trivialized slots of fixed duration equal to 4 ms. To become ad hoc, the waveform must exchange network coordination messages, rather short but recurrent messages. The waveform is also intended to manage the quality of service (QoS) to treat in a differentiated way the different types of flows that have to pass through this web (messaging with more or less constrained delivery time, video

  videoconference, voice over IP or VoIP). In this dual perspective, it is important to have a slice of time or slot as short as possible, in order to be able to pass data packets not necessarily bulky as quickly as possible, even if to create longer slots 25 to pass the more demanding bandwidth flows. In any case, to pass data packets on the radio channel with a good probability of success, one must have a good frequency diversity the time of the slot. To do this, the prior art proposes for example two techniques: a) resorting to frequency evasion or intra-slot EVF, 2908254 3 b) broadband and streaming (slot time) on a single channel, the notion of broadband being defined by a pipe width greater than the maximum coherence band of the propagation channel that one is likely to find all 5 ground conditions combined (the maximum coherence band is in rural terrain flat or very rough and unobstructed); The technique a) is conventional and that which is used, for example, for positioning systems of the GSM type (global system mobile) and, in general, for all the radio systems working in pipeline. instant relatively narrow (less than a few hundred KHz). It is effective against fading or fading and at the same time provides transmission security or TRANSEC and scrambled-type interference resistance. Its main disadvantage is the compromise that must be made between slot duration, hop speed, number of steps per slot and redundancy coding in the channel coding scheme. With technique b) the variables are decoupled. The slot can be more easily pruned, depending on higher-level system parameters (short-rate short slot, high-speed long slot, more or less rugged slot against scrambling, narrow or pulsed band) while keeping the same basic modem, but configurable. OFDM modulation (orthogonal frequency division multiplexing) is perfectly suited to this setting. On the other hand, this technique does not make it possible to obtain the same resistance against broadband intentional interference (partial / sequential band jamming, jamming). For the most severe cases of interference, frequency evasion or EVF remains unavoidable. But in this case, the extra ECCM resistance is paid in terms of latency (longer slots).

The invention relates to a method for modulating-demodulating a modulated signal, the modulation being based on an OFDM modulation, transmitted in a W-bandwidth channel, the signal being represented in a time-frequency space by a set of elementary boxes TF, a box comprising several subcarriers, characterized in that it comprises at least the following steps: • during the modulation of the signal, for each elementary box TF assign a part M of the subcarriers of this box as reference subcarriers, • for the demodulation of the signal after crossing the W channel, • to determine for each elementary box at least one of the 10 parameters of the mean vector of the reference symbols associated with the box, • to use the averaged parameter value associated with an elementary box TF to demodulate the received OFDM symbol for the elementary box TF, 15 • reiterate the demodulation step po all the elementary boxes TF. The invention also relates to a modulation-demodulation system characterized in that it comprises at least one transmitter comprising at least one modulator adapted to implement a modulation scheme 20 according to the steps of the method described above, an antenna of transmission, and a receiver comprising at least one receiving antenna, a demodulator adapted for demodulating the received signal according to the principles of the invention described above.

The method according to the invention has the following advantages in particular: • It optimizes the compromise reached * speed in mobile tactical communications, • In particular, the invention optimizes the performance of demodulation on real propagation channel, that is to say frequency-selective channel and non-stationary, because of the presence respectively of multipath and Doppler associated with the mobility of pions. To do this, the method relies on a large FFT fast Fourier transform and uses a hybrid modulation method, taking advantage of the advantages of each of the following two schemes of coherent modulation and differential modulation type. by eliminating the bearing holes, absence of frequency evasion or intra-slot EVF frequency evasion, • It makes possible the constitution of short slots (slots much less than 4 ms) by capitalizing on the instantaneous frequency diversity inherent in the Broadband transmissions (frequency selective channel), which improves network performance, in terms of end-to-end traversal as well as in terms of convergence time of route establishment, • It can offer to the reception , an almost real-time BE interference detection mechanism, by taking advantage of a part of the OFDM-related FFT resources, more precisely by performing a brief and rapid spectral analysis between each OFDM symbol. This makes it possible, in particular, to check whether the spectral envelope does not exhibit any anomalies, such as a well-marked peak of local power, which would be characteristic of the superposition in the spectrum of a narrow-band interference (the spectral envelope of a signal affected by selective fading naturally presents pronounced power holes, never peaks).

Other features and advantages of the present invention will appear better on reading the description which follows, of a detailed example of at least one embodiment, given by way of indication and in no way limiting, appended figures which represent: FIG. 1 a block diagram of a device according to the invention comprising a modulator and a demodulator; FIG. 2 in a 3-dimensional diagram power, time, frequency, a curve representing the spectrum of the envelope of the signal received by a demodulator of the device of FIG.

In order to better understand the principle embodied in the invention, the following description is given for illustrative purposes for OFDM signal modulation. The demodulation modulation system represented in FIG. 1 comprises a transmitter 1 comprising a modulator 2 adapted to implement a modulation scheme according to the invention, a transmitting antenna 3 and a receiver 4 comprising a reception antenna 5, a demodulator 6 adapted for the demodulation of the received signal according to the principles of the invention described hereinafter. The demodulator allows the demodulation of the signal after transmission in a channel 7 of width W.

The OFDM multiplex signal can be schematized as a chessboard of elementary time-frequency boxes, called TF boxes (an elementary demodulation box is represented in FIG. 2 in a power-time-frequency diagram), a box comprising several subcarriers .

According to the frequency axis, the optimum size of the elementary box corresponds to the maximum bandwidth, below which the channel can still be considered as non-selective in frequency, for the most severe multipath situations against which we want to protect ourselves. It must correspond to the minimum coherence band that one is likely to find, all field conditions combined. Depending on the time axis, the size of the elementary box must not exceed the stationary time of the channel (temporal stability). The method according to the invention is based on a modulation and demodulation scheme using decorrelation of time-frequency boxes between them. The degree and granularity of time and frequency diversity depends on the propagation environment, i.e. fading and / or scrambling conditions. The method implements a hybrid demodulation judiciously combining the advantages of coherent demodulation (theoretical gain of 3dB in terms of energy efficiency Eb / NO, if perfect synchronization, in practice the gain varies from 1 to 2 dB): those of differential demodulation (no problem related to the aging of the channel estimate, conferring a better quality of demodulation in mobility and less vulnerability to pulsed jammers).

The method according to the invention executes, for example, the following steps: For the modulation of the OFDM signal, For each OFDM symbol, assign to a part of the N sub-carriers contained in an elementary TF box the function of subcarriers 15 of references, that is to say that there are M subcarriers carrying reference signals which are known by the transmitter and the demodulator (knowledge of the phase and amplitude parameters associated with a reference subcarrier) and NM subcarriers containing the useful signal information. The reference subcarriers carry a known time index at the transmitter and the receiver. These reference subcarriers are, for example, evenly distributed in the channel width per group of M all N. For demodulation of the signal, at each new OFDM symbol received, the demodulator has a new reference for the signal. current symbol forgetting what happened before. Within a received OFDM symbol, the receiver executes a Fourier transform allowing the passage of the time domain to the frequency domain.

For each of the elementary TF boxes, the demodulator carries out a vector average on the various reference symbols contained in this box (or assigned to this box), it then determines at least one of the components of the vector average, for example the value of the phase. The phase value thus determined is then used to demodulate the received signal corresponding to an elementary box TF.

Without departing from the scope of the invention, the demodulator can also determine the value of the amplitude and use it alone or with the value of the phase to demodulate the received signal. The number M of the subcarriers assigned as reference subcarriers is chosen, for example, according to the compromise between spectral efficiency and desired energy efficiency. In the case where M = 3 distributed subcarriers distributed all the N subcarriers within an elementary cell, each group of three reference subcarriers is used to locally demodulate the surrounding subcarriers ( +/- N / 2) according to the principle of coherent demodulation which consists in demodulating symbols with reference to a single symbol, serving as an absolute phase reference. These absolute references are thus refreshed with each OFDM symbol, thus providing an optimum updating of the local channel estimate for each of the TF boxes in near real-time.

For the implementation of the method according to the invention, the dimensioning of fast, FFT, primary and secondary Fourier transforms is for example as described below. The iterative procedure is as follows: Step 1: Define the multipath propagation channel for the intended RF frequency range (VHF or UHF), by determining the range of multipath temporal spreads [AT MIN; AT MAX] all terrain conditions combined. Step 2: Determine the compatible channel width W of the desired radio rate and respond to the following inequality: W 1 / AT MIN 2908254 9 In practice, the inequality is interpreted, for example, as follows: line W should be be, for example, at least 5 times larger than 1 / AT MIN. Step 3: Determine the size of the secondary FFT (intersymbol spectral analysis FFT). Take NFFTsec as the power of two, immediately inferior to the dimensionless product W. AT MAX. Step 4: Calculate the size of the primary FFT. NFFTprimary is a power of two, an order of magnitude greater than the product without the size W. AT MAX. In practice, take NFFTprimary equal to 8 or 16 times NFFTsec. Step 5: Make sure that the OFDM symbol TOFDM lifetime, equal to the inverse of the subcarrier spacing: TOFDM = NFFTprim / W 15 remains lower than the duration during which the propagation channel can be considered stationary (Limitation related to mobility and associated Doppler effects). This procedure, iterative, makes it possible to size the FFTs. The orders of quantities given can be modulated, in particular according to the requirements of realization (in particular if one wishes to make the two synchronous FFTs of the same sampling clock) and any margins offered by the channel. Propagation (The compromise between multi-path robustness and Doppler robustness can be more or less constrained).

In the case of an example of application in a tactical network, of the Lower Tactical Internet type, comprising several users in total and random mobility, equipped with antennas at 3m maximum height, the RF band retained is the VHF band. -88MHz.

By applying the FFT sizing procedure described above, the characteristics for application in a tactical network of the Lower Tactical Internet type are as follows: 1 Time delay of a mobile-mobile VHF propagation channel, antennas at 3m: All field conditions combined, we take the beach [5ps; 60ps].

2 Width of pipe: the value is 1.5MHz Value which is effectively much higher than 200kHz (= 1 / 5ps) and that we consider compatible the desired useful rate, for the example 10 of realization given.

3 FFT secondary the FFT chosen is a FFT 64 points, which is the power of two immediately less than 90 (90 = 1.5MHz x 60ps) 4 FFT primary 15 the FFT chosen is a FFT 1024 points, 16 times larger than the FFT secondary. In Fig. 2 the traced curve shows the spectral envelope of a broadband signal affected by fading shown in three dimensions in a Power-Time-Frequency space. In this case, it is the envelope spectrum of an OFDM signal in 5 MHz channel operating on one of the eight channels taken in the band 225-400 MHz (for illustration). The fact that the spectrum is not flat in the band is characteristic of a frequency-selective propagation channel, i.e. the instantaneous bandwidth of the signal (channelization) is greater than the coherence band of the propagation channel. Within an elementary TF demodulation box, the propagation channel can be considered as: stationary (time stable) 30 - frequency selective (flat spectrum in the band) 2908254 11 In the system according to the invention proposed for the VHF range, the indicative dimensions of the elementary TF demodulation box are: - 25kHz wide corresponding to the sub-multiple of 25 KHz closest to the minimum coherence band that one is likely to find, all conditions of propagation for the intended application (coherence band min = 16.7 kHz = 1 / 60ps). 743ps long corresponding to the duration of an OFDM symbol (= 1 / Subcarrier spacing + Symbol guard = 1 / 1.46kHz + 60ps).

The OFDM is based on an FFT 1024 giving a very good frequency resolution (spacing between subcarriers), of the order of 1.46 kHz therefore (= 1.5MHz / 1024). Within this box, the symbols are demodulated in a coherent manner, taking as a reference the local phase estimator obtained by averaging the frequencies of the 3 central subcarriers (More precisely by averaging the symbols carried by the 3 sub-carriers). -reported carriers) among the 17 that contains the box (It remains so 14 s / useful carriers per box). This phase reference is recalculated with each OFDM symbol. According to an alternative embodiment, the method offers on reception a Narrow Band, BE, quasi-real time interference detection mechanism, by taking advantage of a portion of the FFT resources. linked to LOFDM, performing a quick and summary spectral analysis between each OFDM symbol. This makes it possible, in particular, to verify whether the spectral envelope does not exhibit anomalies, such as a well-marked local power peak, which would be characteristic of the superposition in the spectrum of a narrowband interference (the spectral envelope of a signal affected by selective fading naturally has pronounced power holes, never peaks). This FFT called secondary FFT is preferably carried out: • In 25 kHz submultiple frequency resolution, the exact value is the submultiple of 25 kHz closest to the width of the elementary box 408, to minimize the FFT leaks in spectral analysis of 25 kHz channel jammers; • In rectangular windowing, the secondary FFT is performed during the guard time of the useful 5 OFDM symbol. The estimate of the interference state of the channel (scrambled narrow band / a priori not scrambled narrow band is therefore done in masked time, partially or completely depending on the relative duration of the guard time of the OFDM symbol vis-à-vis the Secondary FFT, which leads to a minimal or even no impact on the useful rate, and several secondary FFTs can be performed as a result, depending on the compromise between the reliability of the spectral estimation required for the anti-jamming treatment and the acceptable loss of flow rate The method and device described above can be used in the context of a mobile communications network or for fixed communication networks In summary, the method according to the invention maximizes the capacity of a modem to pass through. correctly very high bit rate bits even in the presence of severe fading phenomena (high Doppler multi-path-strong Doppler spreading) and / or under interference at least as long as these interference conditions are not too flat or monolithic (such as broadband jamming may be) but remain selective: • Frequency (scrambler BE narrowband) 25 • Temporal (pulsed scrambler) • Frequency and time (EVF emissions 25 KHz, for example) ... FT: MODULATION-DEMODULATION METHOD HIGH SPEED AND LOW LATENCY

Claims (2)

  A method for modulating-demodulating a modulated signal, the modulation being based on an OFDM modulation, transmitted in a W bandwidth channel, the signal being represented in a time-frequency space by a set of elementary boxes TF, a box comprising a plurality of sub-carriers, characterized in that it comprises at least the following steps: during the modulation of the signal, for each elementary box TF assigning a part M of the sub-carriers of this box as subcarriers reference, • for the demodulation of the signal after crossing the W channel, • determine for each elementary box at least one of the vector mean parameters of the reference symbols associated with the box, • use the average parameter value associated with a box elementary TF to demodulate the received OFDM symbol for the elementary box TF, • repeat the demodulation step for all elementary boxes TF. 2 - Process according to claim 1, characterized in that the modulation is an OFDM modulation. 3. Process according to claim 1, characterized in that the parameter used is the phase, the average value of the phase is determined in an elementary box TF and this average phase value is used to demodulate the symbol corresponding to the box TF. elementary. A method according to claim 2, characterized in that it comprises, for determining the FFT Fourier transforms to be used for the demodulation of the signal, at least the following steps: Step 1: Define the propagation channel in multipath term for the targeted RF frequency range (VHF or UHF), by determining the range of multipath temporal spreads [AT MIN; AT MAX], Step 2: Determine the compatible W channelization width of the desired radio rate and corresponding to the following inequality: W 1 / AT MIN Step 3: Determine the size of the secondary FFT (intersymbol spectral analysis FFT) ), Take NFFTsec as the power of two, immediately smaller than the dimensionless product W. AT MAX, Step 4: Calculate the size of the primary FFT, with NFFTprimary a power of two, an order of magnitude greater than the dimensionless product W AT MAX, Step 5: Make sure that the TOFDM useful life of an OFDM symbol, equal to the inverse of the spacing between subcarriers: TOFDM = NFFTprim / W 20 remains lower than the duration during which the channel propagation can be considered stationary. 5. Process according to claim 4, characterized in that the pipe W is chosen at least 5 times larger than 1 / AT MIN and NFFrprimary is equal to 8 or 16 times NFFTsec. 6. The method as claimed in claim 4, characterized in that the dimensions of the elementary demodulation box are: 25 KHz wide corresponding to the sub-multiple of 25 KHz closest to the minimum coherence band likely to be found for a target application, 2908254 - 743 ps long corresponding to the duration of an OFDM symbol. 7. Process according to claim 1, characterized in that it comprises a step of spectral analysis between each received OFDM symbol. 8 - Process according to claim 6, characterized in that it comprises a secondary FFT realized in frequency resolution of 25 KHz and in rectangular windowing, said secondary FFT being performed during the guard time of a useful OFDM symbol. Demodulation modulation system comprising at least the following elements: a transmitter (1) comprising at least one modulator (2) adapted to implement a modulation scheme by using the method according to one of claims 1 to 6; , a transmitting antenna (3) and a receiver (4) comprising at least one receiving antenna (5), a demodulator (6) adapted for demodulating the received signal using the method according to one of claims 1 to 8. 520