JP2009081846A - System and receiver for receiving a P-length vector of a received signal via a nested channel - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for minimizing a complication degree of a system for transmitting information data via a nested channel. <P>SOLUTION: In the receiver of a P-length vector, a nested channel has M time slots, the P-length vector of transmission signal carries an M-length output vector associated with one transmission via the nested channel, some components of the M-length output vector are obtained by liner combination of s modulation symbols of M-length vector called combined modulation symbols, and other components are modulation symbols of the M-length vector of modulation symbols called non-combined modulation symbols. The receiver has a first detector which outputs an estimated value of the s combined modulation symbols of the M-length vector of the modulation symbol, the value allowing to recover the diversity order and a second detector which outputs an estimated value of other M-s pieces of non-combined modulation symbols. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a system for transmitting information data from a transmitter to a receiver via a nested channel. The invention also relates to a receiver for such a system.

  In the following, channel resource means frequency bandwidth, time interval, and possibly some spatial dimensions provided by transmit and receive antennas located at various spatial locations.

  Signals transmitted over the wireless channel are severely degraded by channel variations such as fading, shadowing, and interference from other transmitters. This makes it possible to regard the channel as a random variable. In the following, gradual channel variation is considered for the time required to transmit an information word, but the channel realization should have changed between two transmissions of the information word. One main way to combat so-called quasi-static fading is to provide diversity in either time, frequency, or space.

  The channel diversity order is defined as the number of independent fading random variables observed in the channel resource. A transmission / reception scheme can collect a given diversity called the diversity order of the system. The upper limit of the diversity order of this system is suppressed by the channel diversity order, also called the full diversity order.

  An error event occurs when the information word is not correctly estimated by the receiver. The main parameters that make it possible to calculate the probability of error associated with a given error event are the noiseless received signal associated with the transmitted information word and the recovered noiseless received signal associated with the decoded information word Euclidean distance between. The error event diversity order is defined as the number of independent random variables involved in the Euclidean distance associated with the error event. Finally, the diversity order of the system is equal to the minimum diversity order of all possible error events. That is, equal to the minimum diversity order of all possible information word pairs.

  Multiple antenna systems that provide high orders of spatial diversity and high capacity have been extensively studied. However, due to the expensive analog radio frequency components, the number of antennas in a mobile terminal is often limited to one transmit radio frequency path and two receive radio frequency paths. In such a mobile terminal, a high diversity order of a transmission signal cannot be obtained on a narrowband channel such as an OFDM-based (orthogonal frequency division multiplexing) transmission system.

  Therefore, the concept of “cooperative communications” where multiple terminals use each other's capabilities was introduced. This makes it possible for the signal to obtain a high diversity order, and the signal transmitted by other terminals causes the paper “User cooperation diversity. Part I. system description” (IEEE Transactions on Communications, vol. 51, no. 11, pp. 1927-1938, Nov. 2003 '), a virtual antenna array as proposed by Sendonaris et al. is obtained. Hereinafter, the concept of the cooperative transmission protocol will be described. This is one application of the invention described below.

  One possible application of such a system is a wireless ad hoc network, such as a mesh network, which does not rely on a central control unit and does not have a fixed infrastructure. Nodes communicate by forming a network based on channel conditions and mobile location.

  Another application of such a system is the coordination of users within a cell of a cellular system. Reliable communication can be achieved through diversity and by relaying signals from terminals remote from the base station. The advantage of such a system compared to conventional systems is that the more users are on the network, the more reliably users can communicate. This is a result of the non-rigidity of the collaborative system infrastructure. In contrast, the rigidity of the non-cooperative system infrastructure involves increasing the blocking probability with the number of terminals sharing the network.

  The disadvantage of the cooperative system is that the channel between users has noise. In order to disguise this drawback, a plurality of cooperation protocols have been developed, which define a method for cooperation between users.

  Cooperative protocols can be classified into two main categories: Amplify-and-Forward (AF) method and Decode-and-Forward (DF) method. In the following, the focus is on the AF protocol.

  A repeater is, for example, a user mobile radio device or a device used to expand the range of cells. Using the AF protocol, the repeater only amplifies the signal received from the source (base station or another repeater) before transferring it to the destination (user equipment or another repeater). . These protocols are easy to implement in practice because the computational complexity that the protocol brings to the repeater is limited to scaling operations.

  Multiple AF protocols are designed for the case of a single repeater, such as the non-orthogonal amplify-and-forward (NAF) protocol, also known as TDMA-based protocol I Yes. In this case, the transmission source broadcasts a signal to both the repeater and the transmission destination in the first phase. In the second phase, the repeater scales the signal and forwards it to the destination, while the sender sends another message to the destination (“Fading relay channel: performance limits and space-time signal design” (IEEE Journal on Selected Areas in Communications, vol. 22, no. 6, pp. 1099-1109, August 2004).

  The following will focus on the case where multiple repeaters are used by the transmission system. The selected transmission protocol transmits a signal over each time slot of a predetermined set of M time slots.

  For example, the known Slotted Amplify-and-Forward (SAF) protocol can be used. However, the present invention is not limited to transmission via a half-duplex non-orthogonal amplification transfer cooperative channel.

  FIG. 1 is a schematic diagram showing a transmission system according to the present invention. The system SYST comprises a transmitter TRD, a nested channel NC comprising M time slots and described by an M × P channel matrix H, and a receiver RCV. The transmitter TRD is for example a base station and the receiver RCV is for example a mobile user equipment.

  The transmitter TRD includes an encoder ENC according to an error correction code structure, a binary modulator MOD, and a linear combiner LC. Roughly speaking, digital transmission occurs as follows. That is, the information data bits {b} to be transmitted are supplied to the encoder ENC at the rate Rc = Ki / Ko. Here, Ki is the number of input data bits {b}, and Ko is the number of encoded bits of the output codeword {c}. The encoder ENC can follow any type of error correction code structure, such as LDPC (low density parity check) code, turbo code, block code (eg Reed-Solomon), binary convolutional code, etc. . The encoded bits are then supplied to the binary modulator MOD.

The binary modulator MOD is, for example, BPSK (2 phase shift keying) or QPSK (4 phase shift keying), preferably 2 ^m quadrature amplitude modulation (2 ^m -QAM). The input of the binary modulator MOD is the encoded bits {c}, or according to a variant of the system SYST, an interleaved version {c} of such encoded bits. In a variant of this system SYST, as shown in FIG. 1, the transmitter TRD comprises a bit interleaver INT and the receiver RCV comprises an associated bit deinterleaver DINT. The output of the binary modulator MOD is a vector of modulation symbols. This modulation symbol vector is divided into M-length vectors Z of modulation symbols Z (i).

  The modulation symbols Z (i) in the same vector Z are then combined with each other by the linear combiner LC. The linear combiner LC provides an output vector X of M combined symbols associated with one transmission over the nested channel NC.

  The combined symbol X (i) of each vector X is obtained by multiplying the vector Z by the M × M complex linear combination matrix S and transmitted via the nested channel NC. Here, it should be noted that the matrix S may be a unit matrix in some cases.

  In the following, the modulation symbol Z (i) of the M-length vector Z used to define the linear combination will be referred to as a combined modulation symbol and M not used to define such a linear combination. The modulation symbol of the long vector Z is called a non-combined modulation symbol.

The nested channel NC is defined from a set of N independent fading random variables a (j), and the norm of the i-th row of the channel matrix H is the portion of D (i) random variables a (j). It is a function of the set Γ (i). The diversity order of uncoded modulation performance transmitted in the i th row of the M × P matrix H is equal to D (i). Without loss of generality, the rows of the matrix are sorted such that i ₁ ≦ i ₂ ⇒Γ (i ₁ ) ⊇Γ (i ₂ ) and D (i ₁ ) ≧ D (i ₂ ).

  According to one embodiment of the invention, M = P and the nested channel NC is

Is a sliding-triangular channel described by the upper triangular matrix H given by

here

Denotes a product for each term, and a1, a2,..., AM are coefficients of fading random variables that define the diversity characteristics of the channel. The γi, j value is a random variable or a fixed parameter. A channel is equivalent to a sliding triangular channel if the configuration of fading coefficients is the same.

  The modulation symbol Z (j) is confused by other channel imperfections besides the channel imperfection associated with the random variable aj, eg additive noise, other multiplicative random variables Please note that there is.

  According to one embodiment of the sliding triangular channel, this channel is defined by the SAF protocol for cooperative transmission with several repeaters.

  FIG. 2 shows the inter-repeater communication protocol of the SAF channel model. Note that in practice, an attenuated version with noise of the combined symbol X (j) is received at the repeater and receiver. According to this example, β repeaters are used. This entails that one use of this channel is chosen such that it contains (β + 1) time slots Tsi of the SAF protocol and the size S of the complex linear combination matrix is M = β + 1. During the first time slot TS1 of the SAF protocol, the transmitter TRD transmits the first symbol X (1). This first symbol X (1) is received by the first repeater R (1) and the receiver RCV. During the second time slot TS2 of the general SAF protocol, the transmitter TRD transmits the second symbol X (2), and the relay R (1) transmits the previously received symbol X (1). Send version. These symbols are received by the receiver RCV and the second repeater R (2). During the third time slot TS3 of the general SAF protocol, the transmitter TRD transmits the third symbol X (3) and the repeater R (2) receives the previously received and combined symbol X ( Send 1) and X (2) versions. These symbols are received by the receiver RCV and the third repeater R (3). During the last time slot TS (β + 1) of the general SAF protocol, the transmitter TRD transmits the last symbol X (β + 1) and the last repeater R (β) was received and combined earlier. The version of the symbol X (1),..., X (β) is transmitted. These symbols are received by the receiver. As a result, the version of the first symbol X (1) is transmitted during each of (β + 1) time slots in one channel use. The version of the i-th symbol xi is transmitted during the last (β + 1) -i + 1 time slots. The version of the last symbol X (β + 1) is received by the receiver RCV only during the last time slot of the SAF protocol. The relay R (1) receives the version of the first symbol xi during the first time slot TS1 and retransmits it during the second time slot TS2. The i-th repeater R (i) receives the combined version of the first i symbols xi during the i-th time slot and re-transmits it during the i + 1-th time slot. Transmit (the i-th repeater receives the signal transmitted by the transmitter and the (i-1) -th repeater). As a result, the encoded data bit estimate carried by the first symbol X (1) has a diversity order D (β + 1) as perceived by the decoder DEC of the receiver RCV. The estimate of the encoded data bits carried by the i-th symbol X (i) has a diversity order D (β + 2-i) perceived by the decoder DEC. The estimate of the encoded data bits carried by the last symbol X (β + 1) has a diversity order D (1) perceived by the decoder DEC. The receiver RCV receives the version of the first symbol X (1) during the first time slot TS1 and the combined version of the first i symbols X (j) during the time slot Tsi. And the combined version of all (β + 1) symbols X (j) during the last time slot.

  Mathematically referring to FIG. 1, the signal model is

Given by.

Here, Yd is the M length vector of the received signal, Z is the M length vector of the modulation symbol Z (i), Wc is the additional noise M length vector, and X is the M length of the combined symbol X (i). Output vector.

  The mathematical model of the signal transmitted between the transmitter TRD and the receiver RCV when the nested channel NC is a SAF channel and β repeaters R (i) are considered is

Given by.

Here, i = 1,..., B + 1. The subscript sr (representing the transmission source) corresponds to the transmitter TRD, d (representing the transmission destination) corresponds to the receiver RCV, and ri (represents the repeater) corresponds to the i-th repeater. The complex symbol X (i) of variance 1 is transmitted in the i-th time slot, the received signal in the i-th time slot at the receiver is Yd (i), and Yr (i) is transmitted by the i-th repeater Ri. It is a received signal. The coefficient ε i represents the energy transmitted in the i th time slot by the transmitter TRD. hk, l is a fading coefficient, and Wd (i) and Wr (i) are additive white Gaussian noise (AWGN) components. γi is an energy normalization coefficient in the i-th repeater Ri on condition that E | γiγri−1 | ² ≦ 1 and γ0 = 0. From this channel model, it can be seen that the SAF channel has the same block diversity characteristics as the sliding triangular channel.

  The receiver RCV comprises a detector DET and a decoder DEC.

The detector DET has a hard output estimate or a soft output estimate of the coded bits associated with the M length vector Z of the modulation symbol Z (i) carried by the P length vector Y _d . one of the estimates).

  The hard output estimate is an estimate of the encoded bits, and its value is equal to either 0 or 1. The soft output estimate relates to the probability that the encoded bits are equal to one.

A maximum likelihood (ML) detector provides the best hard output estimate for the encoded bits. Furthermore, to maximize the likelihood p (Y _d | Z), ie

According to the symbol

To find, or to minimize the performance index such as the Euclidean distance ^{_{‖Y d -Z × S × H‖ 2}} , i.e.

Symbol to minimize

When an exhaustive search is performed on all possible candidate vectors Z to find the optimal detection performance can be achieved with the ML detector.
Thus, an estimate of the encoded bits associated with the M length vector Z of the modulation symbol Z (i) carried by the P length vector Y _d

Is a symbol

Obtained from.

The complexity of such an exhaustive search cannot be handled quickly because the number of possible candidate vectors Z is actually enormous. For example, if the binary modulator MOD is 2 ^m quadrature amplitude modulation, the exhaustive search proceeds for 2 ^mM candidates and becomes unhandled as the number mM increases.

  There are several receivers with reduced complexity but best, such as Sphere Decoder ([64] by E. Viterbo and J. Boutros, “A universal Lattice Decoder for Fading Channels” (IEEE Trans. Inform. Theory, vol. 45, pp. 1639-1642, July 1999)), it is still too complex for low-cost communication devices.

A maximum posterior probability (MAP) detector provides a soft output estimate in the form of a probability for the coded bits associated with the M length vector Z of the modulation symbol Z (i) carried by the P length vector Y _d . Often used for.

  An exhaustive search over all possible candidate vectors Z maximizes the figure of merit function that minimizes the error probability of each coded bit cj associated with the M-length vector Z of the modulation symbol Z (i) MAP detector achieves optimal detection performance. For example, the estimated value is extrinsic probability or posterior probability (A Posteriori Probability).

  The a posteriori probability (APP) for the encoded bit ci is the following exhaustive marginalization:

Calculated by

here

N0 is the noise variance, Ω (ci = 1) is the set of possible candidate vectors Z labeled with the i th bit equal to 1, and Ω is for all possible candidate vectors Z Is a set, and f (j, Z) is the value of the jth bit of the binary labeling associated with the possible candidate vector Z.

  Thus, an estimate of the encoded bits associated with the M length vector Z of the modulation symbol Z (i) carried by the M length vector Yd

Is obtained from the maximum posterior probability (APP) for each encoded bit ci.

  Note that the soft output estimate for the encoded bit ci is

Often expressed as a log-likelihood ratio (LLR) given by

  According to a variant of the MAP detector, list-based marginalization ("Soft-input / soft-output lattice sphere decoder for linear channels" by J. Boutros, N. Gresset, L. Brunel, and M. Fossorier) ( Proc. GlobeCom, San Francisco, Dec. 2003)) is used to select a subset of all possible vectors of symbols and reduce computational complexity.

  The decoder DEC has such an estimate

Is an estimate of the transmitted information data bits {b}

Convert to

  Alternatively, to reduce the complexity of the receiver, a sub-optimal detector such as a linear equalizer is used to estimate non-optimal encoded bits.

Is often used to generate Unfortunately, with these types of detectors, it is not possible to observe the diversity order caused by the linear combination at the detector output.

  In the following, an optimal detector means a detector that can recover the diversity caused by the linear combination. A sub-optimal detector is also referred to as a detector that cannot recover the diversity caused by the linear combination.

  For example, consider a symbol-based zero-forcing (ZF) hard output detector. Estimates for the coded bits associated with the modulation symbol Z (j) of the M-length vector Z of modulation symbols are calculated independently of each other.

Here, Dec (.) Is a decision function, Sj is the jth row of the matrix S, and t indicates the conjugate transpose of the matrix.

  Alternatively, the detector is a ZF per vector, and estimates for the encoded bits associated with the M uncombined modulation symbols of the M-length vector Z of modulation symbols are computed together.

  The symbol-wise detector or vector-wise detector presented above can be classically an MMSE linear equalizer, serial interference canceller (SIC), or parallel interference canceller (PIC). In addition, a soft-bit conversion that converts an estimate for symbols to an estimate for encoded bits can be used instead of the decision module Dec ().

  The problem solved by the present invention is to optimize the transmit / receive system SYST in order to minimize the complexity of the receiver RCV and at the same time obtain the performance indicating the target diversity order.

Indeed, according to the present invention, a receiver of a P-length vector of a received signal transmitted by a transmitter via a nested channel, the nested channel comprising M time slots, M × The P-length vector of the received signal described by a P-channel matrix carries an M-length output vector associated with a single transmission over a nested channel, and several components of the M-length output vector called combined symbols Is a linearly combined modulation symbol (X (i)), which is obtained by a linear combination of s modulation symbols of M length vector called combined modulation symbol. The other components of the M-length output vector that are acquired and referred to as non-combined modulation symbols that are not used to define such a linear combination. Is a modulation symbol Torr, the receiver includes a detector and a decoder, the receiver,
A first detector, the output of which is an estimate for the coded bits associated with the s combined modulation symbols of the M-length vector of the modulation symbols, the first detector comprising: A first detector that makes it possible to recover the diversity order s given by the linear combination of the estimated values;
A second detector, the output of which is an estimate of the coded bits associated with the other M-s uncombined modulation symbols of the M-length vector of the modulation symbols And
It is characterized by providing.

  According to one embodiment of the detector, the output estimate of the first detector is obtained before the output estimate of the second detector.

  According to a variant of this detector embodiment, the output estimate of the first detector is taken as an additional input of the second detector.

  According to another embodiment of the detector, the output estimate of the second detector is obtained before the output estimate of the first detector.

  According to a variant of this alternative embodiment of the detector, the output estimate of the second detector is taken as an additional input of the first detector.

  According to another embodiment of the detector, the output estimate of the first detector is taken as an additional input of the second detector, and the output estimate of the second detector is the first detection. It is taken as an additional input of the instrument.

  According to a variant of this embodiment, the final estimate of the combined symbol and the non-combined symbol is calculated more than once for the first detector and more than once for the second detector. Obtained according to the estimate calculation.

  According to an embodiment of the first detector, the output estimate of the first detector does not take into account interference from Ms uncoupled modulation symbols of the M length vector of the modulation symbols.

  According to another embodiment of the first detector, the output estimate of the first detector takes into account interference from Ms uncoupled modulation symbols of the M-length vector of the modulation symbols.

  According to another embodiment of the first detector, the output estimate of the first detector is a hard output estimate for the coded bits associated with the s combined modulation symbols, all Is obtained from a search with reduced complexity compared to an exhaustive search over all possible candidate modulation vectors.

  According to another embodiment of the first detector, the output estimate of the first detector is a soft output estimate for the coded bits associated with the s combined modulation symbols, all Is obtained from a calculation of marginalization with reduced complexity compared to the comprehensive marginalization calculated over all possible candidate modulation vectors.

  According to one embodiment of the second detector, the output estimate of the second detector does not take into account interference from the s combined modulation symbols of the M-length vector of the modulation symbols.

  According to a variant of this embodiment, the estimated values for the coded bits associated with the non-combined modulation symbols of the M-length vector of the modulation symbols are calculated independently of each other.

  According to another variant of this embodiment, the estimated values for the coded bits associated with the non-combined modulation symbols of the M-length vector of the modulation symbols are calculated together.

  The present invention also relates to a system for transmitting information data from a transmitter to a receiver as described above.

  In addition to the features of the present invention described above, other features will become more apparent from the following description, given in conjunction with the accompanying drawings.

  Generally speaking, the M × M complex linear combination matrix S is given by the matrix multiplication P1 × S ′ × P2. Here, P1 is a row permutation matrix, P2 is a column permutation matrix, and S ′ is

Defined as a block diagonal matrix given by.

Here, I is an (M−s) × (M−s) unit matrix, and the submatrix S ″ is an s × s complex linear combination matrix that satisfies the following characteristics. That is, consider an s × s diagonal fading channel whose diagonal elements are independent fading random variables. Measured at the output of the detector DET if an optimal hard output estimate is calculated (maximum likelihood) to receive s symbols linearly combined by the matrix S ″ and transmitted on the diagonal fading channel The performance achieved shows a diversity order that is s times higher than the performance without the linear combination matrix S ″. This is achieved by algebraic rotation, such as the known cyclotomic rotation.

  The parameter s of the linear coupler LC is called the coupling size of the linear coupler. If 0 <s <M, the linear combiner LC is a linear combination of the modulation symbols Z (j) of the M-length vector Z, only the part of the above, ie some symbols X (i) of the output vector X. It is. It should be noted that the present invention is also relevant for the choice of s = 1, i.e. no linear combination.

  Without loss of generality, according to the P1 matrix, in the following, it is assumed that the combined symbols are always in the s first positions. Similarly, according to the P2 matrix, in the following, the symbols X (1), X (M−s + 2),..., X (M) of the output vector X are the first s of the M length vector Z of the modulation symbol Z (i). Consider a linear combination of modulation symbols (Z (1),..., Z (s)). The other (M−s) symbols X (2),..., X (M−s + 1) of the output vector X are the same as the modulation symbols Z (s + 1),. It is.

  According to one embodiment of the system SYST, the coding rate Rc of the encoder ECN, the parameter M of the nested channel, and the linear combination size s are selected depending on each other to achieve the target diversity order of the system SYST. Is done. For example, if the nested channel is a sliding triangular channel, the four variables have the relationship

Tied together by

  For parameters M and Rc fixed and high spectral efficiency modulation (eg, 16-QAM), several values of s may result in the same target diversity order s. The receiver complexity is minimized by minimizing s. For a sliding triangle channel this is

It becomes.

According to the invention, the detector DET is a low complexity near-optimal detector and is related to the M length vector Z of the modulation symbol Z (i) carried by the P length vector Y _d . An estimate for the encoded bits to be generated is generated. A detector is said to be near-optimal if it can achieve the same diversity order performance after decoding as it does for an optimal detector.

The detector DET is intended to estimate the encoded bits associated with the s combined modulation symbols of the M length vector Z of the modulation symbols Z (i) carried by the P length vector Y _d . A detector DET1 and a second detector DET2 intended to estimate the encoded bits associated with the (M−s) non-combined modulation symbols of the M-length vector Z of the modulation symbols. . Note that these detectors can be processed in series or in parallel, as described below, and can be processed once or repeatedly.

According to one embodiment of the detector DET, the output estimate of the detector DET1 is obtained before the output estimate of the detector DET2. Estimate of s combined modulation symbols are obtained from the detector DET1, the output estimate of the detector DET1, may cancel the contribution of these combined symbols from the P-length vector Y _d. That is, the interference level sent by DET2 can be reduced.

  According to a variant of this embodiment, the output estimate of the detector DET1 is taken as an additional input of the detector DET2.

According to another embodiment of the detector DET, the output estimate of the detector DET2 is obtained before the output estimate of the detector DET1. An estimate of (M−s) uncoupled modulation symbols is obtained from detector DET2, and this output estimate of detector DET2 can cancel the contribution of these uncoupled symbols from P-length vector Y _d . That is, the interference level sent by DET1 can be reduced.

  According to a variant of this embodiment, the output estimate of the detector DET2 is taken as an additional input of the detector DET1.

  According to another embodiment of the detector DET, the output estimate of the detector DET1 is taken as an additional input of the detector DET2, and similarly, the output estimate of the detector DET2 is taken as an additional input of the detector DET1. It is done. The estimate of the non-combined symbol helps reduce the interference seen from the detector DET1, and the estimate of the combined symbol helps reduce the interference seen from the detector DET2. The information provided by these additional inputs is used, for example, to subtract the contribution of uncombined symbols in the received vector Yd before applying the detector DET1, or before applying the detector DET2. Used to subtract the contribution of the combined symbol in the received vector Yd.

  According to a variant of this embodiment, the final estimated values of the combined symbol and the non-combined symbol are calculated two or more estimates of the combined symbol and two or more estimates of the non-combined symbol. Get according to.

  According to a feature of the invention, the first detector DET1 provides an estimate of the encoded bits associated with the s combined modulation symbols by means of a linear combiner LC, ie a submatrix S ″. It is possible to recover the diversity order. When the matrix S is an identity matrix, i.e. processed with a non-linear combination, the first detector DET1 estimates the coded bits associated with the first modulation symbol Z (1) of the M-length vector Z I will provide a.

  According to one embodiment of the detector DET1, an estimate for the coded bits associated with the s combined modulation symbols of the modulation symbol M length vector Z is the (M− s) Do not consider interference from uncoupled modulation symbols.

  According to a variant of this embodiment, the estimate for the coded bits associated with the s combined modulation symbols is preferably

Is based on the calculation of a function of the incomplete likelihood probability p (Yd | Zs, 0) given by

Here, N0 is noise variance, and Zs, 0 is an M length vector Z of modulation symbols in which (M−s) non-combined symbols are equal to zero. That is, Zs, 0 = [Z (1),..., Z (s), 0,.

  According to another embodiment of the detector DET1, an estimate for the coded bits associated with the s combined modulation symbols of the modulation symbol M length vector Z is the (M) of the modulation symbol M length vector Z (M -S) Consider interference from uncoupled modulation symbols.

  According to a variant of this embodiment, the estimate for the coded bits associated with the s combined modulation symbols is preferably

Is based on the calculation of the function of the incomplete likelihood probability p (Yd | Zs, 0) given by

Here, N0 is noise variance, and Z0 and M−s are M-length vectors Z of modulation symbols in which s combined symbols are equal to zero. That is, Z0, Ms = [0,..., 0, Z (s + 1),..., Z (M)].

According to one embodiment of the detector DET1, an estimate for the coded bits associated with the s combined modulation symbols of the modulation symbol M length vector Z is the (M− s) Consider or do not consider interference from uncoupled modulation symbols. Detector DET1 provides a hard output estimate from a search performed over 2 ^ms M-length vectors Z of modulation symbols instead of 2 ^mM vectors of modulation symbols as in the prior art.

  According to a variant of this embodiment, the detector DET1 is a maximum likelihood (ML) detector as described in the opening paragraph, and the likelihood probability p (Yd | Z) of equation (1) is Replaced by the incomplete likelihood probability p (Yd | Zs, 0) calculated according to either equation (6) or (7).

  Since a lower dimensionality (s instead of M) is considered, the complexity of the sphere decoder is also reduced. In fact, it is known that the complexity of the sphere decoder is a polynomial of dimensionality.

  According to another variation of this embodiment, LLL lattice reduction is used to reduce the computational complexity of DT1 while maintaining the diversity utilization capability of DET1. In this case, the detector DET1 is a near-optimal hard output detector and the complexity of the linear combination (LLL reduction) is reduced as the number of dimensions is reduced.

  According to one embodiment of detector DET1, detector DET1 provides a soft output estimate.

According to a variant of this embodiment, the detector DET1 is a maximum posterior probability (MAP) detector as described in the opening paragraph, and the likelihood probability p (Yd | Z) of equation (2) is Replaced by the incomplete likelihood probability p (Yd | Zs, 0) calculated according to either equation (6) or equation (7). Here, marginalization is performed over 2 ^ms possible candidates for the combined modulation symbol s-length vector instead of the 2 ^mM vector of modulation symbols.

The MAP detector ensures that the diversity order s of the coded bits associated with the s combined modulation symbols of the M-length vector Z of modulation symbols is recovered. Also, marginalization performed over 2 ^ms possible candidate vectors Z is accompanied by a significant reduction in computational complexity compared to the exhaustive search that is usually handled in practice.

According to another variation of this embodiment, the marginalization is a list-based marginalization (eg, a list sphere decoder) over a subset of 2 ^ms possible candidate vectors that are s-length vectors of the combined symbols. ). In this case, the detector DET1 is a probability sub-optimal soft output detector, and the complexity of the linear combination and the size of the list are reduced by the lowest number of dimensions (s instead of M). Is done.

  According to one of the previous two variants, the soft output estimate for the encoded bits associated with the s combined symbols is the encoded bits of the s combined modulation symbols. Consider the prior probability π (cj) that is relevant and given by the output of the decoder DEC.

  This modification makes it possible to improve the detection performance of the receiver RCV.

  According to another variant of one of these previous two variants, the soft output estimate for the coded bits associated with the s combined symbols is associated with the coded bits of all modulation symbols. And consider the prior probability π (cj) given by the output of the decoder DEC.

  According to another feature of the invention, the complexity of the second detector DET2 is low compared to the complexity of the first detector DET1, and is preferably negligible.

  According to one embodiment of the second detector DET2, the estimated value for the coded bits associated with the Ms non-combined modulation symbols of the modulation symbol M length vector Z is the M length of the modulation symbol. It does not consider interference from s combined modulation symbols of vector Z.

  According to a variant of this embodiment, estimates for the encoded bits associated with the combined modulation symbol Z (s + j) of the M-length vector Z of modulation symbols are calculated independently of each other.

When the binary modulator MOD is 2 ^m quadrature amplitude modulation, the estimate for the coded bits associated with the uncombined modulation symbol Z (s + j) is

Given by.

Here, Dec (.) Is a decision function, S _{s + j} is the s + jth row of the matrix S, and t indicates the conjugate transpose of the matrix.

  In this case, the detector DET2 is a linear detector based on the decision function given by equation (8). The linear detector provides either a hard output estimate or a soft output estimate. For example, the hard output estimation value is provided using an MMSE linear equalizer, a serial interference canceller (SIC), and a parallel interference canceller (PIC) as described in the first paragraph.

  According to another variant of this embodiment of the second detector DET2, the estimated values for the coded bits associated with the Ms uncombined modulation symbols of the M-length vector Z of modulation symbols are: Calculated.

When the binary modulator MOD is 2 ^m quadrature amplitude modulation, the estimate for the coded bits associated with the uncombined modulation symbol is

Given by.

Here, S _{s + j; M} is a matrix obtained by selecting M from row s + j for matrix S.

  In this case, the detector DET2 is a vectorial detector based on the estimation function given by equation (8). The vector detector provides either a hard output estimate or a soft output estimate. For example, the hard output estimation value is provided using an MMSE linear equalizer, a serial interference canceller (SIC), and a parallel interference canceller (PIC).

  According to a variant of this embodiment of the detector DET2 for soft output estimate calculation, the soft output estimate for the coded bits associated with the Ms uncoupled modulation symbols is iteratively calculated and the detector Consider the prior probability π (cj) associated with the coded bits of the M−s uncoupled modulation symbols given by the output of DEC (“Turbo-equalization for multilevel modulation by A. Dejonghe and L. Vanderdope” : an efficient low-complexity scheme "(Proc. of the IEEE ICC 2002, vol. 25, no. 1, pp. ~ 1863-1867, April 2002)).

1 is a schematic diagram illustrating a transmission system according to the present invention. It is a figure which shows the communication protocol between repeaters of a SAF channel model.

Claims (29)

A receiver of a P-length vector of a received signal (Y _d ) transmitted by a transmitter via a nested channel (NC),
The nested channel (NC) comprises M time slots and is described by an M × P channel matrix (H),
The P length vector of the received signal (Y _d ) carries an M length output vector (X) associated with one transmission via the nested channel (NC),
Some components of the M-length output vector (X), called combined symbols, are linearly combined modulation symbols (X (i)), and the linearly combined modulation symbols (X (i)) are combined Obtained by linear combination of s modulation symbols (Z (i)) of M length vector (Z), called modulation symbols,
The other components of the M-length output vector (X) are modulation symbols of the M-length vector (Z) of the modulation symbols, called uncombined modulation symbols, that are not used to define such a linear combination;
The receiver (RCV) comprises a detector (DET) and a decoder (DEC),
The receiver
A first detector (DET1), the output of which is an estimate for the encoded bits associated with the s combined modulation symbols of the M-length vector (Z) of the modulation symbols; A first detector (DET1) that makes it possible to recover the diversity order given by the linear combination for the estimate;
A second detector (DET2), the output of which is an estimate for the encoded bits associated with the other M-s uncoupled modulation symbols of the M-length vector (Z) of the modulation symbols; A second detector (DET2);
A receiver of a P-length vector of a received signal (Y _d ) transmitted by a transmitter via a nested channel (NC).   Receiver according to claim 1, characterized in that the output estimate of the first detector (DET1) is obtained before the output estimate of the second detector (DET2). .   Receiver according to claim 2, characterized in that the output estimate of the first detector (DET1) is taken as an additional input of the second detector (DET2).   The receiver according to claim 1, characterized in that the output estimate of the second detector (DET2) is obtained before the output estimate of the first detector (DET1). .   The receiver according to claim 4, characterized in that the output estimate of the second detector (DET2) is taken as an additional input of the first detector (DET1).   The estimated output value of the first detector (DET1) is taken as an additional input of the second detector (DET2), and the estimated output value of the second detector (DET2) is Receiver according to any one of claims 1, 2 or 4, characterized in that it is taken as an additional input of one detector (DET1).   Final estimates of combined and non-combined symbols are calculated for two or more estimates of the first detector (DET1) and for two or more estimates of the second detector (DET2). The receiver according to claim 6, wherein the receiver is obtained according to a calculation.   The output estimate of the first detector (DET1) does not take into account interference from the Ms uncoupled modulation symbols of the M-length vector (Z) of the modulation symbols. Item 5. The receiver according to any one of Items 1 to 4. The output estimate of the first detector (DET1) is

Is a function of the incomplete likelihood probability p (Y _d | Z _{s, 0} ) given by
Where N ₀ is the noise variance, Z _{s, 0} is the M-length vector (Z) of the modulation symbol in which the Ms uncoupled symbols are equal to 0, and S is the modulation symbol 9. The M × M complex linear combination matrix used to obtain the M length output vector (X) by multiplying the M length vector (Z) by the matrix S. The listed receiver.   The output estimate of the first detector (DET1) takes into account interference from the Ms uncoupled modulation symbols of the M-length vector (Z) of the modulation symbols. Item 8. The receiver according to any one of Items 4 to 7. The output estimate of the first detector (DET1) is

Is a function of the incomplete likelihood probability p (Y _d | Z _{s, 0} ) given by
Here, N ₀ is noise variance, Z _{0, Ms} is the M length vector (Z) of the modulation symbol in which the s combined symbols are equal to 0, and S is the M length of the modulation symbol. Reception according to claim 10, characterized in that it is an MxM complex linear combination matrix used to obtain the M-length output vector (X) by multiplying the vector (Z) by the matrix S. Machine.   The output estimate of the first detector (DET1) is a hard output estimate for the coded bits associated with the s combined modulation symbols, and an exhaustive search across all possible candidate modulation vectors Receiver according to any one of claims 8 to 11, characterized in that it is obtained from a search with reduced complexity compared to.   The receiver according to claim 12, characterized in that an LLL lattice reduction is used in the detector DET1.   The receiver according to claim 12, characterized in that the first detector DET1 is a maximum likelihood (ML) detector.   The receiver according to any one of claims 8 to 11, characterized in that the first detector (DET1) is a maximum posterior probability detector.   The output estimate of the first detector (DET1) is a soft output estimate for the coded bits associated with the s combined modulation symbols and is calculated over all possible candidate modulation vectors 12. Receiver according to any one of claims 8 to 11, characterized in that it is obtained from a calculation of marginalization with reduced complexity compared to global marginalization. The output estimate of the first detector (DET1) is a soft output estimate for the encoded bits associated with the s combined symbols, and the encoded bits of the s combined symbols The receiver according to claim 15, characterized in that it takes into account the prior probability (π (c _j )) that is related to and given by the output of the decoder (DEC). The output estimate of the first detector (DET1) is a soft output estimate for the encoded bits associated with s combined symbols and is associated with the encoded bits of all modulation symbols. 17. A receiver according to claim 15 or 16, characterized by taking into account the prior probability (π (c _j )) given by the output of the decoder (DEC).   The output estimate of the second detector (DET2) does not take into account interference from the s combined modulation symbols of the M length vector (Z) of the modulation symbols. The receiver according to claim 2 or any one of claims 4 to 18.   20. Receiver according to claim 19, characterized in that the estimates for coded bits associated with non-combined modulation symbols of the M-length vector (Z) of the modulation symbols are calculated independently of each other.   20. Receiver according to claim 19, characterized in that the estimated values for the coded bits associated with the uncombined modulation symbols of the M-length vector (Z) of the modulation symbols are calculated together. The output estimate of the second detector (DET2) is a soft output estimate, and the soft output estimate is iteratively calculated and the encoded of the Ms uncoupled modulation symbols Receiver according to any one of claims 19 to 21, characterized in that it takes into account prior probabilities (π (c _j )) associated with bits and given by the output of the decoder (DEC). A system for transmitting information data from a transmitter to a receiver according to any one of claims 1 to 22 via the nested channel (NC),
The transmitter (TRD)
An encoder (ENC) according to an error correction code structure;
A binary modulator (MOD) whose output is an M length vector (Z) of modulation symbols (Z (i));
A linear combiner (LC) whose parameter is the number s called the bond size;
The output of the linear combiner (LC) is the M length output vector (X).
A system for transmitting information data from a transmitter to a receiver via a nested channel (NC). The nested channel (NC) is the following upper triangular matrix

Is a sliding triangular channel described by
here

Denotes a product for each term, a ₁ , a ₂ ,..., A _M are coefficients of fading random variables that define the diversity characteristics of the channel, and γ _{i, j} values are random variables or fixed parameters. 24. The system of claim 23, wherein:   25. The system of claim 24, wherein the sliding triangular channel is defined by a SAF protocol for cooperative transmission with several repeaters. The combined symbol (X (i)) of the M length output vector (X) is obtained by the product of the M length vector (Z) of the modulation symbol and the M × M complex linear combination matrix (S),
The M × M complex linear combination matrix (S) is given by the matrix multiplication P ₁ × S ′ × P ₂ , where P ₁ is a row permutation matrix, P ₂ is a column permutation matrix, and S ′ Is a block diagonal matrix

Defined as
Here, I is a unit matrix, and the submatrix S ″ is an s number of symbols transmitted by an optimal hard output detector linearly combined by the submatrix S ″ and transmitted on a diagonally independent fading channel. When used to receive, the s × s complex such that the performance measured at the output of the detector DET exhibits a diversity order that is s times higher than the performance without the linear combination matrix S ″. 26. A system according to any one of claims 23 to 25, characterized in that it is a coupling matrix. The submatrix S '' is
The first symbol (X (1)) and the last s−1 symbols (X (M + 2−s),..., X (M)) of the M length output vector (X) are the M length of the modulation symbol. Be a linear combination of the first s modulation symbols (Z (1),..., Z (s)) of the vector (Z), and
Each of the other symbols (X (i)) of the M-length output vector (X) is one of the last Ms modulation symbols (Z (i)) of the M-length vector (Z) of the modulation symbol. 27. The system of claim 26, wherein the system is defined to be identical. The encoding rate (Rc) of the encoder (ENC), the number of time slots M of the nested channel (NC), the combined size s, and the achievable target diversity order δ have the following relationship:

28. The system according to any one of claims 23 to 27, characterized in that they are selected depending on each other. The nested channel (NC) is a sliding triangular channel;
The combination size s of the linear combiner (LC) depends on the target diversity order δ, the rate Rc of the encoder (ENC), and the number of time slots M of the nested channel (NC) as follows:

29. The system according to claim 28, wherein the system is selected using:

Images