EP1180289A1 - Procede de communications radiomobiles amrt iteratif - Google Patents
Procede de communications radiomobiles amrt iteratifInfo
- Publication number
- EP1180289A1 EP1180289A1 EP00936982A EP00936982A EP1180289A1 EP 1180289 A1 EP1180289 A1 EP 1180289A1 EP 00936982 A EP00936982 A EP 00936982A EP 00936982 A EP00936982 A EP 00936982A EP 1180289 A1 EP1180289 A1 EP 1180289A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- channel
- symbols
- estimate
- paths
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/005—Control of transmission; Equalising
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/0204—Channel estimation of multiple channels
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/24—Radio transmission systems, i.e. using radiation field for communication between two or more posts
- H04B7/26—Radio transmission systems, i.e. using radiation field for communication between two or more posts at least one of which is mobile
- H04B7/2643—Radio transmission systems, i.e. using radiation field for communication between two or more posts at least one of which is mobile using time-division multiple access [TDMA]
Definitions
- the subject of the present invention is an iterative TDMA radio communication method ("time division multiple access", in English TDMA for "Time Division Multiple Access”). It can be applied to the European GSM mobile radio system [1] and its GSM ++ evolution or to the American D-AMPS system.
- TDMA systems make it possible to distribute several users communicating over the same radio channel over time [1, 3].
- time slot or interval, or time slot, or "timeslot” in English
- guard intervals are provided on either side of each time slot.
- Figure 1 attached thus shows time slots TA X , TA 2 , .. '. , TA N allocated to N users, with guard intervals IG.
- the radioelectric signal emitted during a time slot is obtained by frequency transposition of the signal into equivalent baseband.
- the baseband signal is the result of the filtering of the data stream to be transmitted by a so-called transmission filter.
- the transmitted data stream is composed of two sequences of data symbols separated in time by a sequence of symbols known to the receiver, called the reference sequence.
- the two sequences of data symbols may possibly originate from the coding and interleaving of the information to be transmitted.
- the reference sequence In order to allow an accurate estimation of the channel at the receiver, the reference sequence often has good time correlation properties.
- CAZAC sequences extension of "Constant Amplitude - Zero Autocorrelation” in English [1, 4, 5] are component sequences taken from the 'bipolar alphabet ⁇ -1, 1 ⁇ and having a circular autocorrelation function zero everywhere except at the origin.
- the mobile radio channel used during a communication between a transmitter and a receiver is generally of the multi-path type with rapid fading known as RAYLEIGH.
- the existence of several paths comes from the fact that the radioelectric wave propagates along several paths between the place of emission and the place of reception.
- the received signal is then the sum of several more or less delayed replicas and more or less altered in phase and amplitude.
- the receiver operates a filtering adapted to the transmission filter and to the channel and it optimally combines the energy contributions of all the replicas of the transmitted signal.
- the signal at the output of the matched filter is sampled at the rate of the symbols and it is whitened by means of a discrete filter called the whitener.
- the samples at the output of the whitening filter provide a filtered and noisy version of the transmitted data stream.
- the discrete filter associated with this filtered version also called discrete channel, presents a finite impulse response which varies from sample to another, it indirectly characterizes the realization of the multipath channel during the corresponding reception time slot.
- the samples at the output of the whitening filter corresponding to the reference sequence are used to estimate the discrete channel [1, 4, 5]. This estimation of the discrete channel makes it possible to equalize the remaining samples possibly making it possible to detect the two sequences of data transmitted.
- the usual equalizer is known as the VITERBI equalizer with flexible decisions [2]. As its name suggests, this equalizer provides, thanks to an application of the VITERBI algorithm with weighted outputs ("Soft-Output Viterbi Algorithm" or SOVA) [2], flexible decisions on all data symbols issued. These flexible outputs are possibly deinterleaved and decoded to detect the information sent.
- the transmitter E comprises a data source 10, an encoder / interleaver 12 delivering symbols a *, a modulator 14.
- the ulti-path channel is symbolized by block 20.
- the receiver R comprises a demodulator 30 (matched filter / filter whitener) delivering samples R k , a discrete channel estimator 32, an equalizer 34, a deinterlacer / decoder 36 and finally a recipient 38.
- the discrete channel seen at the output of the whitening filter can vary significantly from one time slot to the next. This variation is mainly due to the change in the propagation conditions between the transmitter and the receiver and to the frequency stability at the level of the receiver.
- the propagation conditions have a direct influence on the observed multipath channel. They change due to a change in the environment or a displacement of the transmitter and / or the receiver. They cause a temporal variation of the discrete channel, both between successive time slots and within the same slot.
- the variation in the discrete channel between two successive time slots allocated to the same user is greater the greater the time interval between these slots. This variation is accentuated, even at the level of a single time slice, due to an increase in the carrier frequency or the speed of the transmitter and / or receiver.
- the variation of the discrete channel between two time slots is large enough to prevent any adaptive estimation thereof.
- the channel estimation at the time slot level must then be based on the samples of the corresponding reference sequence only.
- the use of this reference sequence provides an unbiased estimate of the discrete channel in the middle of the time slot.
- the estimate obtained in the middle of the frame can be used without equalization to equalize the rest of the frame.
- the high-speed movement of certain terminals and the sustained demand for high-speed services operating on increasingly large radio frequencies make this hypothesis less and less justified.
- the discrete channel can indeed undergo significant variations between the beginning and the end of a given time slot.
- the gap between this real discrete channel and its estimate widens more and more as we move away from the reference sequence. This can cause significant and irreversible degradation of the quality of reception and / or performance of the TDMA system.
- the purpose of the present invention precisely is to remedy these drawbacks.
- the present invention aims to increase the performance of TDMA systems by reducing the ratio between the power received and the multiple access interference for a given quality of reception or, in other words, improving the quality of reception at constant reception power.
- This improvement in quality makes it possible, among other things, to increase the capacity and coverage of TDMA systems, such as the GSM system and especially its EDGE extension to offer high-speed services.
- This improvement is obtained in particular thanks to an optimization of the channel estimation at the receiver. This optimization makes it possible to counter any degradation in performance caused by a rapid variation of the channel between the start and the end of the same time slot. It also makes it possible to significantly reduce the degradations generated by an increase in the number of states of the modulation and this without recourse to an increase in the length or the power of the reference sequence.
- Another object of the invention is to reduce the cost price of the terminals by making them, with unchanged reception quality, less sensitive to the frequency instability of a cheap local oscillator allowing, among other things, the transposition of the signal received in baseband.
- the invention also makes it possible to reduce, for equal reception quality, the number and / or the power of the reference symbols transmitted during each time slot.
- This object is achieved by taking into account, in the estimation of the discrete channel, data symbols which are generally more numerous than the reference symbols.
- This object is also achieved by optimally taking into account the samples corresponding to the reference and / or data symbols of an arbitrary number of time slots allocated to the user. This is taken into account even in the event of significant variations in the channel between two successive tranches belonging to the same user.
- This goal is finally achieved by taking optimal account of the samples from the common signaling channel and potential users on the downlink.
- the reference symbols are distributed in groups over the entire duration of the allocated time slot, which guarantees a better continuation of the variation of the channel and therefore an increased robustness with respect to the high travel speeds and the frequency instability of the local oscillator.
- the reference sequences can be chosen independently of each other, but they must always have good autocorrelation properties. They can be derived from CAZAC type sequences of reduced length.
- the energy transmitted per symbol can vary from one symbol to another without this posing difficulties for the estimation of the channel, which will always be carried out in an optimal manner.
- the processing of a time slot is done in blocks each time the samples corresponding to this slot and possibly to those of other slots (coming from the same user, from other users or from the common signaling channel ) are available to perform channel estimation.
- TDMA time division multiple access
- One of the characteristics of the invention is the use of a trellis to represent the multi-path channel. This feature reduces the complexity of the receiver. Indeed, if we take as an example sequences of data symbols comprising n symbols, without coding, with a two-state phase modulation (MDP2), each bit transmitted can take one or the other of two values . As these values are equiprobable, this means that there are 2 n possible sequences of data symbols. The complexity of a traditional calculation using partial probabilities for all possible sequences therefore increases exponentially (2 n ) with the number of symbols per sequence.
- MDP2 two-state phase modulation
- the labels of the branches of the trellis sections change as the temporal variation of the discrete channel changes at the level of each of the constituent time slots of the block.
- a algorithm different from the VITERBI algorithm with flexible outputs (SOVA) [2] is used for a more precise characterization of the probabilities of the branches of the trellis conditionally to the samples received.
- SOVA VITERBI algorithm with flexible outputs
- the only weighted outputs provided by the SOVA algorithm are used directly after deinterlacing to decode the information transmitted.
- conditional probabilities indirectly characterizing the data symbols are used, in addition to the reference symbols, to provide a better quality estimate of the discrete channel.
- the improved estimate of the discrete channel obtained at the end of a given iteration can be used to increase the quality and reliability of the conditional probabilities of the branches of the trellis characterizing the discrete channel. These improved conditional probabilities can, in turn, be used, together with the reference symbols, to make an additional improvement to the estimation of the discrete channel.
- an iterative estimate leads to an optimal estimate of the discrete channel only after an infinite number of iterations. In practice, however, a few iterations are sufficient without significantly degrading performance compared to an optimal estimate.
- the corresponding trellis can be used either by the optimal BAHL algorithm (or any other simplification thereof), or by the SOVA algorithm to provide improved weighted outputs to the decoder.
- the optimal nature of the receiver according to the invention is linked to the quality of the estimation of the discrete channel seen during a given time slot.
- This optimal character is based on the use of an iterative algorithm called SAGE (abbreviation of "Space-Alternating Generalized Expectation-Maximization Algorithm”) [7] to find the most likely channel realization conditionally to the samples received from the block to be processed.
- SAGE algorithm is an extension of the EM (contraction of "Expectation-Maximization”) algorithm [8, 9] which makes it possible to get rid of the problem of coupling between coefficients of the discrete channel during their estimation.
- the estimation of the discrete channel is also based on the decomposition of each path of the discrete channel to be estimated according to an expansion algorithm called KARHUNEN-LOEVE [3].
- KARHUNEN-LOEVE an expansion algorithm
- This decomposition provides, on the one hand, a flexible characterization of the variations temporal of each of the discrete paths and, on the other hand, fits naturally into the SAGE algorithm. It also makes it possible to deal with the case where the energy emitted by symbol varies from one symbol to another.
- the estimation of the discrete channel is finally based on the use of the BAHL algorithm, which provides at each iteration of the SAGE algorithm the probabilities of the branches of the trellis characterizing the discrete channel, conditionally to the samples received and to the estimation of channel provided in the previous iteration.
- the subject of the invention is therefore a method of radiomobile communications of the time division multiple access type, in which time slots are allocated to several users for the transmission and reception of radioelectric signals, and in which :
- data symbols and reference symbols are constituted in each allocated time slot, a radioelectric signal containing these data symbols and these reference symbols is emitted, this signal using a multi-path radio channel ,
- this channel is represented by a trellis comprising branches symbolizing the transitions between two successive states of the channel and characterizing the possible sequences at the exit of the channel, ii) the probabilities of these branches by implementing an iterative algorithm in which we calculate, by a finite number
- FIG. 9 illustrates the estimation of the coefficients of the filter of the discrete channel by correlation of the samples received with a CAZAC sequence and its circular offsets
- FIG. 11 illustrates a lattice representation of the normalized discrete channel seen at the output of the whitening filter for a two-path channel and a two-state modulation and a time slot length equal to four;
- FIG. 12 shows the chronology of the processing performed on reception for a time slot received
- FIG. 13 shows the chronology of the treatments carried out in an elementary iteration of the channel estimation algorithm.
- the performance of the method according to the invention has been compared with that of a variant of the conventional method used in the context of the GSM system [1].
- the conventional receiver of the GSM system first estimates the discrete channel in the middle of the time slot to be processed, by correlation of the samples received depending solely on the reference sequence with the corresponding CAZAC sequence. It then equalizes the other samples received using the SOVA algorithm in order to decode the information symbols sent.
- the sub-optimal SOVA algorithm has been replaced by the optimal BAHL algorithm used in the invention.
- the comparison is obtained through the evolution of the raw bit error rate (BER) (without taking into account any corrective coding or error detector), for reference symbols also grouped in the middle of the slice time to process (like the GSM system) that distributed in small groups over the duration of this section (in accordance with the invention), in order to guarantee a better continuation of the temporal variation of the channel.
- BER bit error rate
- FIG. 3 thus represents the evolution of the BER according to the normalized Doppler spreading B D T S , product of the Doppler spreading B D and the duration of a symbol T s , which characterizes the intensity of variation of the channel due to movement of the transmitter and / or receiver.
- Doppler spreading is both proportional to the speed of the mobile terminal and the frequency of the carrier.
- the frequency instability of the local oscillator which is not taken into account here, leads to additional degradations in performance.
- the modulation used is the MDP4 modulation (4-state phase shift modulation).
- the E s / M ratio between the average energy received per symbol and the noise level is 10 dB.
- Three configurations were considered: m A central reference sequence of 26 bipolar symbols taken at random from the 8 sequences used in the GSM system. These sequences are cyclic extensions of the 8 CAZAC sequences of length 16 [1, 11]. M A central reference sequence, ⁇ -1, +1, +1, +1, -1 ⁇ , of 5 bipolar symbols obtained by cyclic extension (to a sign and an offset) of the single CAZAC bipolar sequence of length 4 namely ⁇ -1, +1, +1, +1 ⁇ . m Five groups of reference sequences, distributed over the entire time slot to be processed, identical to the central reference sequence of 5 previous bipolar symbols (25 reference symbols in total). In Figure 3, the curves shown correspond to the following cases:
- Table 1 characterizes in a quantified way the improvement in performance brought by the invention vis-à-vis the conventional algorithm improved of the GSM system.
- RC means "conventional receiver” and RI receiver according to the invention.
- the E s / NO ratio between the average energy received per symbol and the noise level is 10 dB.
- the improvement provided by the invention is negligible.
- the reference sequence in this case, is long enough to guarantee, without the aid of the unknown data sequences, good quality of channel estimation.
- the receiver according to the invention makes it possible to obtain not only better performance than the conventional receiver, but also performance comparable to that obtained with the sequence of long reference of 26 reference symbols. Consequently, the receiver according to the invention is made, in a certain way, insensitive to an even significant reduction in the size of the reference sequence. This result is explained by taking into account, in the channel estimation, the symbols of the data sequences which come to cooperate with the reference symbols.
- the receiver according to the invention makes it possible to maintain the bit error rate almost unchanged by varying the normalized Doppler spread between 1/2500 and 1/100.
- the modulations used are the MDP4 and MDP8 modulations.
- a central reference sequence ( ⁇ -1, +1, +1, +1, -1 ⁇ for two paths and ⁇ -1, +1, +1, +1, -1,
- the curves relate to the following cases:
- the curves relate to the following cases:
- the curves relate to the following cases: 71 conventional receiver, 72 receiver of the invention, 73 receiver with perfectly known discrete channel.
- the curves in FIGS. 4 and 5 show that, unlike a receiver according to the invention, the performance of the conventional receiver of the GSM system is very sensitive to the length of the reference sequence used and degrades substantially when we go from the first configuration to the second.
- the curves of FIGS. 5 and 6 show, moreover, that the degradations in performance of the conventional receiver increase significantly with the number of paths whereas, conversely, these degradations remain almost unchanged and weak for a receiver according to the invention.
- the receiver according to the invention therefore has excellent robustness with regard to any reduction in the length of the reference sequence and any increase in the number of paths of the channel.
- Table 2 characterizes in a quantified manner, in the case of the MDP4 modulation, the improvement in performance provided by the receiver according to the invention with respect to the improved algorithm of the GSM system.
- a receiver according to the invention always has better performance than an improved conventional receiver.
- the improvement brought by the invention is all the greater as the length of the reference sequence is reduced or as the number of paths of the channel is increased.
- the modulation used is always the MDP modulation.
- One configuration with three sequences distributed benchmark is considered. Each sequence is composed of 4 + Ll bipolar symbols and is obtained by cyclic extension (to a sign and an offset) of the only bipolar CAZAC sequence of length 4, namely ⁇ -1, +1, +1, +1 ⁇ .
- the value of the Doppler spread considered is sufficiently restrictive not to allow a conventional receiver to follow the variations of each of the paths of the discrete channel which can undergo phase rotations of ⁇ ⁇ between the start and the end of each time slot .
- the performance of the method of the invention having been exposed, the reception phase and in particular the processing of the received signal will be described in more detail.
- a receiver according to the invention determines, conditionally on the samples received from the block to be processed, the realization of the most likely discrete channel during a given time slot.
- MAP maximum posterior probability
- the transmitted energy E k associated with each symbol a k is independent of the values taken by this symbol but can vary from one symbol of the time slot to another. This variation is generally the consequence of an allocation by the TDMA system of different energies to the data sequences and to the reference sequences.
- the samples at the output of the whitening filter corresponding to this time slice form a column vector
- R (R0> R ⁇ > - - -> R -.p R ⁇ "- - - '- ⁇ + -.- 2) of N + L - l components, where L is the number of paths of the discrete channel.
- Figure 8a brings together the means between the input of the transmitter modulator and the output of the receiver whitening filter.
- This figure therefore provides a representation of the real radio channel and of the processing carried out at the level of the receiver (demodulation, adapted filtering and whitening filtering).
- the modulator 91 a block 92 symbolizing the multi-path channel
- the demodulator 93 with its adapted filter 94 and its whitening filter 95.
- FIG. 8b shows a representation of these means in the form of a shift register with (Ll) cells Yi, Y 2 , ..., Y L - ⁇ -
- This register receives the current symbol a n and the cells contain the delayed symbols d 'a symbolic period, namely a n . lf a n - 2 , ..., a n - fL -j ) .
- the circuit shown also includes (L) multipliers Z 0 , Zi, ..., Z-. ⁇ receiving respectively
- c k ' is the / th coefficient of the discrete channel filter at k è ⁇ ae sampling instant and W k is a white Gaussian additive noise.
- the coefficients c of the discrete channel, verifying k ⁇ l or k> N + l intervene in the expression of R k when they are zero.
- the symbols a k emitted, verifying k ⁇ 0 or k> N - ⁇ appear in this expression of R ⁇ . even if they are zero.
- the Doppler power spectra generally encountered in mobile radio communications are either of the conventional type (in outdoor environments) or of the flat type (in indoor environments).
- the average power ⁇ Q ' varies from one path to another and characterizes the path intensity profile ("multipath intensity profile" in English [3]) of the discrete channel.
- MAP maximum posteriori criterion
- it needs an adequate representation of the discrete channel allowing it to get rid of the temporal correlation of the coefficients of the same path.
- It also needs a representation which allows it to overcome the possible variation of the energy transmitted from a symbol of the time slot to the other. This representation is obtained first of all by the introduction of the vector of normalized transmitted symbols
- the vectors of the orthonormal base associated with the / th path are none other than the eigenvectors of the covariance matrix H 'of the normalized vector C associated with this path.
- the l, k ) N- ⁇ eigenvalues [To k _ 0 associated are none other than the variances of the components G 0 ', G [, ..., G N l _ of the adequate representation G' of the path corresponding.
- B '* ⁇ associated with each path, as well as the corresponding eigenvalues, can therefore be obtained from the matrix equations
- ⁇ l- n denotes the discrete autocorrelation function of this path defined above.
- the receiver according to the invention must normally have L orthonormal bases associated with each of the L paths of the discrete channel in order to reconstruct or decompose any intermediate or final estimate of the discrete channel.
- L orthonormal bases associated with each of the L paths of the discrete channel in order to reconstruct or decompose any intermediate or final estimate of the discrete channel.
- the receiver To generate the orthonormal basis of the / th path, the receiver must have as precise an estimate as possible of the covariance matrix H 'of this path. To this end, he must have as precise a knowledge as possible of the characteristics of the Doppler power spectrum of the radio channel.
- This knowledge is updated regularly and necessarily includes an estimate of the Doppler B D spreading or an upper bound thereof. It can optionally include an estimate of the average power of each of the paths of the discrete channel and of the shape of the Doppler power spectrum.
- the receiver can either calculate in real time the appropriate orthonormal bases, or search a bank of pre-calculated orthonormal bases for those which best match the characteristics of the real channel.
- the receiver has no precise knowledge of the shape of the Doppler power spectrum, nor perfect and instantaneous knowledge of the Doppler spread. In this typical case, it can possibly assume a flat Doppler power spectrum with, as spreading Doppler, an upper bound of the real Doppler spread. Furthermore, the receiver may not have an accurate estimate of the average power of each of the paths of the discrete channel taken individually. In this case too, it can also assume a discrete channel with L paths of equal average powers.
- the estimation of the vectors c 'characterizing the L paths of the discrete channel is done iteratively according to the MAP criterion through the normalized vectors C' or their adequate representation G. This estimate can optimally take into account both the characteristics of the data and reference symbols transmitted (statistics and energies) as well as the characteristics of the multipath channel.
- the posterior conditional probability density R ( ⁇ C] / 0
- the SAGE algorithm successively re-estimates, by induction, each of the vectors G 'while keeping unchanged the previous estimates of the other vectors G m , with m ⁇ l. In accordance with the properties of the SAGE algorithm, these successive reestimations guarantee a monotonic growth of the conditional probability density a posteriori QG'j
- the receiver Given the vector R of samples received during a time slot to be processed, the receiver begins by calculating initial conditions ⁇ G (0) ⁇ .
- the reference sequences must be such that a rapid and simple separation of the contributions of the L paths to each of these samples is guaranteed.
- one or more reference sequences must be used, each having good properties. autocorrelation.
- these sequences can be constructed from CAZAC sequences by extension or circular permutation. These sequences are composed of symbols of constant modulus and have a circular autocorrelation function zero everywhere except the origin. In the context of the present invention, it is preferable to have CAZAC sequences with symbols belonging to the constellation of the modulation used to convey the data symbols.
- the shortest CAZAC sequence (except for a sign and a circular permutation), namely ⁇ -1, +1, +1, +1 ⁇ , has only four symbols.
- the only CAZAC sequence (always to a multiplicative factor i, -1 or - i, and to within a circular shift) the shortest, to to know ⁇ l, / ⁇ , is composed only of 2 symbols.
- a reference sequence based on a CAZAC sequence of length M only makes it possible to estimate that M paths of the discrete channel at most.
- the length M of the CAZAC sequence must therefore be greater than or equal to the number of paths L of the discrete channel to be estimated.
- the first line, A gives the reference sequence
- the second, Tl corresponds to the first path
- the third, T2 to the second path
- the fourth, S gives the composition of the samples received.
- the elementary reference sequence occupies the positions between n-4 and n of the time slot to be processed.
- the samples received, depending only on this sequence, occupy the positions between n-3 and n. They are given explicitly by:
- the rough estimate of the L coefficients of the channel, at a given instant of the time slot is obtained by correlation of the samples depending only on the reference sequence with L offset versions of the corresponding CAZAC sequence.
- Each reference sequence constructed by circular extension of a CAZAC sequence makes it possible to obtain a noisy sample of each of the paths of the discrete channel.
- the number of samples per time slice necessary for the reconstruction of the channel depends on the normalized Doppler spreading B D T S and on the number N of symbols making up this slice. The larger the product B D T S N, the more the channel varies from one end to the other of the time slot and the greater the number of elementary reference sequences.
- An elementary reference sequence is a subset of grouped symbols of the reference sequence associated with the time slot obtained from a circular extension of a CAZAC sequence.
- the eigenvector of larger eigenvalue is sufficient by itself for a precise representation of each of the paths of the discrete channel. This eigenvector then varies very slowly over time and therefore makes it possible, thanks to a reference sequence grouped in the middle of each time slot, to reconstruct an almost constant discrete channel over the entire time slot. This is the case with the GSM system where the normalized Doppler spreads are assumed to be very low and the channel is assumed to be constant over a given time slot.
- the two eigenvectors associated with the two largest eigenvalues are necessary for an accurate reconstruction of the discrete channel. These two vectors vary enough for a reference sequence composed of two elementary sequences, placed on either side of the middle of the time slot, to be necessary.
- the greater the product B D T S N the greater the number of eigenvectors to be used in the adequate representation of the discrete channel.
- the larger this product the greater the variations in these eigenvectors and therefore the greater the number of elementary reference sequences to be used.
- the greater the number of elementary reference sequences the more the length of each must be reduced.
- the length of an elementary sequence, M + L1 can at best be reduced to 2L-1 by choosing a CAZAC sequence of length L; hence the usefulness of having very short CAZAC sequences.
- the CAZAC sequence is recommended for all phase modulations when the number of paths L is less than or equal to 4 and, that the product B D T S N is weak.
- K be the number of elementary reference sequences per time slot to be processed.
- the optimal channel estimator according to the MAP criterion calculates iteratively, path after path, the reestimation
- the weights w k take into account the quality of the contribution of the vector B * to the adequate representation of the discrete channel.
- C n 'course is obtained by taking into account the contribution of all the values that can be taken by A ".
- the iterative estimation of the discrete channel can be accomplished a finite number D of times such that the global estimation ⁇ C / (D) ⁇ ; 0 (obtained from
- the modulator-multi-path-demodulator channel assembly can be modeled by a discrete-time MARKOV chain.
- Such a trellis represents the expected output of the channel as a function of time and according to the symbols transmitted. In the illustrated case where the channel is supposed to present two paths, the output of the channel is conditioned by the symbol emitted at the instant considered and at the previous instant.
- the branches of the trellis must be labeled by the coefficients of this intermediate estimate and not by those of the real discrete channel.
- the BAHL algorithm makes it possible to efficiently calculate the conditional posterior probabilities of the branches of the trellis by factorizing the expression of the probability of the received vector R, conditionally to each of the sequences A that can be emitted, so as not to duplicate all the common operations to sequences transmitted having the same past or the same future at a given time.
- a (S) ⁇ A + , (S) y Hl (S, S ').
- P (A n ) is the a priori probability of sending the symbol A n .
- These a priori probabilities can characterize a possible coding of the modulated symbols actually transmitted. They can also characterize the reference symbols which are known to the receiver and which take only one possible value. For these reference symbols, the a priori probability P (A n ) takes the value 1 for the value actually transmitted and 0 for the other possible values of the modulation.
- P (A n ) takes the value 1 for the value actually transmitted and 0 for the other possible values of the modulation.
- the VITERBI algorithm searches for the most likely transmitted sequence. It therefore makes it possible to reduce the probability of decision errors on these transmitted sequences.
- the SOVA algorithm is a VITERBI algorithm with weighted outputs. It allows a more precise characterization of the reliability of the decisions made by the VITERBI algorithm.
- the BAHL algorithm makes it possible to determine the posterior probability density of each symbol transmitted, conditional on the samples received and the channel estimation obtained. It therefore makes it possible to determine the most likely value of each symbol emitted and therefore to reduce the error rate.
- the weighted outputs are obtained through the probabilities of the symbols conditionally on the samples received.
- the weighted outputs are only useful if correcting decoding or source coding with weighted inputs is used in order to increase the performance of the receiver.
- the receiver uses the estimation
- the supply to an error correcting decoder or to a source coder of firm or flexible decisions on the symbols transmitted is carried out by means 123 which possibly implement the VITERBI algorithm [3], the SOVA algorithm [2], the BAHL algorithm [6] or the linearized BAHL algorithm [10].
- the iterative reestimation of the channel is performed D + l times and makes it possible to calculate
- FIG. 13 illustrates the L steps composing each iteration and which are necessary for the successive estimation of each of the L paths. Thanks to the means 121 and 122 of FIG. 12, (providing the bases
- the ( ⁇ + l) th iteration 3 d + ⁇ of the channel estimator implements a first step, based on the BAHL algorithm and represented by the BAHL 0 block referenced 130, to deliver the conditional probabilities a posteriori
- Th iteration then implements a second step, based on the SAGE algorithm and represented by the SAGE 0 block referenced 131, to allow the (d + 1) th reestimation G 0 (+1) of the first path.
- the ( ⁇ i + l) th iteration implements a third step, based again on the BAHL algorithm and represented by the BAHL block, referenced 132, to deliver conditional probabilities to
- the (fi. +1) th iteration implements a fourth step, based on the SAGE algorithm and represented by the SAGE block, referenced 133, to allow the (d + l) th reesti ation G / ( ⁇ + I) of the / th journey.
- the (d + l) th iteration implements a last step, based on the BAHL algorithm and represented by the BAHL block L _, referenced 134, to deliver the conditional probabilities a
- the first iteration can not implement algorithms BAHL SAGE and to calculate the first estimation of the discrete channel.
- the reference symbols are then used, the conditional a priori and a posteriori probabilities of which are well known to the receiver to make this first estimate.
- This first estimation is carried out, by correlating the samples received, depending only on these reference symbols, with the corresponding CAZAC sequences. It uses the means 122 of FIG. 12 to calculate the weights ⁇ w k '].
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Mobile Radio Communication Systems (AREA)
- Noise Elimination (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
- Time-Division Multiplex Systems (AREA)
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR9906945A FR2794589B1 (fr) | 1999-06-02 | 1999-06-02 | Procede de communications radiomobiles amrt iteratif |
FR9906945 | 1999-06-02 | ||
PCT/FR2000/001483 WO2000076160A1 (fr) | 1999-06-02 | 2000-05-30 | Procede de communications radiomobiles amrt iteratif |
Publications (1)
Publication Number | Publication Date |
---|---|
EP1180289A1 true EP1180289A1 (fr) | 2002-02-20 |
Family
ID=9546271
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP00936982A Withdrawn EP1180289A1 (fr) | 1999-06-02 | 2000-05-30 | Procede de communications radiomobiles amrt iteratif |
Country Status (5)
Country | Link |
---|---|
US (1) | US6990092B1 (ja) |
EP (1) | EP1180289A1 (ja) |
JP (1) | JP4472904B2 (ja) |
FR (1) | FR2794589B1 (ja) |
WO (1) | WO2000076160A1 (ja) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7139336B2 (en) * | 2002-04-05 | 2006-11-21 | Nokia Corporation | Method and system for channel estimation using iterative estimation and detection |
ATE264582T1 (de) * | 2002-05-03 | 2004-04-15 | Elektrobit Ag | Verfahren zum bestimmen der signalausbreitungseigenschaften eines drahtlos übertragbaren signals |
US7941149B2 (en) | 2002-05-13 | 2011-05-10 | Misonimo Chi Acquistion L.L.C. | Multi-hop ultra wide band wireless network communication |
US7957356B2 (en) | 2002-05-13 | 2011-06-07 | Misomino Chi Acquisitions L.L.C. | Scalable media access control for multi-hop high bandwidth communications |
US8780770B2 (en) | 2002-05-13 | 2014-07-15 | Misonimo Chi Acquisition L.L.C. | Systems and methods for voice and video communication over a wireless network |
DE10303475B3 (de) | 2003-01-29 | 2004-10-07 | Infineon Technologies Ag | Maximum-Likelihood-Abschätzung der Kanalkoeffizienten und des DC-Offset in einem digitalen Basisbandsignal eines Funkempfängers mit dem SAGE-Algorithmus |
US7711035B2 (en) * | 2004-09-17 | 2010-05-04 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for suppressing communication signal interference |
US7672412B2 (en) | 2006-04-05 | 2010-03-02 | Research In Motion Limited | Method and receiver for estimating the channel impulse response using a constant modulus interference removal iteration |
US8175613B2 (en) | 2006-08-04 | 2012-05-08 | Misonimo Chi Acquisitions L.L.C. | Systems and methods for determining location of devices within a wireless network |
CN101611571B (zh) * | 2006-12-07 | 2013-04-03 | 米索尼莫奇获取有限公司 | 时隙和信道分配系统和方法 |
WO2008093716A1 (ja) * | 2007-01-31 | 2008-08-07 | Panasonic Corporation | 無線送信装置及び無線送信方法 |
CN101771476B (zh) * | 2009-01-06 | 2013-04-24 | 华为技术有限公司 | 感知无线电中次要用户的频谱接入方法及装置 |
US8995583B2 (en) * | 2010-07-30 | 2015-03-31 | Telefonaktiebolaget L M Ericsson (Publ) | Decoding technique for tail-biting codes |
WO2012171553A1 (en) * | 2011-06-14 | 2012-12-20 | Nokia Siemens Networks Oy | Apparatus and method for allocating the reference symbols in order to reduce the latency of the lte system |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2747870B1 (fr) * | 1996-04-19 | 1998-11-06 | Wavecom Sa | Signal numerique a blocs de reference multiples pour l'estimation de canal, procedes d'estimation de canal et recepteurs correspondants |
FR2782587B1 (fr) * | 1998-08-20 | 2000-09-22 | France Telecom | Procedes de communications numeriques amrc a repartition des symboles de reference |
-
1999
- 1999-06-02 FR FR9906945A patent/FR2794589B1/fr not_active Expired - Fee Related
-
2000
- 2000-05-30 EP EP00936982A patent/EP1180289A1/fr not_active Withdrawn
- 2000-05-30 JP JP2001502312A patent/JP4472904B2/ja not_active Expired - Lifetime
- 2000-05-30 WO PCT/FR2000/001483 patent/WO2000076160A1/fr not_active Application Discontinuation
- 2000-06-02 US US09/585,352 patent/US6990092B1/en not_active Expired - Fee Related
Non-Patent Citations (2)
Title |
---|
None * |
See also references of WO0076160A1 * |
Also Published As
Publication number | Publication date |
---|---|
FR2794589A1 (fr) | 2000-12-08 |
WO2000076160A1 (fr) | 2000-12-14 |
JP2003512747A (ja) | 2003-04-02 |
JP4472904B2 (ja) | 2010-06-02 |
US6990092B1 (en) | 2006-01-24 |
FR2794589B1 (fr) | 2001-08-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1282968B1 (fr) | Procede et systeme de detection et de decodage iteratif de symboles recus, couple a une reestimation des coefficients du canal de transmission | |
EP1423921B1 (fr) | Procede et systeme de reception iterative sous optimale pour systeme de transmission haut debit cdma | |
EP1362447B1 (fr) | Procede et systeme de codage-decodage iteratif de flux de donnees numeriques codees par combinaisons spatio-temporelles, en emission et reception multiple | |
EP0802656B1 (fr) | Procédé d'estimation de canal et récepteur correspondant | |
EP0808538B1 (fr) | Dispositif de reception de signaux numeriques a structure iterative, module et procede correspondants | |
EP1180289A1 (fr) | Procede de communications radiomobiles amrt iteratif | |
EP1212872A1 (fr) | Recepteur a multiplexage par repartition en frequences orthogonales avec estimation iterative de canal et procede correspondant | |
EP1378089A1 (fr) | Procede de codage/decodage d'un flux de donnees numeriques codees abec entrelacement sur bits en emission et un reception multiple en presence d'interference intersymboles et systeme correspondant | |
EP2274852A2 (fr) | Procédé de décodage d'un signal mettant en oeuvre une construction progressive d'un arbre de décodage, produit programme d'ordinateur et dispositif de décodage correspondants | |
FR2913161A1 (fr) | Decodeur a maximum de vraisemblance pour systeme multi-source a modulation de position d'impulsion et d'amplitude | |
EP1168739A1 (fr) | Procédé et dispositif d'estimation de la réponse impulsionelle d'un canal de transmission d'informations, en particulier pour un téléphone mobile cellulaire | |
FR2825551A1 (fr) | Procede d'estimation de la fonction de transfert d'un canal de transmission d'un signal multiporteuse, procede de reception d'un signal numerique, et recepteur d'un signal multiporteuse correspondants | |
FR2783120A1 (fr) | Procede d'egalisation numerique, et recepteur de radiocommunication mettant en oeuvre un tel procede | |
FR2805102A1 (fr) | Procedes et dispositifs d'emission et de reception d'information, et systemes les mettant en oeuvre | |
EP1105975B1 (fr) | Recepteur en rateau iteratif et procede de reception correspondant | |
WO2007028917A2 (fr) | Estimation iterative de canal de mimo dans un reseau cdma | |
EP3912317B1 (fr) | Procédé de réception d'un signal soqpsk-tg en décomposition pam | |
EP1213884B1 (fr) | Procédé et dispositif d'estimation des valeurs successives de symboles numériques, en particulier pour l'égalisation d'un canal de transmission d'informations en téléphonie mobile | |
EP0704981A1 (fr) | Dispositif de synchronisation de branches d'un décodeur de viterbi compris dans un récepteur de données numériques codées en treillis multidimensionnel | |
EP0648037B1 (fr) | Asservissement de phase, en bande de base | |
EP1212873B1 (fr) | Procede de decodage et d'egalisation conjointe d'un signal numerique protege par un code defini par un treillis | |
EP0858196A1 (fr) | Dispositif d'égalisation et de décodage de canal | |
EP1211857A1 (fr) | Procédé et dispositif d'estimation des valeurs successives de symboles numériques, en particulier pour l'égalisation d'un canal de transmission d'informations en téléphonie mobile | |
FR2767983A1 (fr) | Procede de decodage et de synchronisation de phase simultanes exploitant le critere de maximum de vraisemblance et dispositif correspondant |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
17P | Request for examination filed |
Effective date: 20011009 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE |
|
GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
18D | Application deemed to be withdrawn |
Effective date: 20070613 |