JP5287846B2 - High performance transmission system, transmission method, receiver, and transmitter - Google Patents

Description

  The present invention relates to a high-performance transmission system using a multipath transmission line as a transmission medium, a transmission method, a receiver, and a transmitter.

  In a radio wave propagation path in mobile communication, radio waves arriving from a transmitting antenna are reflected according to the surrounding terrain and the like, and arrive at a receiver as a group of waves. Therefore, what has always been an obstacle to realizing high-quality mobile communications is a phenomenon called multipath fading that results from these results. One of the wireless communication methods for solving this problem is frequency domain equalization. Since this is excellent in frequency utilization efficiency and exhibits excellent characteristics even in a multipath environment, its application to next-generation mobile communication systems has been actively studied.

On the other hand, in next-generation mobile communication, higher-speed information transmission is required. As a related technology, a high-speed turbo decoder for providing error correction means with a small time lag without causing an increase in circuit scale such as an increase in memory capacity in a mobile communication requiring high speed and interactivity is provided. It has been proposed by the inventor (see Patent Document 1).
JP 2001-320282 A

  In order to realize high-speed information transmission, the modulation method used in mobile communication continues to evolve from the conventional QPSK (Quadrature Phase Shift Keying) to 16 QAM (Quadrature Amplitude Modulation) and further to 64 QAM. However, the influence of multipath interference peculiar to mobile communication becomes larger as the density becomes higher, and the characteristics are extremely deteriorated.

  In this regard, the related technology described above provides a powerful error correction means in mobile communication that requires high speed and the like, but since it has not been made by specifically recognizing the above problems, frequency utilization efficiency It is difficult to overcome the characteristic degradation in a multipath environment caused by using a good high-density modulation.

  The present invention has been made in view of the above problems, and provides a transmission method that overcomes characteristic degradation in a multipath environment caused by using high-density modulation with high frequency utilization efficiency. Furthermore, the present invention provides a method in which the processing scale is reduced with respect to the reception processing adopting such a high-performance transmission method.

  In order to achieve the above object, a high-performance transmission system according to the present invention is a transmission system for transmitting information between a transmission side and a reception side via a transmission line, and is equivalent to a high-density modulated wave transmitted by a Nyquist rate. In order to realize the transmission speed, it has means for transmitting a plurality of low density modulated waves at a speed exceeding the Nyquist rate, and the receiving side receives a semi-ring when receiving the plurality of low density modulated waves. The transmission information is estimated by circulating between the detector and the decoder by using an external likelihood value.

  The high-performance transmission method according to the present invention is a transmission method for transmitting information between a transmission side and a reception side via a transmission line, and realizes a transmission rate equivalent to a high-density modulated wave transmitted by a Nyquist rate. Therefore, when transmitting a plurality of low density modulated waves at a speed exceeding the Nyquist rate and receiving the plurality of low density modulated waves on the receiving side, the detector and the decoder used a semi-ring. The transmission information is estimated by performing a metric calculation and circulating between the detector and the decoder by using an external likelihood value.

  The receiver according to the present invention is a receiver of a transmission system that transmits information between a transmission side and a reception side via a transmission line, and realizes a transmission rate equivalent to a high-density modulated wave transmitted by a Nyquist rate. Therefore, when receiving a plurality of low density modulated waves transmitted at a speed exceeding the Nyquist rate, a detector and a decoder for performing a metric calculation using a semi-ring are provided. Transmission information is estimated by cycling between decoders with external likelihood values.

  Further, the transmitter according to the present invention realizes a plurality of low density modulated waves at a speed exceeding the Nyquist rate in order to realize a transmission rate equivalent to the high density modulated wave transmitted by the Nyquist rate to the above receiver. It is characterized by transmitting.

  According to the present invention, it is possible to provide a transmission method that overcomes characteristic degradation in a multipath environment caused by using high-density modulation with high frequency utilization efficiency. Furthermore, it is possible to provide a method in which the processing scale is reduced with respect to the reception processing adopting such a high-performance transmission method.

It is a schematic block diagram which shows the whole structure of the high performance transmission system which concerns on the Example of this invention. FIG. 2 is a schematic block diagram illustrating a configuration of a state metric update circuit used in a detector having a semi-ring configuration on the reception side illustrated in FIG. 1. FIG. 3 is a schematic block diagram showing an overall configuration of a gamma metric supply circuit shown in FIG. 2. FIG. 4 is a schematic block diagram illustrating a configuration of a first stage arithmetic processing unit illustrated in FIG. 3. It is a schematic block diagram which shows the structure of the ACS arithmetic process part which consists of the cascade connection of the 2nd stage or more stage shown in FIG. In the state metric update circuit used in the receiving-side semi-ring detector (semi-ring detector) shown in FIG. 1, an unrolling method for outputting an LLR for calculating an external likelihood value at a processing speed of K times is used. It is a schematic block diagram which shows the structure at the time of using. It is a schematic block diagram which shows the structure of the processing means based on the MMSE norm which degenerates the number of states of the receiving side shown in FIG. FIG. 6 is a schematic block diagram showing a configuration of a receiver in a transmission system according to a second embodiment of the present invention when a residual interference component obtained from a turbo loop is reproduced and subtracted from a received signal, and then degenerate processing based on the MMSE standard is performed. FIG. It is a schematic block diagram which shows the structure of the transmission system used with the simulation model of 64QAM of the comparative example with respect to the Example of this invention. In the Example of this invention, it is a graph which shows the simulation result in case a transmission line is an AWGN channel. In the Example of this invention, it is a graph which shows the simulation result in the case of a multipath transmission line. In the Example of this invention, it is a graph which shows the simulation result in the case where the state of a transmission line is unknown on the transmission side, and a known case.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Coding part 102 Interleaver 103 Modulator 104 which produces | generates several low density modulation waves Multiple low density modulation waves 105 of the speed | rate exceeding Nyquist rate Decoupling processing means 106 Processing means based on MMSE norm which performs degeneracy of the number of states 107 detector 108 interleaver 109 deinterleaver 110 decoder 111 receiver 201 alpha metric update circuit 202 alpha metric memory 203 address control up / down counter 204 gamma metric supply circuit 301 first stage arithmetic processing section 302 ACS arithmetic processing section 303 Gamma metric supply circuit 601 Beta metric update circuit 602 ACS operation processing unit 603 having a cascade connection structure of a plurality of stages Gamma metric supply circuit 604 Addition connection of beta metric and gamma metric Result 605 Gamma metric calculation result 701 Guard interval removal unit 702 Series-parallel conversion unit 703 FFT unit 704 Degeneration frequency domain MMSE unit 705 IFFT unit 706 Parallel serial conversion unit 805 Residual interference estimation unit 806 MMSE standard for reducing the number of states Processing unit 807 based detector 808 interleaver 809 deinterleaver 810 decoder 901 encoding unit 902 interleaver 903 64QAM modulator 904 Nyquist rate modulator output waveform 905 equalizer 906 demapper 907 deinterleaver 908 Decoder 909 Receiver

  Next, embodiments of the present invention will be described with reference to the drawings. The high-performance transmission method according to each embodiment of the present invention described below is applied to a transmission system, a transmission method, a receiver, and a transmitter that transmit information between a transmission side and a reception side via a transmission line. is there.

[First Embodiment]
The high-performance transmission system according to the first embodiment includes means for realizing a transmission rate equivalent to a high-density modulated wave transmitted by a Nyquist rate by transmitting a plurality of low-density modulated waves at a rate exceeding the Nyquist rate. . The receiving side has a detector and a decoder using a metric calculation method using a semi-ring when receiving a plurality of low density modulated waves, and an external likelihood is provided between the detector and the decoder. The transmission information is estimated by cycling according to the degree value. Accordingly, high-performance transmission can be realized at a transmission rate equivalent to that of a high-density modulated wave even in a multipath environment by using a low-density modulation anti-multipath interference characteristic and a transmission rate exceeding Nyquist rate.

[Second Embodiment]
In the high-performance transmission system according to the second embodiment, the metric calculation using the half ring used in the detector is a trellis line formed by the waveform shaping filter of the low-density modulated wave on the transmission side and the impulse response of the transmission line. For the calculation in the figure, a semi-ring in the Max-log domain, and as a definition of addition and multiplication on the semi-ring, the sum of elements a and b is set as a maximum value calculation (MAX), and elements a and b The product is normal addition,

の式に従い演算を行う。これにより、複数の低密度変調波の受信の処理規模を削減することができる。

The calculation is performed according to the following formula. Thereby, the processing scale of reception of a plurality of low density modulated waves can be reduced.

[Third Embodiment]
In the high-performance transmission system according to the third embodiment, the receiving side has processing means based on the MMSE (Minimum Mean Square Error) standard for reducing the number of states of the trellis diagram before the detector. Performs a metric operation using a trellis diagram based on the number of degenerated states. According to this, the processing scale can be reduced exponentially according to the amount of degeneration.

[Fourth Embodiment]
In the high-performance transmission system according to the fourth embodiment, the processing means based on the MMSE standard for reducing the number of states of the trellis diagram in the preceding stage of the detector includes a waveform shaping filter for a low-density modulated wave on the transmission side, and The trellis line formed by the impulse response of the transmission line is degenerated so as to become the trellis line formed by the waveform shaping filter of the low density modulated wave in transmission. The detector performs a metric operation using the trellis diagram formed by the waveform shaping filter of the low density modulated wave as the trellis diagram based on the number of degenerated states. As a result, a trellis diagram formed by a known low-density modulated wave waveform shaping filter on the transmission side can be used for the MMSE processing on the reception side, and the processing scale can be reduced.

[Fifth Embodiment]
In the high-performance transmission method according to the fifth embodiment, the waveform shaping filter of the low-density modulated wave on the transmission side has a linear phase characteristic and is composed of only amplitude components. Thereby, it can be set as a process only for a real axis, and the processing scale can be halved.

[Sixth Embodiment]
In the high-performance transmission method according to the sixth embodiment, the transmission side transmits a plurality of low density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high density modulated wave transmitted by the Nyquist rate. The waveform shaping filter of the low-density modulated wave on the transmission side is configured only with an amplitude component having a linear phase characteristic. The reception side has a detector and a decoder using a metric calculation method using a half ring when receiving a plurality of low density modulated waves, and an external likelihood value is provided between the detector and the decoder. The transmission information is estimated by circulating. With this configuration, processing means based on the MMSE norm that performs degeneration of the number of states in the trellis diagram is provided in the preceding stage of the detector. When performing metric calculation using a trellis diagram based on the number of states after degeneration, the detector uses a trellis line consisting of only a real or imaginary axis component formed by a waveform shaping filter composed of only amplitude components based on linear phase characteristics. Calculation is performed using the figure. As a result, the detector can process only the real axis as a trellis diagram by the number of degenerated states, and the processing scale can be halved.

[Seventh Embodiment]
In the high-performance transmission system according to the seventh embodiment, the metric calculation using the half ring in the detector is an alpha metric calculation in the BCJR (Bahl, Cocke, Jelinek, Raviv) algorithm or ACS (Add, Compare, Select calculation). And means for supplying a plurality of pipelined gamma metrics in at least one of the metric operations. In addition, it has an ACS computing means having a cascaded configuration of a plurality of stages with the pipelined gamma metric as an input. Furthermore, it has means for updating the state metric for each of a plurality of stages (k stages) with the calculation result as an input. As a result, the processing speed can be increased k times, and high-speed high-performance transmission can be realized.

[Eighth Embodiment]
In the high-performance transmission system according to the eighth embodiment, the processing means based on the MMSE standard for reducing the number of states of the trellis diagram in the previous stage of the detector includes a waveform shaping filter and a transmission of a low-density modulated wave on the transmission side This is done by a weighting process created using a covariance matrix for the result of the direct product of the impulse response of the path, and a cross-correlation vector of the waveform shaping filter of the low-density modulated wave on the transmission side and the impulse response of the transmission path. As a result, the degeneration process can be performed as a high-performance linear operation suitable for digital signal processing.

[Ninth Embodiment]
In the high-performance transmission scheme according to the ninth embodiment, when receiving a plurality of low density modulated waves, the receiving side estimates transmission information by circulating between a detector and a decoder using an external likelihood value. , Based on the transmission information obtained from the circulation process, insert a decoupling process to independently process the received signal with the real and imaginary axis components, and correlate the real and imaginary axis components. To reduce the number of states. Thereby, the complexity of the product of the real axis component and the imaginary axis component can be reduced to the sum complexity.

[Tenth embodiment]
In the high-performance transmission system according to the tenth embodiment, the processing means based on the MMSE standard for reducing the number of states of the trellis diagram in the previous stage of the detector is processing in the frequency domain, and after the conversion to the frequency domain After weighting the signal after the number of states is reduced as much as possible, conversion to the time domain is performed, and the signal after the reduction is used as an input of the detector. As a result, the degeneracy processing can be realized with low complexity even for a transmission line having a large delay spread.

[Eleventh embodiment]
In the high-performance transmission system according to the eleventh embodiment, when receiving a plurality of low density modulated waves, the processing means estimates transmission information by circulating between the detector and the decoder with an external likelihood value. , The residual interference component is reproduced based on the transmission information obtained from the circulation process, the residual interference component is subtracted from the received signal, and then the number of states of the trellis diagram is reduced by processing based on the MMSE norm. The signal after degeneration is used as the detector input. As a result, even if the number of trellis diagrams on which the detector is based is smaller than the number of states on the trellis diagram on the transmission side, interpolation is performed using the reproduced residual interference component, so the processing scale on the receiving side is further reduced. Can be reduced.

[Twelfth embodiment]
In the high-performance transmission system according to the twelfth embodiment, the transmission side transmits a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high-density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low density modulated wave on the transmission side can change its impulse response or frequency response according to the state of the transmission path. Thereby, it is possible to approach the capacity that the transmission path can realize.

[Thirteenth embodiment]
In the high-performance transmission system according to the thirteenth embodiment, the transmitting side transmits a plurality of low density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low-density modulated wave on the transmission side can change the impulse response according to the state of the transmission path. The impulse response is a response that is band-limited to a waveform obtained by taking a complex conjugate with respect to the impulse response of the transmission line and reversing the time axis. As a result, the maximum signal-to-noise ratio (SNR) can be achieved at the receiving end, and the capacity that can be realized by the actual transmission path can be approached.

[Fourteenth embodiment]
In the high-performance transmission system according to the fourteenth embodiment, the transmission side transmits a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high-density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low density modulated wave on the transmission side can change the transfer function according to the state of the transmission path. The transfer function is a function obtained by multiplying a function having a complex conjugate with respect to the transfer function of the transmission path by a transfer function for band limitation. As a result, the processing scale can be reduced even when the delay spread is large.

[Fifteenth embodiment]
In the high-performance transmission system according to the fifteenth embodiment, the transmission side transmits a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high-density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low-density modulated wave on the transmission side can change the impulse response according to the state of the transmission path. The impulse response is a response obtained by band-limiting a waveform obtained by taking a complex conjugate with respect to the impulse response of the transmission line and reversing its time axis. The reception side has a detector and a decoder using a metric calculation method using a half ring when receiving a plurality of low density modulated waves, and an external likelihood value is provided between the detector and the decoder. Transmission information is estimated by circulation. With this configuration, a processing unit based on the MMSE norm that reduces the number of states of the trellis diagram is provided at the front stage of the detector. The processing means takes a complex conjugate of the impulse response of the transmission line according to the state of the transmission line as a trellis diagram having the number of states after degeneration, and band-limits the waveform to a waveform obtained by reversing the time axis. Process to be a response. Thereby, detection by a trellis diagram capable of realizing the maximum SNR at the receiving end can be performed, and the capacity that can be realized by an actual transmission path can be approached.

[Sixteenth Embodiment]
In the high-performance transmission system according to the sixteenth embodiment, the transmission side transmits a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high-density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low density modulated wave on the transmission side can change the transfer function according to the state of the transmission path. The transfer function is a function obtained by multiplying a function having a complex conjugate with respect to the transfer function of the transmission path by a transfer function for band limitation. The reception side has a detector and a decoder using a metric calculation method using a half ring when receiving a plurality of low density modulated waves, and an external likelihood value is provided between the detector and the decoder. Transmission information is estimated by circulation. With this configuration, a processing unit based on the MMSE norm that reduces the number of states of the trellis diagram is provided at the front stage of the detector. The processing means is a trellis diagram having the number of states after degeneration and is a function obtained by multiplying the transfer function of the transmission path according to the state of the transmission path by a complex conjugate with the transfer function for band limitation. Process as follows. As a result, detection by a trellis diagram capable of realizing the maximum SNR at the receiving end can be performed, and the capacity that can be realized by an actual transmission path can be approached.

[Seventeenth embodiment]
In the high-performance transmission system according to the seventeenth embodiment, the transmission side transmits a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, thereby realizing a transmission speed equivalent to the high-density modulated wave transmitted by the Nyquist rate. And means for detecting the state of the transmission line. The waveform shaping filter of the low density modulated wave on the transmission side can change the transfer function according to the state of the transmission path. The transfer function is a function obtained by multiplying a transfer function of the transmission path by a complex conjugate and a transfer function for band limitation, and has means for executing the processing by calculation using the matrix T. Further, a means for converting the transmission signal vector S obtained by blocking the transmission signal by calculation using the discrete Fourier transform matrix D, and a transmission signal represented by the matrix as H, and the reception signal vector r blocked by

の式で表現したときに、

When expressed as

の式に従い特異値分解する手段とを有する。受信側は、複数の低密度変調波の受信を行うことに際し、受信ユニタリーマトリクスＵ^Ｈによって変換する手段を有し、送信ユニタリーマトリクスＶによって変換された送信信号ベクトルＳ’を

Means for performing singular value decomposition according to the following formula. The receiving side has means for converting with a reception unitary matrix U ^H when receiving a plurality of low density modulated waves, and transmits a transmission signal vector S ′ converted with the transmission unitary matrix V.

の式に従い検出する。これにより、検出器と復号器の間を外部尤度値によって循環させることにより送信情報を推定することが無いので、循環による遅延のない高性能伝送方式を実現することができる。

It detects according to the formula of As a result, the transmission information is not estimated by circulating between the detector and the decoder using the external likelihood value, so that a high-performance transmission system without delay due to circulation can be realized.

  FIG. 1 illustrates the configuration of a high-performance transmission system according to an embodiment of the present invention. In the transmission system shown in the figure, the upper part shows a transmission system (transmission side or transmitter), and the lower part shows a reception system (reception side or receiver).

  Among these, the transmission system includes an encoder 101, an interleaver (Π) 102, and a modulator (Tx Matrix) 103 as shown in FIG. In order to achieve the same transmission speed as that of, a plurality of low density modulated waves are transmitted at a speed exceeding the Nyquist rate. The modulator 103 includes a waveform shaping filter that performs waveform shaping of a plurality of low density modulated waves.

Also, as shown in the figure, the receiving system uses a semi-ring type detector (Semi-ring type Detector) that performs metric computation using a semi-ring (semi-ring) when receiving multiple low-density modulated waves. The detector 107 and the decoder 110 are configured, and the detector 107 and the decoder 110 are cycled by an external likelihood value via a pair of interleaver (１０８) 108 and deinterleaver ( ^{−１ −1} ) 109. To estimate the transmission information. In the preceding stage of the detector 107, a decoupling processing means (Decoupling) 105 and a processing means (Rx Matrix) 106 based on the MMSE standard for reducing the number of states, which will be described later, are provided. Hereinafter, each component will be described in detail.

  In the transmission system shown in the upper part of FIG. 1, transmission information is first encoded by the encoder 101, then stirred by the interleaver 102, and input to the modulator 103 that generates a plurality of low density modulated waves. . In the present embodiment, the modulator 103 generates three QPSK as a plurality of low density modulated waves in order to realize a transmission rate equivalent to 64QAM which is high density modulation transmitted by Nyquist rate. The plurality of low density modulated waves are transmitted at a speed exceeding the Nyquist rate. In this embodiment, the low density modulated waves are transmitted at one third of the Nyquist interval (Nyquist) as indicated by reference numeral 104 in FIG. The transmission rate in this case is 2 bits × 3 = 6 bits when converted by the Nyquist interval. On the other hand, the transmission speed of the high-density modulated wave is 6 bits × 1 = 6 bits when converted by the Nyquist interval in the case of 64QAM transmitted by Nyquist rate. Therefore, the transmission system of FIG. 1 has the same transmission rate as that of 64QAM which is equivalent to high density modulation. It should be noted that an interference component has already occurred at the stage of transmission before passing through the multipath transmission line as indicated by reference numeral 104 in FIG.

  The plurality of low density modulated waves generated in this way arrives at the receiving end through the multipath transmission path. A plurality of low density modulated waves are received by the receiving system shown in the lower part of FIG. The receiver 111 constituting this reception system has a detector 107 using a semi-ring metric calculation and a decoder 110 corresponding to the encoder 101 in the transmission system. Between the detector 107 and the decoder 110, an interleaver 108 that performs the same agitation processing as the interleaver 102 used in the transmission system and a deinterleaver 109 that performs the reverse process are interposed. A loop called feedback is formed. In addition, an amount of information called external likelihood passes between the detector 107 and the decoder 110, and high-performance transmission information can be estimated by this loop processing.

  Hereinafter, the configuration of the metric calculation by the half ring used in the detector 110 (hereinafter also referred to as “semi-ring configuration” or “half-ring configuration”) will be described.

  The semi-ring configuration realizes high speed and low complexity, and is a means for realizing the turbo principle based on the turbo feedback described above. The semi-ring, known as an algebraic structure with an abstract concept, is closely related to Bayesian statistics, and it is possible to realize marginalization processing that plays a central role in Bayesian statistics with a low degree of complexity. It is described in the following documents.

"July 1996, IEE Transactions on Information Theory, Vol. 42, No. 4, pp. 1072-1922 (IEEE Transactions on information theory, Vol.42, No.4, pp .1072-1092) "(Reference 1)
“July 2000, IEE Transactions on Information Theory, Vol. 46, No. 2, pages 325-343 (IEEE Transactions on information theory, Vol.46, No.2, pp .325-343) "(Reference 2)

  Processing such as BCJR algorithm for decoding a turbo code and Belief Propagation (BP) in LDPC can be regarded as calculation of probability propagation in a Bayesian network (BN). In calculating this probability propagation in the transition process on the graph, it is necessary to calculate the marginal probability distribution. An algorithm called Max-Log MAP in turbo decoding is used to calculate the marginal probability distribution necessary for each transition in a semi-ring of the Max-Log region and obtain the maximum posterior probability. Note that in turbo decoding, probability propagation is performed for operations on a finite field modulo 2. However, in the detector 107 of this embodiment, attention should be paid to the fact that operations are performed on real numbers.

  In the present embodiment, a half ring is formed of a matrix and a turbo principle is performed at a low speed with a low complexity to realize a high-performance transmission system when performing arithmetic operations on real numbers. This eliminates interference for signal waveforms that can be seen as partial response evolution systems realized by multiple low-density modulation schemes that exceed Nyquist rate instead of high-density modulation schemes that are vulnerable to multipath interference. Instead, the interference is used to realize high frequency utilization efficiency.

  Hereinafter, the semi-ring detector used in this embodiment will be described.

  The detector 107 shown in FIG. 1 uses the BCJR algorithm using a trellis diagram corresponding to waveform shaping by the modulator 103 that generates a plurality of low density modulated waves. The half ring is a method for realizing the BCJR algorithm at high speed and with low complexity. Here, it is possible to calculate probability propagation by a semi-ring known as an algebraic structure having an abstract concept, and it is possible to use properly according to the complexity and reliability such as the Max-Log area and the Log area. The half ring will be described below.

  A semi-ring is two operators

からなる。この半環上の加法と乗法は、以下の関係がある。

Consists of. The addition and multiplication on this half ring have the following relationship.

半環は、加法に対して単位元を持ち、ｍで表している。同様に、乗法に対しても単位元を持ち、

A semi-ring has unit elements for addition and is represented by m. Similarly, it has a unit element for multiplication,

で表している。加法における零元は、ｍである。

It is represented by The zero element in the addition is m.

  Hereinafter, A) Semi-ring expression in max-log domain, B) Semi-ring expression in log domain, C) Max-Log domain Extended Semi-ring expression in max-log domain, D) Extended Semi-ring expression in Log domain, E) Hybrid in Max-Log area and Log area Hybrid semi-ring expression (1) in max-log and log domain, F) Hybrid semi-ring expression (2) in Max-Log region and Log region (2) (Hybrid Semi-ring expression (2) ) in max-log and log domain).

A) Semi-ring in the Max-Log domain First, the Max-Log-MAP algorithm is related as follows using the relationship of the above [Equation 10]. In this case, the BCJR algorithm is interpreted as a semi-ring in the Max-Log domain and is expressed as:

この式で示されるように、半環上の加法は、２つの元ａ、ｂの和を最大値演算として、また半環上の乗法は、２つの元ａ、ｂの積を通常の加算として、それぞれ定義される。これらの関係より、上記［数１０］式の“ｍ”は、“−∞”に、

As shown in this equation, the addition on the half ring is the sum of the two elements a and b as the maximum value operation, and the multiplication on the half ring is the product of the two elements a and b as a normal addition. Are defined respectively. From these relationships, “m” in the above [Equation 10] becomes “−∞”,

は、“０”に、それぞれ対応することになる。

Corresponds to “0”.

  A half ring according to this definition is an approximation of the BCJR algorithm, but can be realized with low complexity.

B) Semi-ring in the Log domain Next, another interpretation of the BCJR algorithm is a semi-ring in the Log domain, which is expressed by the following equation.

この式で示されるように、半環上の加法は、２つの元ａ、ｂの和をヤコビアン対数に基づく演算として、また半環上の乗法は、２つの元ａ、ｂの積を通常の加算として、それぞれ定義される。

As shown in this equation, the addition on the half ring is an operation based on the Jacobian logarithm of the two elements a and b, and the multiplication on the half ring is a normal product of the two elements a and b. Each is defined as an addition.

  Of these equations, an operation based on the Jacobian logarithm is expressed as follows.

この式において、ｆ（＊）は補正関数を示し、ルックアップテーブル（look-up table）を用いて演算される。補正関数ｆ（＊）が理想的に作用すると、この定義による半環は、ＢＣＪＲアルゴリズムそのものになる。

In this equation, f (*) represents a correction function and is calculated using a look-up table. If the correction function f (*) works ideally, the semi-ring according to this definition becomes the BCJR algorithm itself.

C) Extended semi-ring in the Max-Log region Next, the extended semi-ring in the Max-Log region is obtained by adding the half ring addition and multiplication in the Max-Log region of A) to two elements a, Expanded from the sum and product of b to M (M is a natural number) elements a ₁ , a ₂ , ..., a _m , ..., a _M (m is a natural number satisfying 1≤m≤M). It is expressed as follows.

この式の定義による半環は、ＢＣＪＲアルゴリズムの近似値となる。

The semi-ring according to the definition of this equation is an approximate value of the BCJR algorithm.

D) Extended Semi-ring expression in log domain
Next, the extended semi-ring in the Log domain is obtained by adding the semi-ring addition and multiplication in the Log domain of B) to M (M is a natural number) elements a from the sum and product of the two elements a and b. ₁ , a ₂ ,..., A _m ,..., A _M (m is a natural number satisfying 1 ≦ m ≦ M) and is expressed as follows.

この式のうち、初めの式をヤコビアン対数に基づく演算とすると、以下のようになる。

If the first of these equations is an operation based on the Jacobian logarithm, the following is obtained.

この式において、ｆ（＊）は補正関数を示し、ルックアップテーブル（look-up table）を用いて演算される。補正関数ｆ（＊）が理想的に作用すると、この定義による半環は、ＢＣＪＲアルゴリズムそのものになる。

In this equation, f (*) represents a correction function and is calculated using a look-up table. If the correction function f (*) works ideally, the semi-ring according to this definition becomes the BCJR algorithm itself.

E) Hybrid half ring in Max-Log region and Log region (1)
Next, the hybrid half ring (1) in the Max-Log region and the Log region is defined as follows.

この式で示される

Indicated by this formula

は、Ｍａｘ値と二番目のＭａｘ値との差を示している。これによって、ルックアップテーブル（look-up table）のサイズと入力信号の数を減らすことができる。

Indicates the difference between the Max value and the second Max value. This can reduce the size of the look-up table and the number of input signals.

F) Hybrid half ring in Max-Log region and Log region (2)
Next, the hybrid semi-ring (2) in the Max-Log region and the Log region is defined as follows.

Ｔｒｅｅ構造を用いた場合、この式中のｍｉｎ（＊）の処理結果は、二番目のＭａｘ値となる。この式の定義による半環は近似値であるが、性能はＭａｘ−Ｌｏｇ領域とＬｏｇ領域の間になる。

When the Tree structure is used, the processing result of min (*) in this expression is the second Max value. The semi-ring according to the definition of this equation is an approximate value, but the performance is between the Max-Log region and the Log region.

  Based on the above definition, the half-ring configuration will be described.

  In the following example, for convenience of explanation, the number of states is set to 4, but it goes without saying that other numbers are possible. Now, the state metric and the γ metric (gamma metric) of the BCJR algorithm are respectively expressed as a 1 × 4 (1 × 4) vector S [n] and a 4 × 4 (4 × 4) matrix Γ [. n]. That is, the matrix Γ [n] is expressed as follows.

上記［数２２］式を用いて、時刻ｎのステートメトリックベクトルＳ［ｎ］を表すと、以下のようになる。

The state metric vector S [n] at time n is expressed using the above [Equation 22] as follows.

この式は、ＢＣＪＲアルゴリズムにおけるαメトリック（アルファメトリック）とβメトリック（ベータメトリック）の両方に適用することができる。

This equation can be applied to both α metric (alpha metric) and β metric (beta metric) in the BCJR algorithm.

  From the above equation, the state metric vector for every arbitrary K times is expressed as follows.

例えば、式［数２４］を元にαメトリックを例にとると、本実施例のセミリング構成の検出器で用いられるステートメトリック更新回路は、図２のように、アルファメトリック更新回路２０１、アルファメトリック用メモリ（Memory for alpha in metric）２０２、アップダウンカウンタ（U/D converter）２０３、及びガンマメトリック供給回路２０４により、構成される。

For example, taking the α metric as an example based on the formula [Equation 24], the state metric update circuit used in the detector of the semi-ring configuration of the present embodiment includes an alpha metric update circuit 201, an alpha metric as shown in FIG. A memory for alpha in metric 202, an up / down counter (U / D converter) 203, and a gamma metric supply circuit 204.

  In the figure, an alpha metric update circuit 201 on the left side performs an alpha metric sequential process, and alpha metric is changed from α [K · (n−1)] to α [K · (n)] in one clock. Update the vector. In the case of FIG. 3, since K = 3, the alpha metric is updated every three stages, and the calculation result is accumulated in the alpha metric memory 202. The memory address to be stored is designated by the up / down counter 203 for address control. Thereby, the memory capacity for alpha metric can be reduced to 1 / K, and the update speed can be improved K times.

  Gamma-metric matrix product over a half ring

は、予め計算されており、同図では、

Is calculated in advance,

として、

As

なる表現のガンマメトリックを用いる。このガンマメトリックは、ガンマメトリック供給回路２０４により、上記更新を行うアルファメトリック更新回路２０１へと供給される。また、次のアルファメトリックの更新のために、次のガンマメトリックＡ１［ｎ−１］、Ａ２［ｎ−１］、Ａ３［ｎ−１］及びＡ４［ｎ−１］が待機していて、淀みなく供給されるようになっている。

The gamma metric of the expression The gamma metric is supplied from the gamma metric supply circuit 204 to the alpha metric update circuit 201 that performs the update. Also, the next gamma metrics A1 [n-1], A2 [n-1], A3 [n-1] and A4 [n-1] are waiting for the next alpha metric update, and itching It comes to be supplied without.

  Next, ACS calculation results a11 to a14, a21 to a24, a31 to a34, a41 to a44 having a plurality of cascade connection configurations of the alpha metric update circuit 201 are obtained using the gamma metric supplied as described above as an input. The configuration will be described with reference to FIG.

  In the alpha metric update circuit shown in the figure, a gamma metric is supplied from the gamma metric supply circuit 303 to the first stage arithmetic processing unit 301 and the second and subsequent ACS arithmetic processing units 302.

  The first-stage arithmetic processing unit 301 performs a gamma metric addition operation. The detailed configuration is shown in FIG. As shown in the figure, the first stage arithmetic processing unit 301 can simplify the ACS operation and can be configured only by an adder. The calculation result by the first-stage calculation processing unit 301 is input to the ACS calculation processing unit 302 configured with a trellis structure including the second and subsequent-stage parallel components shown in FIG.

  The ACS arithmetic processing unit 302 has a trellis structure composed of parallel components. In the present embodiment, the ACS arithmetic processing unit 302 has a single-stage configuration, but this is because K = 3. When K of a larger value is used, the ACS arithmetic processing unit 302 has a multi-stage configuration. FIG. 5 shows details of the ACS arithmetic processing unit 302 configured with a trellis structure including the parallel components. As shown in the figure, the ACS arithmetic processing unit 302 includes logic circuits such as an adder, a comparator (comparator), and a multiplexer (data selector, signal switcher).

  3 to 5, the state metric update circuit used in the semi-ring detector of the present embodiment can achieve K times faster than the normal BCJR algorithm.

  The comparator used in FIGS. 2, 3 and 5 can also include a correction function f (*), and can adopt a semi-ring associated by the above-described various expressions. is there. For example, a semi-ring in the Log region using a Jacobian logarithm, a hybrid semi-ring in the Max-Log region, Max-Log and Log region, or the like. Depending on the degree of complexity and the degree of accuracy of the BCJR algorithm, various uses can be made.

  The LLR (Log-Likelihood ratio) for calculating the external likelihood value can be directly calculated by, for example, the following equation (in this example, when the number of states is 8 and K = 4, Yes.)

ここで、

here,

である。これによって、ステートメトリック更新の場合で、四つのＬＬＲが同時に得られ
４倍速となり、

It is. As a result, in the case of state metric update, four LLRs are obtained at the same time, resulting in quadruple speed,

となる。なお、上式は、状態数が８の場合で記述している。

It becomes. The above equation is described when the number of states is 8.

  Another example of the state metric update circuit used in the semi-ring detector of this embodiment is based on the unrolling method. FIG. 6 shows an example in which this method is adopted and the number of states is four. In the example shown in the figure, the K metric β metric is generated in one clock.

  The state metric update circuit shown in FIG. 6 includes a beta metric update circuit 601, an ACS arithmetic processing unit 602 having a cascade connection configuration of a plurality of stages, and a gamma metric supply circuit 603. In the figure, reference numeral 604 indicates the addition result of the beta metric and gamma metric, and reference numeral 605 indicates the calculation result of the gamma metric.

  In the state metric update circuit shown in the figure, the beta metric update circuit 601 on the right side updates the beta metric from β [K · (n + 1)] to β [K · n] every K = 3 stages in one clock. The The calculation processing result is input to an ACS calculation processing unit 602 having a cascade connection configuration of a plurality of stages.

The ACS arithmetic processing unit 602 has a multistage configuration of processing used in normal beta metric update. Therefore, the ACS arithmetic processing unit 602 performs the beta metric β [K · n].
That is,

を初期値として後方演算処理を縦続的に行うことになる。

As a default value, backward calculation processing is performed in a cascade manner.

  As a result, the gamma metric supplied by the gamma metric supply circuit 603 and necessary for the backward calculation processing is also pipelined for each stage. The gamma metric supply circuit 603 supplies the gamma metric with a delay of one step each time the stage advances. The reason why the step is delayed by one step every time the stage advances is because it is a backward operation, and thus the time is reversed. As a result, the process can proceed smoothly. As a result of such an operation, a beta metric and gamma metric addition result 604 can be obtained from each stage of the ACS arithmetic processing unit 602 having a cascade connection configuration. The addition result 604 is output in a pipelined configuration and delayed by one step as each stage progresses.

  Therefore, in this embodiment, three values are output in parallel without any stagnation, and the processing speed is K = 3 times that of the normally used BCJR algorithm. It goes without saying that any processing speed can be improved by increasing the value of K. Also, the gamma metric calculation result 605 in the figure is the same as that described in the case of alpha metric, so the description is omitted. However, since there is one point of difference, the update direction is that the alpha metric is increasing, while the beta metric is decreasing. The LLR can be calculated based on the addition result of the beta metric and the gamma metric obtained in this way and the above-described alpha metric.

  As described above, the semicircular configuration can reduce the complexity and improve the processing speed. However, in this embodiment, as shown in FIG. One of them is to arrange the processing means 106 based on the MMSE norm that performs the degeneration of the number of states shown in FIG.

  The number of states increases the turbo processing scale with the semi-ring configuration. Therefore, before the loop processing by the detector 107 and the decoder 110, the number of states is reduced to reduce the processing scale. In the case of the present embodiment, the degeneracy is realized by a matrix operation based on the MMSE standard that uses a signal including an interference wave. The method used in the present embodiment performs a degeneracy process in which only a state formed by a trellis diagram formed by a known low density modulated wave waveform shaping filter performed on the transmission side is performed. The processing based on the MMSE norm is originally processed so as to output a desired signal having no interference wave, but since it is used for the purpose of reducing the number of states, autocorrelation occurs when each element of the covariance matrix is viewed. . Furthermore, it should be noted that the cross-correlation vector is a correlation with the degenerated interference wave. In consideration of the above, degenerate matrix calculation is performed.

  In the case of the present embodiment, the waveform shaping filter of the low-density modulated wave on the transmission side has a linear phase characteristic and is composed of only the amplitude component. The detection can be performed independently of the axis component, and the detector only needs to have two surfaces for the real axis and the imaginary axis with reduced complexity. That is, the complexity has been reduced from the order of “(real axis complexity) × (imaginary axis complexity)” to the order of “(real axis complexity) + (imaginary axis complexity)”. Become.

  Now, let r be a received signal including a multipath transmission path, and d be a signal including ISI interference (inter-symbol interference) formed by a waveform shaping filter of a low-density modulated wave on the transmitting side.

によって、縮退用の重みベクトル

The weight vector for degeneration

を得ることができる。

Can be obtained.

  The above is a case where the waveform of the waveform shaping filter of the low-density modulated wave on the transmission side, which is known, is used as the degenerated signal, and has a linear phase characteristic and is composed only of an amplitude component. Therefore, the waveform after the degeneration can be processed independently for the real axis component and the imaginary axis component, and the detector 107 can be reduced in complexity by the effect of reducing the number of states due to degeneration and the effect of reducing the independent real axis imaginary axis. It is the Example implement | achieved by.

  Next, an embodiment will be described in which the processing scale of the processing means 106 based on the MMSE norm that reduces the number of states is reduced. As a means used for this purpose, in this embodiment, the I (In-phase) axis and the Q (Quadrature) axis can be processed independently before the processing 106 based on the degenerate MMSE norm. A decoupling processing means 105 which is a mechanism is arranged. Normally, IQ components are received in a mutually correlated form, but the decoupling processing means 105 separates the IQ correlations to reduce the processing scale of the processing means 106 based on the MMSE standard.

  That is, in order to perform processing based on the MMSE standard for signals having a mutually correlated form of IQ components, a complex operation is required as shown in [Equation 32], and complexity increases. On the other hand, if the correlation between IQs is separated and signals for the I axis and Q axis can be processed independently, complex arithmetic processing is eliminated, and the processing scale can be reduced.

  Hereinafter, a specific embodiment of the decoupling processing unit 105 will be described.

  There are two methods for decoupling processing, linear processing and non-linear processing. In this embodiment, non-linear processing is used.

  Now, the signal received through the multipath transmission line is

とすると、

Then,

ここで、ｓ（ｎ）は送信信号で、

Here, s (n) is a transmission signal,

で表される。

It is represented by

  F (l) is a coefficient of the multipath transmission line,

である。

It is.

  Received signal after decoupling processing

とすると、

Then,

の演算を施すことによって、

By performing the operation of

と

When

が一対一の関係として描ける。

Can be drawn as a one-to-one relationship.

  Similarly,

の演算を施すことによって、

By performing the operation of

と

When

が一対一の関係として描ける。

Can be drawn as a one-to-one relationship.

  here,

及び

as well as

は、ターボループからの送信信号推定値である。

Is a transmission signal estimation value from the turbo loop.

  That is, a signal having a correlation with IQ is separated by decoupling processing, and can be processed independently of the I axis and the Q axis. Note that these processes can be performed together with a method of performing MMSE processing by subtracting residual interference components described later.

  Next, degeneration processing using the MMSE norm in the frequency domain by the processing means 106 of FIG. 1 will be described.

  FIG. 7 shows an example of the configuration of processing means for performing degeneration according to the MMSE standard using estimated values in the frequency domain of a multipath transmission path. The processing means shown in the figure performs processing corresponding to the processing means 106 of FIG. 1, and includes a guard interval removal unit (GI deletion) 701, a serial / parallel conversion unit (S / P) 702, an FFT unit 703, and a degeneration unit. It has a frequency domain MMSE unit (MMSE weight) 704, an IFFT unit 705, and a parallel-serial converter (P / S) 706.

  In the figure, the received signal is first subjected to removal of the cyclic prefix (CP) added on the transmission side by the guard interval removing unit 701, and then parallelized by the serial-to-parallel converter 702 and input to the FFT unit 703. Is done. The input is converted from a time-domain signal to a frequency-domain signal by the FFT unit 703, and then the following weighting processing is performed by the frequency-domain MMSE unit 704 to perform degeneration.

  In the present embodiment, the signal generated by the waveform shaping filter of the low-density modulated wave on the transmission side is degenerated as much as possible, so that

ここで、

here,

は、マルチパス伝送路の周波数領域での推定値、Ｎはサンプル数、σｎ^２はノイズ分散である。

Is the estimated value in the frequency domain of the multipath transmission line, N is the number of samples, and σn ² is the noise variance.

  Next, the IFFT unit 705 converts the signal into a time domain signal, and the parallel / serial conversion unit 706 obtains serialized data. With the above processing, the degeneration processing using the MMSE norm in the frequency domain is completed. Since the degeneration process using the MMSE norm in this frequency domain changes all delays to phase information, the degeneration process can be realized with low complexity even for a transmission line with a large delay spread.

  Next, a second embodiment of the present invention will be described. In the transmission system according to the present embodiment, the residual interference component obtained from the turbo loop is reproduced on the receiving side and subtracted from the received signal, and then the degeneration processing based on the MMSE standard is performed.

  FIG. 8 shows a receiver to which the present embodiment is applied, and corresponds to the receiver 111 of FIG. 8 and 1 correspond to processing means 806 and 106, detectors 807 and 107, interleavers 808 and 108, deinterleavers 809 and 109, and decoders 810 and 110, respectively. In the following, the description of the parts overlapping with the components in FIG. 1 will be omitted or simplified, and only the characteristic parts excluding the overlapping parts will be described.

  In the receiver shown in FIG. 8, the LLR that circulates between the detector 807 and the decoder 810 is fed back from the output of the interleaver 808, and based on the LLR, the residual interference estimation unit (Replica) 805 generates a residual interference component. It is reproduced. The processing unit 806 based on the MMSE standard that uses the component to reduce the number of states subtracts from the received signal, performs MMSE processing, and outputs the result to the detector 807.

  The detector 807 performs metric calculation in a simplified form of the trellis diagram formed by the low-density modulation waveform shaping filter on the transmission side. For this reason, the processing scale of the detector 807 can be reduced. Even if such a reduction is performed, the residual interference estimation unit calculates the error between the trellis diagram formed by the low-density modulation waveform shaping filter on the transmission side and the trellis diagram assumed by the detector on the reception side. Since the residual interference component generated by 805 cancels, the characteristic does not deteriorate.

  The loop processing can include the above-described decoupling processing, and the residual interference component cancellation processing and decoupling processing can be performed simultaneously.

  The embodiment described above is a case where the state of the transmission path is unknown on the transmission side, but the embodiment described below is a case where the state of the transmission path is known on the transmission side.

  When the state of the transmission path is known on the transmission side, better performance can be obtained by increasing ISI interference within the same transmission power. In this case, the waveform shaping filter used in the modulator 103 that generates a plurality of low density modulated waves in FIG. 1 can change its impulse response or frequency response according to the state of the transmission path. In the example, it is controlled based on the following method.

  From the information theoretical point of view, this method is equivalent to increasing the capacity by introducing interference rather than canceling it in the channel with interference, and it is accompanied by the interference. This corresponds to extracting the maximum channel capacity of the propagation path. Until now, a reception method that can realize this has not been achieved at a level that can be realized, but according to the present embodiment, it is possible to draw out characteristics close to the maximum channel capacity at a relatively low complexity and at high speed. .

Now, the Fourier transform of the transmission signal wave s (t) is S (ω). If the transfer function of the multipath transmission path is H (ω) and the noise variance at the receiving end is σ ² , the received SNR is

となる。ここで、ｔｍは信号判定時刻である。

It becomes. Here, tm is a signal determination time.

  Assuming that the correlation with interference waves from other symbols is zero, consider the transmission signal waveform that maximizes the received SNR based on the constant transmission power in the above equation. It should be noted that SNR is maximized, not SINR (Signal to Interference plus Noise Ratio).

From the Schwarz inequality, if F ₁ (ω) and F ₂ (ω) are complex functions,

等号は、

The equal sign is

の時だけ成立する。ここで、kは任意の定数である。

It is established only when Here, k is an arbitrary constant.

と置くと、

And put

となる。送信電力が一定であるから、右辺は一定である。その条件で、左辺が最大になるのは、

It becomes. Since the transmission power is constant, the right side is constant. Under that condition, the left side is maximized

の時だけである。この時、受信ＳＮＲは最大になる。

Only when. At this time, the received SNR is maximized.

  Therefore, the waveform shaping of the transmission signal is as follows under band limitation.

ここで、ｈ＊（ｔ）は、マルチパス伝送路のインパルス応答の複素共役である。

Here, h * (t) is a complex conjugate of the impulse response of the multipath transmission line.

  That is, the waveform shaping filter for the low-density modulated wave on the transmission side can maximize the reception SNR by shaping the complex conjugate of the impulse response of the multipath transmission line and reversing its time axis. I understand. Furthermore, if the band is limited so as to transmit at a speed exceeding the Nyquist rate, the modulator 103 of this embodiment is obtained.

  Alternatively, depending on the transmission state, as shown in [Equation 54], a complex conjugate may be taken with respect to the transfer function of the transmission line, and this may be multiplied by a band limiting transfer function for transmission at a speed exceeding the Nyquist rate. The modulator 103 of this embodiment can be obtained.

  The plurality of low density modulated waves generated in this way passes through the multipath transmission path and arrives at the receiving end. The processing means 106 based on the MMSE standard for reducing the number of states of the trellis diagram installed in the front stage of the detector 107 on the receiving side is an impulse of the multipath transmission line as a trellis diagram having the number of states after the reduction. A complex conjugate is taken for the response, and processing is performed so that the response is band-limited to a shaped waveform with the time axis reversed. The detector 107 performs metric calculation using a trellis diagram based on the number of states after the degeneration.

  Alternatively, the processing means 106 based on the MMSE standard configured by frequency domain processing and performing reduction of the number of states of the trellis diagram installed in the front stage of the detector 107 on the receiving side may include a trellis having the number of states after the reduction. As a diagram, processing is performed such that a function obtained by taking a complex conjugate of the transfer function of the transmission path is multiplied by a transfer function for band limitation. On the other hand, the detector 107 performs metric calculation using a trellis diagram based on the number of states after the degeneration. Such a configuration is also possible.

  Next, a third embodiment of the present invention will be described. In the transmission system according to the present embodiment, the above-described method is further advanced, and the transmission waveform exceeding the Nyquist rate, the SNR maximizing transmission matrix processing, and the multipath interference waveform are represented in matrix, and the orthogonalization method using singular value decomposition is used. The transmission method is applied.

  A transmission path obtained by performing band-limiting processing on a function obtained by taking a complex conjugate with respect to the transfer function of the transmission path, where r is a received signal vector, s is a transmission signal vector, and D is a discrete Fourier transform matrix. If the processing that can change the transfer function in accordance with the state is an interference-emphasized band limiting matrix T, and the transmission path expressed in matrix is H and the noise vector is N, the received signal vector r is expressed by the following equation.

次に、送信ユニタリーマトリクスＶおよび受信ユニタリーマトリクスＵ^Ｈによる行列処理を行って、ユニタリー処理後の受信ベクトルをｙとすると、受信ベクトルｙは次式で表される。

Next, when matrix processing is performed using the transmission unitary matrix V and the reception unitary matrix U ^H and the reception vector after the unitary processing is y, the reception vector y is expressed by the following equation.

ここで、Ｎ’は受信ユニタリー演算後のノイズベクトルで、雑音強調は生じないで雑音電力の変化はない。また、ｓ’は送信ユニタリー演算後の送信信号ベクトルである。Λは、特異値行列で対角行列であるから、ｙより簡単にｓ’を、更に送信信号ベクトルＳ’を得ることができる。即ち、送信信号ベクトルＳ’は、

Here, N ′ is a noise vector after the reception unitary calculation, and noise enhancement does not occur and there is no change in noise power. Further, s ′ is a transmission signal vector after transmission unitary calculation. Since Λ is a singular value matrix and a diagonal matrix, s ′ and transmission signal vector S ′ can be obtained more easily than y. That is, the transmission signal vector S ′ is

の如く検出することができる。

It can be detected as follows.

  Therefore, in this embodiment, since there is no circulation between the detector and the decoder by the LLR, it is possible to realize a high-performance transmission system without a cyclic delay due to the turbo loop.

  Next, in order to show the effect of the embodiment of the present invention, based on the above-described configuration, Proakis's Channel model B, Channel model C, equal level 5-pass model, and AWGN (Additive White Gaussian Noise) channel without multipath are used. The result of having used it for the simulation is shown.

  A comparative example to be compared with the present embodiment shows a case where three QPSK signals are used as a plurality of low-density modulated waves that realize 64QAM and a transmission rate equivalent to 64QAM. Both signal bandwidths are the same. Further, processing on the transmission side is performed when the state of the propagation path is unknown on the transmission side and when the state of the propagation path is known on the transmission side.

  First, simulation results (FIGS. 10 and 11) when the propagation path state is unknown on the transmission side are shown, and then simulation results (FIG. 12) when the propagation path state is known on the transmission side are shown.

1) When the transmission path is an AWGN channel FIG. 9 shows the configuration of the transmission system used in the 64QAM simulation model of the comparative example. In the transmission system shown in the figure, the upper part shows a transmission system and the lower part shows a reception system.

  In the transmission system of FIG. 9, transmission information is first encoded by the encoder 901, stirred by the interleaver 902, and input to the 64QAM modulator 903. The output waveform of the modulator 903 coincides with the Nyquist interval as indicated by reference numeral 904 in FIG.

  In the receiver 909 constituting the reception system of FIG. 9, the received signal is processed by an equalizer 905 based on the MMSE norm, and a metric calculation is performed from a signal point by a demapper 906 to calculate an LLR, and a deinterleaver 907 The reverse process of the interleaver used in the transmission system is performed and input to the decoder 908. Then, the output of the decoder 908 is output from the receiver 909.

  The encoding method, interleaver, decoding method and processing method based on the MMSE standard when using the high-density modulated wave (64QAM) of the comparative example shown in FIG. This is exactly the same as the method using QPSK.

FIG. 10 shows the results of simulation in the case of using three QPSKs of the present example (Proposed method) and the case of using 64QAM of a comparative example (horizontal axis: E _b / N _o , vertical axis: BER). The simulation conditions were as follows: coding rate R = 0.793651, information bit length = 500 bits, frame length = 630 bits, and number of signals = 630 / M.

  According to this, in the case of 64QAM of the comparative example, since error correction is performed at the coding rate R = 0.793651, the characteristics corresponding to the coding gain are improved compared to the uncoded 64QAM itself. On the other hand, the characteristics of the system using the three QPSKs of this embodiment are further improved. This confirms that higher frequency utilization efficiency can be achieved with the same transmission power by transmitting a plurality of low density modulated waves (QPSK) at a speed exceeding the Nyquist rate instead of the high density modulated waves (64QAM). It was.

2) When the transmission path is Proakis's Channel model B and Channel model C Next, FIG. 11 shows a case where three QPSKs of this embodiment (Proposed method) are used and a case where 64QAM of a comparative example is used. , result of a simulation of a multipath transmission channel shows a (horizontal _{axis:: E b / N o (} energy-per-bit to noise density ratio), the vertical axis BER (bit Error Rate)). In the simulation model using 64QAM of the comparative example, the characteristic deterioration is significant, and as a comparative example other than that, the frequency utilization efficiency is 1/3, but QPSK transmitted by Nyquist rate is also used as a comparative example. The simulation conditions were set as follows: coding rate R = 0.793651, information bit length = 500 bits, frame length = 630 bits, and number of signals = 630 / M.

  This simulation result is obtained when the state of the propagation path is unknown on the transmission side. The multipath transmission lines used are Proakis's Channel model B and Channel model C, both of which are channel models for equalizer evaluation. Therefore, as the channel model B changes to the channel model C, considerable characteristic degradation occurs. Using this channel model, the characteristics were compared between the case of using the three QPSKs of the present example and the case of using 64QAM and QPSK of the comparative example.

  According to the simulation results shown in FIG. 11, the method of this example shows better characteristics than QPSK with a frequency utilization efficiency of 1/3 even in Proakis's Channel model C, which is the most severely deteriorated. Therefore, it was found that the higher the density modulation, the more affected by multipath interference peculiar to mobile communication. It was also found that the method of this example is much more resistant to multipath interference. As a result, by transmitting a plurality of low density modulated waves (QPSK) instead of high density modulated waves (64QAM) at a speed exceeding the Nyquist rate, a higher frequency can be obtained with the same transmission power without being affected by multipath interference. It was confirmed that the utilization efficiency could be achieved.

3) When the transmission path state is known on the transmission side and the equi-level five-path model is used Next, FIG. 12 shows a scheme (QPSK × 3 (degenerate turbo reception) using the semi-ring configuration of this embodiment. = 64QAM)), and when the state of the propagation path is known on the transmission side and when the state of the transmission line is unknown on the transmission side, the multi-path transmission line using the equi-level 5-path model The results of simulation (horizontal axis: E _b / N _o , vertical axis: BER) are shown. The simulation conditions were set as follows: coding rate R = 0.793651, information bit length = 500 bits, frame length = 630 bits, and number of signals = 630 / M.

  When the state of the transmission line is known on the transmission side, the interference wave is enhanced according to the state of the transmission line. That is, the impulse response or frequency response of the waveform shaping filter of the low-density modulated wave is changed according to the state of the transmission line, according to [Equation 55] or [Equation 54], within the band limitation and under the same transmission power. . In this method of enhancing interference, processing on the transmission side is performed within the band limitation using the state of the transmission path so that the maximum is based on SNR, not SINR at the receiving end. The channel model used is an equilevel 5-pass model.

  According to the simulation result shown in FIG. 12, when the state of the propagation path is known on the transmission side, if the interference is increased within the same transmission power, the transmission side shows a better performance than when the state of the transmission path is unknown. I understood. From an information-theoretic perspective, this is equivalent to increasing the capacity by incorporating the interference rather than canceling it in a channel with interference, and propagating with the interference further. This is equivalent to extracting the maximum communication channel capacity of the channel. As a result, the conventional method has not been able to achieve a reception method capable of realizing it, but according to the present embodiment, the characteristics close to the maximum channel capacity at high speed with relatively low complexity. It was confirmed to pull out.

  Although the present invention has been described with reference to the embodiments and examples, the present invention is not limited to the above embodiments and examples. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2008-049886 for which it applied on February 29, 2008, and takes in those the indications of all here.

  As described above, according to the present invention, the present invention can be used for a high-performance transmission system, a transmission method, a receiver, and a transmitter using a multipath transmission line as a transmission medium.

Claims (20)

A transmission system for transmitting information between a transmission side and a reception side via a transmission line,
In order to realize a transmission speed equivalent to a high-density modulated wave transmitted by Nyquistrate, it has means for transmitting a plurality of low-density modulated waves at a speed exceeding Nyquistrate,
The receiving side has a detector and a decoder for performing a metric operation using a semi-ring when receiving the plurality of low density modulated waves, and the detector and the decoder are externally connected. A high-performance transmission system, characterized in that transmission information is estimated by cycling according to a likelihood value. The transmission side has a waveform shaping filter that performs waveform shaping of the plurality of low density modulated waves,
The detector is a semi-ring in the Max-log region as a metric operation using the semi-ring, with respect to an operation on a trellis diagram formed by the impulse response of the waveform shaping filter and the transmission line, As the definitions of addition and multiplication on the half ring, the sum of elements a and b is the maximum value operation (MAX), and the product of elements a and b is normal addition.

The high-performance transmission system according to claim 1, wherein the calculation is performed according to the equation: The receiving side has processing means based on a MMSE (Minimum Mean Square Error) standard for performing degeneracy of the number of states of the trellis diagram in the previous stage of the detector,
The high-performance transmission system according to claim 2, wherein the detector performs a metric operation using a trellis diagram based on a degenerated state number. The processing means performs the degeneration of the number of states so that the trellis line formed by the waveform shaping filter with respect to the trellis line formed by the impulse response of the waveform shaping filter and the transmission line,
4. The high-performance transmission system according to claim 3, wherein the detector performs a metric operation using a trellis diagram formed by the waveform shaping filter as a trellis diagram according to a degenerated state number.   5. The high-performance transmission system according to claim 4, wherein the waveform shaping filter has a linear phase characteristic and includes only an amplitude component.   The detector, when performing a metric calculation using a trellis diagram based on the number of states after degeneration, performs an operation using a trellis diagram based only on a real axis or an imaginary axis component formed by the waveform shaping filter. 6. The high-performance transmission system according to claim 5, wherein The detector is
A pipelined multi-stage gamma metric is used in at least one of the alpha metric calculation and beta metric calculation in the BCJR (Bahl, Cocke, Jelinek, Raviv) algorithm or ACS (Add, Compare, Select) calculation. Means for supplying;
ACS calculation means comprising a cascaded configuration of a plurality of stages with the pipelined gamma metric as input,
Means for updating the state metric for each of a plurality of stages (k stages) with the calculation result as an input;
The high-performance transmission system according to claim 1, comprising:   The processing means is created using a covariance matrix for a direct product of the impulse response of the waveform shaping filter and the transmission path, and a cross-correlation vector of the impulse response of the waveform shaping filter and the transmission path. The high-performance transmission system according to any one of claims 3 to 6, wherein weighting processing is performed.   When the reception side estimates transmission information by circulating between the detector and the decoder with an external likelihood value, based on the transmission information obtained from the circulation process, the received signal is taken as a real axis. 3. A means for performing a decoupling process for processing independently of an imaginary axis component, wherein the correlation between the real axis component and the imaginary axis component is separated to reduce the number of states. High performance transmission system as described.   The processing means is processing in the frequency domain, and after conversion to the frequency domain, weights the signal after degeneration of the number of states as much as possible, performs conversion to the time domain, and converts the signal after the degeneration. The high-performance transmission system according to any one of claims 3 to 6, wherein the high-performance transmission system is an input of the detector.   The processing means reproduces the residual interference component based on the transmission information obtained from the circulation process when estimating the transmission information by circulating between the detector and the decoder with an external likelihood value. Then, after subtracting the residual interference component from the received signal, the number of states of the trellis diagram is reduced by processing based on the MMSE norm, and the signal after the reduction is used as the input of the detector. Item 11. The high-performance transmission system according to any one of Items 3, 4, 6, 8, and 10. The transmission side has means for detecting the state of the transmission path,
The high-performance transmission system according to any one of claims 3 to 11, wherein the waveform shaping filter is capable of changing an impulse response or a frequency response in accordance with a state of the transmission path. The transmission side has means for detecting the state of the transmission path,
The waveform shaping filter can change the impulse response according to the state of the transmission line, and the impulse response can be obtained by taking a complex conjugate with respect to the impulse response of the transmission line and reversing its time axis. The high-performance transmission system according to any one of claims 2 to 12, wherein the response is a band-limited response to the waveform. The transmission side has means for detecting the state of the transmission path,
The waveform shaping filter can change its transfer function according to the state of the transmission line, and the transfer function is a band limiting function in a function that takes a complex conjugate with respect to the transfer function of the transmission line. The high-performance transmission system according to claim 2, wherein the high-performance transmission system is configured to be a function obtained by multiplying a transfer function. The waveform shaping filter can change the impulse response according to the state of the transmission line, and the impulse response can be obtained by taking a complex conjugate with respect to the impulse response of the transmission line and reversing its time axis. This is a response with band limitation applied to
The processing means, as a trellis diagram having the number of states after the degeneration, forms a band in a waveform obtained by taking a complex conjugate with respect to the impulse response of the transmission line according to the state of the transmission line and reversing its time axis. 14. The high-performance transmission system according to claim 13, wherein processing is performed so as to obtain a limited response. The waveform shaping filter can change its transfer function according to the state of the transmission line, and the transfer function is a band limiting function in a function that takes a complex conjugate with respect to the transfer function of the transmission line. A function multiplied by a transfer function,
The processing means, as a trellis diagram having the number of states after the degeneracy, multiplies a transfer function for a band limit by a function that takes a complex conjugate with respect to the transfer function of the transmission line according to the state of the transmission line. 15. The high-performance transmission system according to claim 14, wherein processing is performed so as to be a function. A transmission system for transmitting information between a transmission side and a reception side via a transmission line,
The transmission side detects means for transmitting a plurality of low-density modulated waves at a speed exceeding the Nyquist rate, and detects the state of the transmission path, in order to realize a transmission rate equivalent to the high-density modulated wave transmitted by the Nyquist rate. Means, and a waveform shaping filter for shaping the waveform of the low density modulated wave,
The waveform shaping filter can change its transfer function according to the state of the transmission line, and the transfer function is a band limiting function that is a complex conjugate of the transfer function of the transmission line. A function obtained by multiplying the transfer function by the following, and having means for executing the process by an operation using the matrix T,
Means for converting a transmission signal vector S obtained by blocking a transmission signal by a calculation using a discrete Fourier transform matrix D;
Assuming that the transmission path expressed in a matrix is H, the received signal vector r that is blocked is

When expressed as

Means for singular value decomposition according to the equation
The transmission side has means for converting by the reception unitary matrix UH when receiving the plurality of low density modulated waves, and transmits the transmission signal vector S ′ converted by the transmission unitary matrix V.

A high-performance transmission system characterized by detecting according to the equation: A transmission method for transmitting information between a transmission side and a reception side via a transmission line,
In order to achieve the same transmission speed as high-density modulated waves transmitted by Nyquistrate, multiple low-density modulated waves are transmitted at a speed exceeding Nyquistrate,
When receiving the plurality of low density modulated waves on the receiving side, the detector and the decoder perform a metric calculation using a semi-ring, and an external likelihood is provided between the detector and the decoder. A high-performance transmission method characterized in that transmission information is estimated by cycling according to a value. A receiver of a transmission system for transmitting information between a transmission side and a reception side via a transmission line,
In order to achieve the same transmission speed as high-density modulated waves transmitted by Nyquistrate, metric calculation using a semi-ring when receiving multiple low-density modulated waves transmitted at a speed exceeding Nyquistrate A receiver having a detector and a decoder for performing transmission, and estimating transmission information by circulating between the detector and the decoder with an external likelihood value.   The receiver according to claim 19, wherein a plurality of low density modulated waves are transmitted at a speed exceeding the Nyquist rate in order to realize a transmission speed equivalent to the high density modulated wave transmitted by the Nyquist rate. Transmitter.

Images