Receiver and FBMC reception method with low decoding latency

Abstract

The invention relates to a receiver and a FBMC reception method making it possible to reduce the decoding latency time and to increase the data rate in a communication system using a handshake exchange protocol or a TDMA access protocol. The receiver introduces zero padding values in place of the last samples of the last block of samples of a FBMC packet, without waiting for the end of this packet. The decoding of the FBMC packet is thus decoded more rapidly, without significant degradation of the error rate and without reduction of the out-of-band rejection rate of the FBMC signal.

Claims

1. A filter bank multi-carrier (FBMC) receiver configured to receive at least one packet of FBMC symbols N K, the receiver comprising: a sampler to sample at the frequency Nf a FBMC packet received in baseband, the FBMC symbols in the received FBMC packet being transmitted with a plurality N of frequency sub-channels and following one another at a frequent f=1/T with an overlapping factor K, where T is a symbol duration; a serial-parallel converter to form blocks of successive samples of size KN, a fast Fourier transformation (FFT) module to carry out a FFT of size KN on each of said blocks; a battery of analysis filters to carry out a filtering and a spectral despreading on frequency components at the output of the FFT module; and a first multiplexer to pad with zero values a first plurality (M.sub.z) of last samples of a last block of the FBMC packet at the input of the FFT module, without waiting for an end of reception of the FBMC packet.

2. The FBMC receiver according to claim 1, wherein the overlapping factor is equal to 4 and a first plurality of samples is equal to KN/3 to within 10%.

3. The FBMC receiver according to claim 1, further comprising a second multiplexer at the input of the FFT module to pad with zero values a second plurality of first samples of a first block of the FBMC packet.

4. The FBMC receiver according to claim 1, wherein the overlapping factor is equal to 4 and a second plurality of samples is equal to KN/3 to within 10%.

5. The FBMC receiver according to claim 1, wherein the receiver is configured to analyse multicarrier signals, including the FBMC packet, based on an analysis filter bank including frequency shifted versions of a prototype filter.

6. A method of receiving at least one packet of filter bank multi-carrier (FBMC) symbols, N K said method comprising: sampling at the frequency Nf of a FBMC packet received in baseband, the FBMC symbols of the received FBMC packet being transmitted with a plurality N of frequency sub-channels and following one another at a frequency f=1/T with an overlapping factor K, where T is a symbol duration; a serial-parallel conversion to form blocks of successive samples of size KN, a FFT of size KN on each of said blocks thereby obtained; a filtering and a spectral despreading, in the frequency domain, of frequency components at the output of the fast Fourier transformation (FFT); and prior to the FFT, a first step of padding with zero values a first plurality (M.sub.z) of last samples of a last block of the FBMC packet, the step of padding being carried out without waiting for an end of reception of the FBMC packet.

7. The method according to claim 6, wherein the overlapping factor is equal to 4 and said first plurality of samples is equal to KN/3 to within 10%.

8. The method according to claim 6, further comprising, prior to the FFT, a second step of padding with zero values a second plurality of first samples of a first block of the FBMC packet.

9. The method according to claim 8, wherein the overlapping factor is equal to 4 and the second plurality of samples is equal to KN/3 to within 10% .

10. The method according to claim 5, comprising an equalization of the frequency components prior to the step of filtering and spectral despreading.

11. The method according to claim 6, comprising an offset modulation demodulation after the step of filtering and spectral despreading.

12. The method according to claim 6, said method comprising analysing multicarrier signals, including the FBMC packet, based on an analysis filter bank including frequency shifted versions of a prototype filter.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other characteristics and advantages of the invention will become clear on reading preferential embodiments and referring to the appended figures among which:

(2) FIG. 1 represents in a schematic manner a communication between a transmitter and a FBMC receiver using a handshake protocol;

(3) FIG. 2 represents two chronograms of transmission of FBMC packets in TDMA mode;

(4) FIG. 3 represents the impulse response of a prototype filter, time truncated;

(5) FIG. 4 schematically represents the architecture of a FBMC receiver according to a first embodiment of the invention;

(6) FIG. 5 represents the principle of zero padding of the samples of the last FBMC symbol of a packet received by the receiver of FIG. 4;

(7) FIG. 6 represents the change in the signal to interference ratio at the level of the receiver as a function of the zero padding rate;

(8) FIGS. 7A and 7B represents the variation in the rate in saturated mode as a function of the number of users for two separate sizes of packets;

(9) FIG. 8 schematically represents the architecture of an FBMC receiver according to a second embodiment of the invention.

DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS

(10) The basic idea of the present invention is not to window the FBMC signal at the level of the transmitter and in particular not to truncate the impulse response of the prototype filter but on the contrary to carry out a treatment at the level of the receiver while padding the last samples of the last block received with zeros before carrying out the FFT. It has in fact been able to be shown that it is possible to only use a partial information to decode the last FBMC symbols. FIG. 4 schematically represents the structure of an FBMC receiver according to a first embodiment of the invention.

(11) The FBMC signal received, after having been demodulated in baseband, is sampled by a sampler, 400, at the frequency Nf where f=1/T.sub.s is the symbol frequency. The successive samples are grouped together, by a serial/parallel converter, 410, in the form of blocks of length KN where K is the overlapping factor.

(12) A sliding FFT (the window of the sliding FFT of KT between two FFT calculations) is carried out by means of a FFT module, 420, of KN consecutive samples.

(13) A multiplexer, 415, at the input of the FFT module, replaces the M.sub.z last samples of the block corresponding to the last FBMC symbol of the packet by zeros. This multiplexer is commanded by a command signal LPS, for example from a counter (not represented) indicating the last FBMC symbol of the packet.

(14) Thus, it will be understood that the receiver does not have to wait for the M.sub.z last samples of the last block of the packet to carry out the last FFT operation. The frequency components at the output of the FFT are then subjected, if need be, to an equalization in the frequency domain, in the equaliser 430. The equaliser is nevertheless an optional element of the invention, depending on the transfer function of the transmission channel.

(15) After possible equalization, the outputs of the FFT are filtered and spectrally despread by the battery of analysis filters, 440. More precisely, if P {hacek over (d)}.sub.i,k are the samples corresponding to the 2K1 frequencies (i1)K+1, . . . ,iK, . . . ,(i+1)K1 of the FFT (that is to say the frequencies of the i.sup.th sub-channel), the battery of filters provides (for this i.sup.th sub-channel) the sample:

(16) $d_i=\underset{k=-K+1}{\overset{K-1}{.Math.}}{G}_k{d}_{i,k}$
where the coefficients G.sub.k are the values of the transfer function of the analysis filter (translated to the frequency iK of the transfer function of the prototype filter).

(17) The data d.sub.i thereby obtained may undergo a plurality of operations, reserve to those implemented in the FBMC transmitter. For example, if the data have been subjected to a OQAM modulation (Offset QAM) at the level of the transmitter, the data d.sub.i then undergo a OQAM demodulation, in a manner known per se. Similarly, if the data have been coded by a channel coding and modulated by means of a Q-ary symbol modulation according to an MCS (Modulation and Coding Scheme) at the level of the transmitter, the reverse operations are carried out at the level of the receiver.

(18) FIG. 5 illustrates the principle of the treatment of the last block of samples obtained of the FBMC packet. On the X-axis is represented the rank of the sample for KN consecutive samples at the input of the FFT and on the Y-axis the amplitude. The waveform 510 is that of a FBMC signal corresponding to a single sub-channel and thus corresponds to the impulse response of the prototype filter. It may be seen that the last input block of the FFT module is constituted of KNM.sub.z samples of the signal, 520, and M.sub.z samples of padding constituted by zero values, 530. Thus, the receiver does not have to wait for the end of the time spread of the last FBMC symbol to terminate the decoding of the packet. In other words, the last FBMC symbol(s) (to take account of the overlapping of these symbols) are decoded on the basis of a partial information.

(19) It has been able to be shown that this zeroing of the end of the last block of samples only slightly affects the performances of the decoding of this block.

(20) FIG. 6 represents the change in the signal to interference ratio as a function of the rate of padding the last block with zeros. More precisely, the padding rate designates the ratio M.sub.z/NK. Here K=4 and N=1024 are taken for example.

(21) It may be seen that the signal to interference ratio remains constant as long as the padding rate remains below 0.15 then decreases to observe a plateau up to a padding rate of around , then again decreases.

(22) Depending on the desired minimum SIR level, in other words depending on the maximum acceptable binary error rate (BER), it is possible to determine by means of this curve the maximum padding rate. For example, in the case illustrated if it is wished to have a signal to interference ratio above 45 dB a padding rate close to 0.3 will be chosen. It will thus be understood that a time of around T/3 is gained compared to a conventional FBMC demodulation while retaining a very good out-of-band rejection rate since the FBMC signal transmitted is unchanged.

(23) Generally speaking, for an overlapping factor K=4, the zero padding rate of the last block is chosen equal to KN/3.

(24) FIGS. 7A and 7B represent the rate in saturated mode of a FBMC communication system as a function of the number of users. Two handshake protocols have been envisaged here. The first protocol is a basic version of resource access sharing known as CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) implementing two handshakes and the second protocol is an advanced version, using two additional RTS/CTS (Request To Send/ Clear To Send) signalisation signals and implementing four handshakes between the transmitter and the receiver.

(25) For each of these two protocols, the performances of a FBMC reception with a padding of zeros as described above is also represented.

(26) For these two protocols, a MCS scheme has been used with a channel coding of output R= and a 16 QAM modulation.

(27) FIG. 7A corresponds to a size of packet of 1500 octets and FIG. 7B to a size of packet of 500 octets. It will be noted that at saturation, the gain in rate is some 10% for a size of packet of 1500 octets and some 16% for a size of packet of 500 octets. As predicted, the smaller the packet size the greater the gain.

(28) FIG. 8 schematically represents the structure of a FBMC receiver according to a second embodiment of the invention.

(29) Unlike the first embodiment, the zero padding here intervenes both at the start of the first block of samples and at the end of the last block of samples of the FBMC packet.

(30) As in FIG. 4, the FBMC signal received is demodulated in baseband then sampled at the frequency Nf. The successive samples are grouped together, in a serial/parallel converter, 810, in the form of blocks of length KN where K is the overlapping factor.

(31) The FFT module 820 carries out a sliding FFT on a block of KN consecutive samples.

(32) A second multiplexer, 815, at the input of the FFT module, 820, replaces the M.sub.z last samples of the block corresponding to the last FBMC symbol of the packet by zeros. In a similar manner, a second multiplexer, 813, at the input of the FFT module, 820, replaces the M.sub.z first samples of the block corresponding to the first FBMC symbol of the packet by zeros. According to a variant, the number of zeroed samples may be different for the first and second multiplexers.

(33) The first multiplexer is commanded by an initialisation signal, INI, indicating the start of a new packet. The second multiplexer is commanded by a command signal LPS, from a counter, indicating the last FBMC symbol of the packet, as in the first embodiment.

(34) Thus, the receiver does not take account of the time spread of the first symbol and the last symbol of the FBMC packet for the decoder. It is then possible to reduce the size of the transmission intervals as in the chronogram (b) of FIG. 2 without degrading the level of interference and thus without increasing the error rate.

(35) As in the first embodiment, the frequency components at the output of the FFT are subjected to a potential equalization in the frequency domain by the equaliser 830, then to a filtering and a spectral despreading by the battery of analysis filters, 840.

(36) As in the first embodiment, also, the data at the output of the battery of filters, 840, may then be subject to a OQAM demodulation, a binary symbol demodulation followed by a channel decoding, as a function of the operations carried out on the side of the transmitter.