KR20020095227A - Multicarrier transmission system with reduced complexity leakage matrix multiplication - Google Patents

Abstract

Described is a transmission system for transmitting a multicarrier signal from a transmitter (10) to a receiver (20). The multicarrier signal comprises a plurality of subcarriers. The receiver (20) comprises a channel estimator (28) for estimating amplitudes of the subcarriers and for estimating time derivatives of the amplitudes. The receiver (20) further comprises an equalizer (24) for canceling intercarrier interference included in the received multicarrier signal in dependence on the estimated amplitudes and derivatives (29). The receiver (20) comprises a multiplication by a leakage matrix, and the multiplication is implemented in the following sequence N-point IFFT (82), N pointwise multiplications (84) and an N-point FFT (86).

Description

Multicarrier transmission system with reduced complexity leakage matrix multiplication

Multicarrier modulation methods such as OFDN and MC-CDMA have been used for some time now. OFDM or Orthogonal Frequency Division Multiplexing is a modulation method designed in the 1970s in which multiple user symbols are transmitted in parallel using different subcarriers. These subcarriers have overlapping (sinc-shaped) spectra, although the signal waveforms are orthogonal. Compared to modulation methods such as BPSK, QPSK, or MSK, OFDM, they have a relatively long duration for transmitting symbols, but have a narrow bandwidth. Usually, OFDM systems are designed such that each subcarrier is small enough in bandwidth to experience frequency-flat fading. This also ensures that the subcarriers can select the frequency (as appropriate) but remain orthogonal when received over a time-invariant channel. When an OFDM signal is received over this channel, each subcarrier experiences different attenuation but is not distributed.

The characteristics of OFDM mentioned above avoid the need for a tapped delay line equalizer and avoid the use of Digital Audio Broadcasting (DAB), Digital Terrestrial Television Broadcast (DTTB), which is part of the Digital Video Broadcasting standard (DVB), and more recently wireless local. In some standards, such as HIPERLAN / 2, the area network standard, it is a major motivation for using OFDM modulation methods. Especially in DAB and DTTB applications, mobile reception under adverse channel conditions is predicted with both frequency and time variance. Mobile reception of television is no longer regarded as a major market. Nevertheless, DVB systems promise to be a fast delivery mechanism for mobile multimedia and Internet services. At the IFA '99 Consumer Electronics trade show, a consortium of Nokia, Deutsche Telecom and ZDF demonstrated mobile web browsing, email access and television viewing over an OFDM DVB link with a GSM reply channel. 8k OFDM subcarriers with wireless DVB reception functioned properly at 50 mph for vehicle speed. Mobile reception, ie, reception and corresponding time distribution over channels of Doppler spreads, generally remains one of the problems associated with multicarrier transmission systems, in particular OFDM systems. While its strength to frequency selectivity appears to be an advantage of OFDM, the time varying nature of the channel is known to suggest limitations on system performance. Time variations are known to impair orthogonality of OFDM subcarrier waveforms. In this case, ICI (Intercarrier Interference, also referred to as FFT loss) occurs because signal components from one subcarrier cause to interface with another, mainly neighboring subcarriers.

In April 1998, a multicarrier transmission system is described by Guillaume Geslin in the document "Equalization of FFT-leakage in mobile DVB-T" by stockholm, Master Tesis in Radiocommunication from the Royal Institute of Technology. This known transmission system ICI is canceled at the receiver by the equalizer (ie detected and removed from the received multicarrier signal). This equalizer infers a vector of symbols estimated from the vector of received symbols. The operation of the equalizer is based on a channel model in which the amplitudes of the subcarriers and their time derivative represent ICI. The receiver includes a channel estimator that generates estimates of these amplitudes and derivatives and feeds these estimates to the equalizer. Receivers in known transmission systems are relatively complex. That is, a relatively large number of calculations are required to perform the channel estimator and equalizer.

The present invention relates to a transmission system for transmitting a multicarrier signal from a transmitter to a receiver.

The invention also relates to a receiver for receiving a multicarrier signal from a transmitter.

1 is a block diagram of a transmission system according to the present invention.

2 and 3 and 5 and 6 are block diagrams of embodiments of decision feedback receivers in accordance with the present invention.

4 is some graphs illustrating the performance of a decision feedback receiver as shown in FIG. 3.

7 is a block diagram of an MC-CDMA transmitter.

8 is a block diagram of an embodiment of an MC-CDMA decision feedback receiver in accordance with the present invention.

It is an object of the invention to provide a transmission system according to the above in which the computational burden is largely reduced. This object can be achieved in a transmission system according to the invention, wherein the transmission system is arranged to transmit a multicarrier signal from a transmitter to a receiver, the multicarrier signal comprising a plurality of subcarriers, the receiver having an amplitude of the subcarriers And a channel estimator for estimating the temporal derivatives of the amplitudes, the receiver also includes an equalizer for canceling intercarrier interference included in the received multicarrier signal according to the estimated amplitudes and derivatives, and Multiplication of the leak matrix Ξ of N x N, and the multiplication is performed as a sequence of N-point IFFT, multiplications such as N points, and N-point FFT. The invention is based on recognizing the complexity of the multiplication of the leak matrix Ξ, which is a very commonly used operation with respect to ICI cancellation in multicarrier transmission systems, where the leak matrix Ξ is based on Fourier-based, Ξ = FΔF ^H. It will be reduced by the fact that it is diagonalized. Where F is an N-point FFT matrix with normalized columns and Δ is a positive diagonal matrix. Thus, multiplication by Ξ, which is an N x N matrix, can be performed as a sequence of N-point IFFT 82, multiplications 84 such as N points, and N-point FFT 86.

In an embodiment of the transmission system according to the invention, the receiver is a decision feedback receiver and the channel estimator comprises a smoothing filter for smoothing the estimated amplitudes and / or derivatives. The use of such a smoothing filter has the advantage of using the correlation between derivatives. That is, because of noise or other effects, in particular the estimation of the derivative on the subcarrier is inaccurate and it is useful to consider the values of the derivative in neighboring subcarriers as well. In practice this typically means smoothing the values of the subcarriers of the various subcarriers.

In another embodiment of the transmission system according to the invention the FFT is also arranged to demodulate the received multicarrier signal. In this way the FFT will be used for demodulation and will be used as part of the multiplication of the lossy matrix.

The above objects and features of the present invention will be explained in more detail from the following description of the preferred embodiment with reference to the drawings.

In the figures, like reference numerals are provided to like parts.

The present invention is based on the development of a simple and reliable channel representation. For that purpose, a multicarrier transmission system, for example We will consider an OFDM or MC-CDMA transmission system with N subcarriers arranged by. Each subcarrier It has a finite length rectangular envelope that includes periodic expansions in excess of. If is a vector of N transmitted symbols, then the transmitted continuous time baseband signal is expressed as

In the case of a time-varying additive white Gaussian noise (AWGN) channel with selectable frequency, the received continuous time signal is expressed as follows.

Where the coefficient H _k (t) represents the time-varying frequency response in the k-th subcarrier at 1 ≦ k ≦ N, and n (t) is AGWN in the signal bandwidth. Since the channel changes slowly, assume that only the first change will be considered within the signal data block period. That is, all H _k (t)

Accurately approximated by and, where H _'k (t) is a first order approximation of _r and t H _k (t) is a time reference in the received data block. Note that the time varying channel H _k (t) may also consider the remaining frequency offset after coarse frequency synchronization.

The received baseband signal is sampled at sampling offset t ₀ and rate Nf _s , and the next samples in its Nth block [y (t ₀ ), y (t ₀ + T), ..., y (t ₀ + (N-1) T)] (where ) Is dependent on a fast fourier transform (FFT) of size N. y = [y ₁ , ..., y _N ] If ^T is a vector of N FFT samples

to be. If you use (4) instead of (2) and the approximation of (3),

Get Where n _k are samples of AWGN with some variance σ ² , for 1 ≦ _k ≦ N. It is convenient to rewrite the result (5) in a closed matrix form. To this end, the diagonal matrices A = diag {a ₁ , ..., a _N }, D = diag {d ₁ , ..., d _N } and N x N matrix

Define. By this notation, expression (5) is as follows.

Where n = [n ₁ , ..., n _N ] ^T is the N x 1 vector of AWGN. In the channel model 9, the effect of the channel is represented by two sets of N parameters a = [a ₁ , ..., a _N ] ^T and d = [d ₁ , ..., d _N ] ^T do. , Then coefficients a _k are equal to the complex amplitudes of the channel frequency response rotated by the sampling phase exp (i 2 pi f _s lt ₀ ) for 1 ≦ _k ≦ N. Similarly, the coefficients d _k are the time domain derivatives of the complex amplitudes of the channel frequency response scaled by the sampling period T and rotated by the same sampling phase exp (i 2 pi f _s lt ₀ ) for 1 ≦ k ≦ N. same.

Note that inter-carrier interference occurs when the channel responds to variables in time (ie, d ≠ 0). This interference is defined by the vector d as well as the fixed N × N matrix Ξ. This is according to (8) that the latter matrix is the Toeplitz Hermitian matrix

It is easy to see that.

Later in this specification, a will be referred to as the amplitudes of (vector), d is the derivatives (vector) and Ξ is the loss matrix. In the above representation, the values on the diagonal of Ξ depend on the (optional) selection of the reference time instant t ₀ and may vary depending on the embodiment of the receiver. Typical choices of t ₀ are time start and end or middle of the frame window. For t ₀ selected near the middle of the frame, the diagonal items tend to approximate approximately zero.

For the implementation of a receiver based on the principles discussed herein, multiplication of Ξ will not be particularly complicated for large N (many subcarriers). This, of course, uses only the Taplitz property of the term by using only terms near the diagonal and performing it as a delay-line filter. But there are more efficient implementations. Recognize that the first order ICI terms result from linearly increasing amplitudes. In other words, you can execute Ξ in the following order:

1. I-FFT operation (return to frequency domain from frequency domain representation of subcarriers)

2. Multiplication resulting from a time-domain signal by the weighting of each element by a diagonal matrix, ie, a scalar. In its basic form the weighting is y, a linearly increasing function, but in practice this will be converted from this, for example avoiding noise enhancements that optimize the reduction of ICI together.

3. FFT operation to return to the frequency domain representation in which subcarrier signals are typically processed.

This allows the performance of Ξ in complexity NlogN, instead of N ² multiplications. The inverse of Ξ can be done with this structure. For the inverse, the terms on the diagonal of the matrix of step (2) will be approximated in form or n / (c + n ² ), where n is the index in the output number of the I-FFT. We will use this in some receiver embodiments, which will be given later. In practical practice, this is useful for using the same hardware circuitry for the FFT and IFFT used for Ð as the hardware used for the main FFT operation to retrieve subcarrier signals.

In order to process the received signal, the set of channel parameters a and d will be estimated. The estimation accuracy of these 2N scalar parameters will be enhanced if the stochastic characteristics of the channel are used. first, Assume that these channel variations are sufficiently slow that they do not change continuously within the duration of the symbol. In this case, (6) and (7) can be rewritten as follows.

The quantities a, d and the physical parameters of the spreading channel, ie its K th spread delays {τ ₀ , ..., τ _k }, the corresponding Doppler shifts {f ₀ , ..., f _k }, and Analyze the relationship between the set of complex amplitudes {h ₀ , ..., h _k }. The static characteristics of the channel frequency respond depending on the relative delays and Doppler shifts and the group delay and / or Doppler shift results in the order of h _k for 1 ≦ _k ≦ _K , where the sequences are time and carrier synchronization / Note that it is handled by carrier synchronization / tracking. Thus, we assume no loss of generality τ ₀ = 0 and f ₀ = 0. Now, the channel frequency response H _l and its derivative H ' _l can be written as

The relationships between (10) and (11) can be easily used to estimate the static characteristics of amplitudes a and derivatives d. Whenever the number of diffusion paths is large enough (ideally K »N), the coefficients The sets of can be considered together as a Gaussian distributed. Also this Sets don't matter and It can be seen that are not correlated with each other and that the Doppler spectrum has a symmetrical shape. In this case, vectors a and d mean zero and covariance matrices

It can be assumed to be a statically independent multivariate Gaussian with, where E {.} Is present for the mathematical expectation operator and C _a , C _d are N × N Hermit non-speech defined matrices.

As described in the book Microwave Mobile Communications by C.Jakes, published in 1974 by John Wiley & Sons, Inc., there is a very important case of C _a , C _d corresponding to the standard model for mobile channels. This model (known as the Jakes model) assumes independent contributions of different diffusion pathways, exponential delay profiles and uniquely distributed angles of occurrence for different pathways. In this case,

Can be seen, where f _Δ is the magnitude of the Doppler spread and T _Δ is the root mean square spread delay spread. The last two parameters depend on the speed of travel and the supply environment, respectively.

Although the outlined channel models are characterized by 2N parameters, the number of independent degrees of freedom is actually substantially smaller. This property comes from the fact that the spread of propagation delay is sometimes much smaller than the word retention period. This property also means that the entries of a are strongly related and extend to the fact that the expected matrix C _a can be approximated by a matrix of exactly lower rank.

Similarly, the entries of d are strongly related and the expected matrix C _d can also be approximated by a matrix of exactly lower rank. Considering the Jakes model is (13). Defining the eigendecomposition of C:

Where U is the NxN unitary matrix of eigenvectors of C and Λ is the NxN positive diagonal matrix of its eigenvalues {Λ ₁ , ..., Λ _N }. Assume that eigen values are ordered and thus the sequence of {Λ ₁ , ..., Λ _N } does not increase. In the Jakes model, the elements of this sequence have an exponentially decaying profile:

Thus, a sequence of eigenvalues can be accurately approximated with a relatively small number of r of nonzero values:

The aforementioned characteristics of channel parameters (ie amplitudes and derivatives) can be widely used to extract reduced complexity sequences for channel equalization with ICI removal. Obviously, in the situation where the static channel is switched from the ideal theoretical situation, these models still keep the design of the actual receiver. Mismatches between real channels and idealized channel models can be led to (small) performance penalties. However, this does not mean that receiver principles appear in the present invention, but does not mean that it can not be used successfully.

1 shows a block diagram of a transmission system according to the invention. The transmission system includes a transmitter 10 and a receiver 20. The transmission system may also include transmitters 10 and receivers 20. The transmitter 10 transmits a multicarrier signal to the receiver 20 via a wireless channel. The multicarrier signal may be an OFDM signal or an MC-CDMA signal. Receiver 20 includes a demodulator 22 for demodulating the received multicarrier signal, the received multicarrier signal comprising vectors of received symbols. Demodulator 22 may be performed by an FFT. The demodulated multicarrier signal is supplied by the demodulator 22 to the equalizer 24. Equalizer 24 eliminates intercarrier interference that may be included in the received multicarrier signal. Equalizer 24 outputs vectors of estimated symbols 25 (derived from vectors of received symbols) to (smoothing) slicer 26. Slicer 26 may be used for smoothing matrices (smoothing decisions) and / or signal coded portions of a receiver (not shown) that is, for example, an FEC decoder (hard decisions). Generate a binary that estimates. The output signal of slicer 26 may also be considered to include estimated symbols 27. Receiver 20 also includes a channel estimator 28 for estimating the amplitudes 29 of the subcarriers and estimating the time derivatives 29 of the amplitudes. Equalizer 24 cancels intercarrier interference included in the received multicarrier signal depending on the estimated amplitudes and derivatives 29 provided from channel estimator 28 to equalizer 24. The channel estimator 28 includes a reduced complexity filter for the amplitudes estimated from the vectors of the received symbols 23 and the vectors of the estimated symbols 27 and the vectors of the derivatives 29. can do.

Now we continue with the embodiment of the receiver based on the advanced channel model. OFDM receiver amplitudes Derivatives (including complex values, e.g., phase information) as well as (as in conventional receivers) If even (uncommon for typical OFDM receivers) is extended to the preferred estimation, the user data can be changed as follows.

Symbolize the estimates of amplitudes and derivatives, respectively And Having a matrix Create Note that the receiver receives y = Qs + n (according to (9)).

-after Estimate s.

Such a receiver is called a linear receiver. The receiver 20 shown in FIG. 1 may be considered such a linear receiver when the equalizer 24 performs matrix multiplication Q'y. Where Q 'plays the role of Q. There are at least two approaches. In the zero-proximity approach, Q 'is exactly the arithmetic inverse of Q. Q 'in the MMSE setting That is, chosen to corroborate the conditional expectation of s for a given y. This is typically the mean squared error Minimize. In a zero-proximate receiver, ICI is effectively canceled but noise is augmented. This can lead to undesirable results. The MMSE receiver optimizes the combined strength of noise and the rest of the ICI. This receiver provides instantaneous channel characteristics And It requires an adaptive (typically real-time) inverse of the matrix that depends on.

It may also use what is called a decision feedback receiver. The channel model previously mentioned in this document is expressed in some aspects so as to refine and perform this decision feedback receiver, among which:

The feedback loop in the estimates of the derivatives is more accurate using more knowledge, especially correlation, about the static behavior of these derivatives.

The feedback loop in the error correlation code is used within the loop. Note that reference is made to multicarrier CDMA as a facet, and the spreading code serves as an identification correlation error code. That is, it plays the role of decoding in the loop (does it occur as CDMA despreading, error correlation decoding, or any other form).

2 shows a block diagram of an embodiment of a decision feedback receiver. The decision feedback receiver 20 includes a demodulator 22 for demodulating the received multicarrier signal, the received multicarrier signal comprising vectors of received symbols. Demodulator 22 may be performed by FFT. The demodulated multicarrier signal is supplied by demodulator 22 to eliminator 32. Eliminator 32 removes the estimate of ICI included in the multicarrier signal received from the demodulated multicarrier signal. The resulting "ICI-free" signal is fed to equalizer 24 and to channel estimator 28 for normalization of the signal. Equalizer 24 may also include a slicer. Equalizer 24 operates depending on the estimated amplitudes supplied by channel estimator 28 to equalizer 24. The output signal of equalizer 24, including the vectors of estimated symbols, is provided to multiplier 31. In addition, the output signal of the equalizer 24 is supplied to other signal processing portions (not shown) of the receiver. Channel estimator 28 estimates the amplitudes and time-derivers of the subcarriers. Estimated amplitudes 29 are provided to equalizer 24 and estimated derivatives 29 are provided to multiplier 31. The multiplier 31 multiplies the estimated derivatives and the estimated data symbols and feeds the resulting signal to a filter 30 which performs a loss matrix Ξ. The filtered signal, which is an estimate of ICI, is then fed to eliminator 32.

Based on the general configuration, another decision feedback receiver can be studied as shown in FIG. Here, the FFT demodulator is not shown (but thought to exist). The structure of a conventional OFDM receiver (based on blind or pilot) and a signal path including Y ₀ , Y ₁ , Y ₂ , slicer 26, omnidirectional error control decoding 42 and a channel estimator; similar. In the receiver described here, the removal of the estimated ICI (ΞZ ₅ ) was started using Y ₁ = Y ₀ -ΞZ ₅ . Where Z ₅ is modulated derivatives Is estimated. Signal paths Z ₁ , Z ₂ , Z ₃ , Z ₄ are Is used to estimate the derivatives of the amplitudes. The logic behind the circuit is that Z ₁ regains ICI because the estimate of modulated subcarriers is removed from Y ₀ . Only noise leaves ICI and estimation errors. Filter 50 is used to estimate the modulated derivatives from the ICI. It is not necessary to reverse in exact mathematical recognition, but this inverts the loss matrix Ξ. Preferably this is done while avoiding excessive noise enhancement or enhancement of the estimation error. The modulation of the derivatives is removed in step Z _2- > Z ₃ . Filter 54 is used to produce a better estimate Z ₄ in the correlation between the subcarrier derivatives. Useful implementation of Ξ leads to (inverse) FFT, multiplication (diagonal matrix) and stepwise use of FFT.

Although the circuit is depicted in hardware assembly blocks, a typical implementation involves iterative software processing. We experimented with the included iteration method of the following steps for iterative round i:

Input: amplitudes Derivatives, as well as the (i-1) th estimate of And data Observation of Y ₀ . Where the values between the parentheses represent the number of repeat rounds.

Previous Derivatives And data Using the estimates of Calculate Y ₁ (i) using.

New estimated amplitudes from Y ₁ (i) , And (not shown) explain the knowledge of the amplitudes and derivatives in the previous frame as possible.

ㆍ Data New estimates.

Calculation of Z ₁ (i), Z ₂ (i), Z ₃ (i), Z ₄ (i), Z ₅ (i).

It is possible to use knowledge of amplitudes and derivatives in the previous frame (optional step, not shown in the figure). This step includes operations that use correlation among the subcarrier derivatives.

Output: amplitudes , Derivatives , And data New estimates.

Start condition is , , And For all zeros.

Filter 50 attempts to recover the estimate of ds from Z ₁ by filtering Z ₂ = M ₁ Z ₁ . One mathematical approach is to use the orthogonality principle for MMSE estimation. In this case, the appropriate choice for M _{1 follows} from the request E [(Z ₂ -ds) Z ₁ ^H ] = 0 _N. e is Is defined as a vector of decision errors. this is Where _n is the variable of noise. Any modeling and (pre) calculation of static expectations can be performed here, but this is not practical for receiver designers. Thus, in the following, to simplify the approximations will be searched.

this is Can be simplified as a result of M ₁ , where G is G = c ₁ I _N with a constant c ₁ which can be adapted to the particular spreading environments, eg average BER, average SNR or speed of a mobile receiver. It is determined experimentally.

Z ₃ is Close to, but this is due to error distributions due to AWGN and Wow It includes the estimation error in. Here, channel behavior, for example, can explain the advanced static knowledge of the interrelationships of the derivatives. The multiplication to the multiplier Z ₅ of the circuit from Z ₂ , 1 / x, Z ₃ , M ₂ , Z ₄ is intended to perform this task. Multiplications remove or rewrite the modulated data on the signal. A smoothing operation M ₂ takes place between. MMSE filter is closely approximated Orthogonality Principle Estimate Z ₄ to follow from, and thus, to be. In fact this is approximated as it is To be receivable. Experiments show that R ₃ = c ₂ I _N with the constant c ₂ when c ₂ is an operable solution.

4 shows some graphs illustrating the performance of a decision feedback receiver as shown in FIG. 3. The magnitudes of the amplitudes and derivatives are plotted against the number of subcarriers (in dB). Graph 60 shows the intensity of the actual amplitudes, and graph 62 shows the intensity of the estimated amplitudes. Graph 64 shows the intensity of the actual derivatives, and graph 66 shows the intensity of the estimated derivatives. It can be seen that if the estimated derivatives deviate some from the actual derivatives, the amplitudes are very well estimated by the decision feedback receiver of FIG. 3.

5 shows another decision feedback receiver based on the MMSE structure. This allows iterative calculations to minimize the change in the error between the input and output of the slicer. The slicer box, as used in FIG. 5, may or may not include error correlation decoding. Slicer Wow As well as data It also outputs an estimate of. For QAM modulation OFDM is typically derived from a lookup table Is found. Iterative software processing may also be used in the receiver of FIG. The repetition method includes the following steps for repetition round i:

Input: amplitudes Derivatives, as well as the (i-1) th estimate of And data Observation of Y ₀

Previous Derivatives And data Using, Calculate Y ₂ (i) using.

New estimated amplitudes from Y ₂ (i) , And (not shown) explain the knowledge of the amplitudes and derivatives in the previous frame as possible.

New data , And Corresponding values of (e.g., table lookup) Estimated.

Calculation of Z ₆ (i), Z ₇ (i), Z ₈ (i), Z ₉ (i). Where Z ₉ (i) is estimated Act as (correction of).

Integration of Z ₉ (i) over various rounds, eg

Use knowledge of amplitudes and derivatives in previous frames (not shown in the figure).

Output: amplitudes , Derivatives , And data New estimates.

Start condition is , , And For all zeros.

Filters 72 and 76 can be taken unadapted, which is the same as M ₆ = M ₁ and M ₇ = M ₂ in FIG. 3, respectively. The actual value for the integral constant can be α = 0.9, so to be.

It is shown that smelting some implementations is possible as shown in FIG. 6. Here, the inverse of Ξ is performed as a Finite Impulse Response (FIR) filter. The second filter M ₇ is performed as an IIR smoothing filter. Finally, an FFT-weighting-FFT filter is applied to generate an estimate of ICI.

Many other implementations are envisaged: the use of amplitude and derivatives of preceding frames to better estimate amplitude and derivative. This may be indicated as an 'optional step' in algorithms or for a new frame. end plus We can take an initial condition that repeats the inference of the results of the correlations from the previous OFDM frame, later from during any periodic addition or defense interval, which is equal to.

The filters of the receiver, in particular M ₁ , M ₂ , M ₆ , M ₇ , may be in an actual receiver selected from a library of fixed or precomputed values. For example, a receiver control system may be used to provide a dissemination environment between optimized settings for static reception (largest ICI switching), slow movement reception (several ICI deletion), or rapid movement reception (active ICI deletion). May be knowledge to select.

In addition, adaptive filters may be used. These can be used as confidence information for the estimates. This may be accomplished by the recognition of deletions in the matrices or estimates of adaptability.

MC-CDMA is an extension of the basic OFDM principle. In 1993, this form of Orthogonal Multi-Carrier CDMA was proposed. Basically, an OFDM-type of transmission in a multi-user simultaneous DS-CDMA signal is proposed. Thus, it is exposed to Doppler. As shown in Figure 7, the following vector notation is used. For OFDM, a vector s of length N carries a 'frame' of user data, with s = [s ₀ , s ₁ , ..., s _N-1 ] ^T , where elements s _n are user symbols admit. In MC-CDMA, s = Cx, where C is an NxN code matrix and x = [x ₀ , x ₁ , ..., x _N-1 ] ^T represents a frame of user data. Explicitly we will call N user signals x to identify that all symbols come from the same last user or not. The kth column of C represents the 'spreading code' of the user data stream k, Will be represented. In the special case commonly used, Will be considered, where WH _N is a Walsh-Hadamard matrix of size _N × _N. In this case, C = C ⁻¹ = C ^H , thus CC = I _N , where I _N is the _N × _N unit matrix. In other special cases, ie, C = I _N , the MC-CDMA system is reduced to OFDM. For ease of analysis, , or Similar to generalizing modulation. then to be.

7 shows such an MC-CDMA transmitter. The frames are in parallel with serial-parallel (S / P) conversion of the input stream of data by serial-to-parallel converter 90, code spreading application by spreader 92, I-FFT 94 and fixed insertion Generated by the serial transformation 96. It will be assumed that addressing a transmission of a single frame, interframe interference is avoided by selecting the appropriate defensive intervals. Thus, the components of the vectors s and x are constant over time. Frame duration includes any defense interval T _s , where w _s t _s = 2π.

If the receiver architectures proposed in the previous parts are used for MC-CDMA, then basically the FEC is changed by the (inverse) code matrix. The receiver represented in FIG. 8 is an extension of the receiver of FIG. 5, adjusting MC-CDMA instead of OFDM. The difference from the OFDM receiver is in the dotted box. For OFDM, this basically includes gain control and slicers. For MC-CDMA, it is applied to the code spreading matrix C (used in the despreader 102 of the code spreaders 106 and 110). The receiver is designed so that all matrices are fixed (or taken from a library of precomputed values) and no partitions are needed. The exception is the weight matrix W (used for filter 100), which has channel adaptation settings.

In the MMSE setting, W has only nonzero components in the diagonal,

It can be seen that the constant here depends on the noise floor. Detailed descriptions of the circuit for estimating amplitudes are not shown in FIG. 8, but may be based on known principles.

For MC-CDMA, the slicer is based on its symbol decisions on energy received from all subcarriers, and thus the reliability of the estimate Is much more accurate than in subcarriers at fade.

The principles of the receivers described above can also be combined with an FFT that adjusts more samples than a regular sized FFT. One example is the use of fractionally located FFTs, and the other is a dual sized FFT. It may also design a system to separate the received components via amplitudes from these received derivatives.

Although primarily OFDM transmission systems have been described above, the present invention can equally be utilized in other multicarrier transmission systems, such as MC-CDMA transmission systems. A large portion of the receivers will be performed by digital hardware or software performed by a digital signal processor or general purpose microprocessor.

The scope of the invention is not limited to the explicitly described embodiments. The invention is illustrated in each new feature and each combination of features. Any reference signs do not limit the scope of the claims. The term "comprise" does not exclude the presence of these elements and other elements and steps in the claims. The use of a singular representation of a component does not exclude the presence of a plurality of such components.

Claims (10)

In a transmission system for transmitting a multicarrier signal from a transmitter 10 to a receiver 20, the multicarrier signal includes a plurality of subcarriers, the receiver 20 estimates amplitudes of the subcarriers and the amplitude And a channel estimator 28 for estimating the temporal derivatives of the receiver, wherein the receiver 20 also depends on the estimated amplitudes and derivatives 29 to determine the intercarrier interference included in the received multicarrier signal. An equalizer 24 for erasing, the receiver 20 comprising multiplication by a loss matrix Ξ of N x N, the multiplication being an N-point IFFT 82, multiplications 84 such as N points. And a sequence of N-point FFTs 86. The transmission of claim 1, wherein the receiver 20 is a decision feedback receiver and the channel estimator 28 comprises a smoothing filter 76 that smoothes the estimated amplitudes and / or derivatives. system. The transmission system of claim 1 or 2, wherein the FFT (86) is further arranged to demodulate the received multicarrier signal. The transmission system according to claim 1, 2 or 3, wherein the multicarrier signal is an OFDM signal. The transmission system according to claim 1 or 2 or 3, wherein the multicarrier signal is an MC-CDMA signal. In the receiver 20 for receiving a multicarrier signal from the transmitter 10, the multicarrier signal comprises a plurality of subcarriers, the receiver 20 estimates the amplitudes of the subcarriers and the time derivative of the amplitudes. Channel estimator 28 for estimating the interpolation, and the receiver 20 is adapted to cancel intercarrier interferences included in the received multicarrier signals depending on the estimated amplitudes and derivatives 29. Also included is an equalizer 24, the receiver 20 comprising multiplication by a lossy matrix of N × N, the multiplication being N-point IFFT 82, multiplications 84 such as N points and N Receiver 20, which is performed as a sequence of point FFTs 86. 7. The receiver of claim 6, wherein the receiver 20 is a decision feedback receiver and the channel estimator 28 comprises a smoothing filter 76 which smoothes the estimated amplitudes and / or derivatives. . 8. Receiver (20) according to claim 6 or 7, wherein the FFT (86) is also arranged to demodulate the received multicarrier signal. 9. A receiver (20) as claimed in claim 6 or 7 or 8, wherein the multicarrier signal is an OFDM signal. 9. A receiver (20) as claimed in claim 6 or 7 or 8, wherein the multicarrier signal is an MC-CDMA signal.