JP4763057B2 - Method and apparatus for normalizing metrics input to channel decoder in wireless communication system - Google Patents

Description

  The present invention relates to wireless communication systems, and more particularly to a method and apparatus for normalizing metrics input to a channel decoder.

  CDMA (Code Division Multiple Access) 2000, WCDMA (Wideband-CDMA), and IEEE (Institute of Electrical and Electronics Engineers) 802.16 systems are QPSK (Quadrature Phase Shift Keying), 8PSK, 16-QAM (16-ary Quadrature). Amplitude Modulation) and 64-QAM are performed. Furthermore, these systems perform adaptive modulation and coding (AMC) with a combination of channel codes such as turbo codes. This system obtains the optimum transmission rate corresponding to the channel condition. The reception stage calculates an LLR (Log Likelihood Ratio) per bit with a demapper by various modulations, and acquires an input metric to the channel decoder. The channel decoder receives and decodes the metric.

  FIG. 1 shows a configuration of a transceiver in a conventional wireless communication system.

  Referring to FIG. 1, binary data i (n) to be transmitted is encoded by a channel encoder 110 in the transmitter 100. The channel encoder 110 generates a series of binary code symbols c (n). The mapper 120 generates several code symbol blocks among the generated code symbols, maps them to one point on the constellation, and converts them to complex-valued modulation symbols x (n). To do. This modulation symbol x (n) is applied to the modulator 130. The modulator 130 generates a continuous-time wave by a modulation symbol x (n) by a CDMA method or an orthogonal frequency division multiplexing (hereinafter referred to as “OFDM”) method, and receives a receiver via a channel 140. 150.

  In receiver 150, demodulator / channel estimator 160 performs baseband demodulation and channel estimation processes on the received signal. The demodulator can be realized by various techniques. For example, the demodulator can be an OFDM demodulator implemented with a CDMA rake receiver or an IFFT (Inverse Fast Fourier Transform) process and a channel estimator. After demodulating the baseband, a channel estimate c (n) modulated by QAM or PSK and a received symbol y (n) are obtained.

  The demapper 170 uses the received symbol y (n) and the channel estimation value c (n) to calculate a metric for the bits forming the codeword of the channel code. The sequence Λ (n) corresponding to the metric value calculated by the demapper 170 is input to the channel decoder 180 and decoded into the original transmitted binary data. When the channel decoder 180 completes the decoding operation, the receiver 150 completes the basic operation in the physical layer. At this time, the channel decoder 180 includes a Viterbi decoder corresponding to the convolutional code, a SOVA (Soft Output Viterbi Algorithm) iterative decoder corresponding to the turbo code, a log-MAP (Maximum a Posteriori) iterative decoder, and a max-log-MAP iterative decoder. Etc. can be used.

  In the realization of the conventional wireless communication system operating as described above, the dynamic range of the metric input to the decoder is generally not limited when performing floating point operations. However, when hardware that performs a fixed point operation is realized, the dynamic range is affected by quantization noise, clipping noise, and the like. Thus, each step of the communication system must ensure optimal performance with minimal hardware by performing appropriate normalization on the metric representation. However, since the conventional method does not consider the normalization of the metric calculated by the demapper, the performance is lower at a high code rate and higher-order modulation than the normal code rate and modulation. was there.

  Therefore, the present invention has been proposed in view of the above-described problems of the prior art, and its purpose is to obtain optimum performance by channel decoding based on an LLR metric with a small number of bits in a wireless communication system. It is to provide a method and apparatus that can be used.

  Another object of the present invention is to provide a method and apparatus for improving the decoding performance with a minimum number of bits by performing normalization of a metric used as an input of a channel decoder in a wireless communication system.

  Another object of the present invention is to provide a method and apparatus capable of appropriately normalizing a metric used as an input of a channel decoder according to a modulation order and a current noise level in a wireless communication system. is there.

  Furthermore, an object of the present invention is to appropriately use information on modulation order, channel coding rate, and channel code frame length when there is no information on noise dispersion used as an input of a channel decoder in a wireless communication system. It is an object of the present invention to provide a method and apparatus capable of performing normalization.

According to an aspect of the present invention, a normalization apparatus for soft metrics input to a channel decoder in a wireless communication system, the in-phase component (X _k ) and the quadrature component (Y _k ) of a received modulation symbol (R _k ). ), A demapper for generating a soft metric using at least one of a channel fading coefficient (g _k ) and a constant value (c) determined by the modulation order of the received modulation symbol, receiving the soft metric, and noise The ratio of the constant value to the variance value is multiplied by the soft metric to obtain a normalized LLR (Log Likelihood Ratio), the normalized LLR is converted into a predetermined range and the number of bits, and the channel decoder input LLR is output. And a normalizer.

According to another aspect of the present invention, an apparatus for normalizing a soft metric input to a channel decoder in a wireless communication system, the in-phase component (X _k ) and the quadrature component (Y) of a received modulation symbol (R _k ). _k ), a channel fading coefficient (g _k ), and a demapper that generates a soft metric using at least one of a constant value (c) defined by the modulation order of the received modulation symbol; Multiplying a normalization coefficient obtained from adaptive modulation and coding (AMC) information by a soft metric to obtain a normalized LLR (Log Likelihood Ratio), and converting the normalized LLR into a predetermined range and number of bits And a normalizer that outputs the input LLR of the channel decoder.

According to still another aspect of the present invention, there is provided a method for normalizing a soft metric input to a channel decoder in a wireless communication system, in-phase component (X _k ) and quadrature component (X _k ) of a received modulation symbol (R _k ). Generating a soft metric using at least one of a constant value (c) defined by Y _k ), a channel fading factor (g _k ), and a modulation order of the received modulation symbol; and receiving the soft metric A step of obtaining a normalized LLR (Log Likelihood Ratio) by multiplying a ratio of a noise variance value to a constant value by a soft metric, a step of converting the normalized LLR into a predetermined range and the number of bits, and a channel And outputting an input LLR of the decoder.

Furthermore, according to another aspect of the present invention, there is provided a method for normalizing a soft metric input to a channel decoder in a wireless communication system, the in-phase component (X _k ) and the quadrature component of a received modulation symbol (R _k ). Generating a soft metric using at least one of (Y _k ), a channel fading coefficient (g _k ), and a constant value (c) defined by the modulation order of the received modulation symbol, and receiving the soft metric A step of obtaining a normalized LLR by multiplying a normalization coefficient obtained by adaptive modulation and coding (AMC) information by a soft metric, and converting the normalized LLR into a predetermined range and the number of bits. And outputting an input LLR of the channel decoder.

  According to the embodiment of the present invention, in the wireless communication system, the channel has a different value for each symbol through normalization of the soft output from the demapper. In addition, even in the case of an OFDM system that requires a high metric resolution, the desired performance can be obtained by inputting a smaller number of bits to the turbo decoder.

  Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  When it is determined that a specific description related to a known function or configuration related to the present invention makes the gist of the present invention unclear, a detailed description thereof will be omitted. The terminology described below is defined in consideration of the function of the present invention, and may vary depending on the intention or practice of the operator.

  Embodiments of the present invention provide a method and apparatus for obtaining optimal decoding performance with a low bit number LLR metric when encoding a channel. Also, the embodiment of the present invention can improve the decoding performance with a small number of bits by normalizing the input metric of the channel decoder.

<First Embodiment>
The first embodiment of the present invention provides a structure and operation procedure for performing normalization using information on noise variance used as an input of a channel decoder.

  FIG. 2 shows a configuration of a wireless communication transceiver to which the metric normalizer according to the first embodiment of the present invention is applied.

  Referring to FIG. 2, the transmitted binary data i (n) is encoded by a channel encoder 210 in the transmitter 200. The channel encoder 210 generates a series of binary code symbols c (n). The mapper 220 generates several code symbol blocks among the generated code symbols, maps them to one point on the constellation, and converts them to complex-valued modulation symbols x (n). This sequence x (n) is applied to the modulator 230. The modulator 230 generates a continuous-time wave by a CDMA method or an orthogonal frequency division multiplexing (hereinafter referred to as “OFDM”) method using symbols, and transmits the generated continuous-time wave to the receiver 250 via the channel 240.

  In the receiver 250, a demodulator / channel estimator 260 performs a baseband demodulation and channel estimation process on the signal passing through the channel 240. The demodulator can be realized by a technique applied to the baseband. For example, the demodulator can be an OFDM demodulator implemented with a CDMA rake receiver or an IFFT (Inverse Fast Fourier Transform) process and a channel estimator.

  In the embodiment of the present invention, an IEEE 802.16e and OFDMA system will be basically described. After demodulating the baseband by demodulator / channel estimator 260, the received symbols and channel estimates are output to noise variance estimator 265 and demapper 270.

を多様なアルゴリズムで推定してＬＬＲ正規化器２７５に伝送する。 The noise variance estimator 265 calculates a noise variance value based on the channel estimation value.

Is estimated by various algorithms and transmitted to the LLR normalizer 275.

  The demapper 270 receives the received symbol y (n) and the channel estimation value c (n) modulated by QAM (Quadrature Amplitude Modulation) or PSK (Phase Shift Keying) from the demodulator / channel estimator 260, and performs de-mapping. Output a metric for each bit. The demapper 270 can obtain the metric using various algorithms. In the demapping method, a simple algorithm close to the optimum algorithm is generally used. One of various methods is described in Reference 1 (Y. Xu, H.-J. Su, E. Geraniotis, “Pilot symbol assisted QAM with interleaved filtering and turbo decoding over Rayleigh flat-fading channel”, in Proc. This is a dual minimum metric method proposed in MILCOM '99, pp.86-91).

  The IEEE 802.16e system uses 16QAM or 64QAM high order modulation. Signals transmitted after such modulation may be distorted by channel fading and noise. In the IEEE 250.16 system receiver 250, the convolutional turbo decoder, which is the channel decoder 280, receives and decodes the soft metric corresponding to the reliability information of each bit. Therefore, the received signal distorted in the previous stage of the channel decoder 280. The process of calculating the soft metric from Such a process is performed by a demapper 270 in the receiver 250. Here, a demapping algorithm applicable to the present invention will be described.

The IEEE 802.16 system uses QPSK, 16QAM, and 64QAM modulation. If the number of bits representing one modulation symbol in the output sequence of the binary channel encoder is m, the number of signal points in the signal point arrangement diagram is M = 2 ^m (here And m = 2, 4, 6,... The m bits are mapped to a specific signal point among the signal points. By formulating the M-QAM mapping, the in-phase and quadrature components of the modulation symbol can be obtained from m binary symbols as shown in the following equation (1).

In Equation (1), s _{k, i} (i = 0, 1,..., M−1) represents the i-th symbol in the output sequence of the binary channel encoder mapped to the k-th signal point. x _k and y _k represent the in-phase component and the quadrature component of the kth signal point, respectively. In the case of 16QAM, m = 4.

  3A to 3C show signal point arrangement diagrams of QPSK, 16QAM, and 64QAM, respectively.

As can be seen from FIGS. 3A-3C, the x _k of the modulated symbol is determined by s _{k, m−1} , s _{k, m−2} ,..., S _{k, m / 2} , and y _k is s _k, determined by _{m / 2-1} ,..., _{sk, 0} . A constant c that determines the arrangement point of each signal is defined as shown in Equation (2). This is a value for setting the average energy of the symbol to 1.

Here, _{c 4} is the reference value of the _{QPSK, c 16} is the 16QAM reference _{value, c 64} represents each reference value of the 64QAM,. The modulation symbol has a complex value of x _k + jy _k . After this modulation symbol passes through the channel 240 and the baseband demodulator 260, a signal as shown in Equation (3) is input to the demapper 270.

Here, g _k is a channel fading coefficient and is expressed as g _k = g _xk + jg _yk . n _xk and _{n yk} is the noise and interference components. The LLR (Log Likelihood Ratio) of the bit symbol sk _{, i} corresponding to the element of the QAM symbol can be approximated as shown in Equation (4).

は雑音及び干渉の分散である。 Here, z _k (s _{k, i} = 0) is a modified signal constellation point _obtained by multiplying a symbol with s _{k, i} = 0 by a fading constant g _k ,

Is the variance of noise and interference.

  In Equation (4), the Max-Log MAP method is applied to the calculation of the LLR, and a high reliability estimation value is obtained using a small amount of calculation. Equation (4) can be approximated as equation (5) below.

はｎ_ｋ，ｉの否定(negation）である。 Here, n _{k, i} is the i-th information bit value mapped to the constellation point of the signal closest to the received symbol R _k ,

Is the negation of _{nk, i} .

The bit symbols _{sk, i} forming QPSK, 16QAM, and 64QAM symbols are each related to only one of the in-phase component and the quadrature component of the received symbol. For R _k and z _{k in} equation (5), one of the x and y axis components is removed by s _{k, i} .

FIG. 4 shows an example of LLR calculation when g _k is a real value.

Assuming that R _k is received, the LLR of S ₃ can be defined by equation (6) as shown in FIG.

はすべての場合に存在し、括弧内の部分は入力信号に対する線形計算式(linear equation)である。 When the LLR is calculated by a method like Equation (6), the coefficient

Exists in all cases, and the part in parentheses is the linear equation for the input signal.

In this embodiment, the demapper can be realized by a linear function of a soft metric generator (hereinafter referred to as “SMG”). After the constants including the fading coefficient g _k are input to the SMG, they can be processed in an appropriate scaling method.

を除去してソフトメトリックを生成する関数がＳＭＧ(ａ，ｂ)であると仮定すれば、式(６)に示すＬＬＲ計算は式(７)のように書き換えることができる。 LLR to coefficient

Assuming that the function for generating a soft metric by removing SMG (a, b) is, the LLR calculation shown in equation (6) can be rewritten as equation (7).

Equation (7) shows the LLR calculation formula related only to the in-phase component X _k. Of course, LLR calculation formula related only to the quadrature component _{Y k} is | _{g k} | instead of ^{_{2 X k | g k | 2}} Y k can also be used.

The inputs of this SMG are | g _k | ² X _k , | g _k | ² Y _k , and | g _k | ² c, where g _k is obtained from the channel estimate. Thus, the SMG input can be easily calculated from the received symbols and channel estimates.

When g _k is a complex number, the input of the SMG mapped to the in-phase signal component and the quadrature signal component is as shown in the following equation (8).

  That is, the SMG input in equation (8) is easily calculated from the received signal using equation (9).

において、４の値はＱＰＳＫ、１６ＱＡＭ、６４ＱＡＭの共通の係数であるため、量子化が反映される。 Coefficients applied identically between SMG outputs

Since the value of 4 is a common coefficient for QPSK, 16QAM, and 64QAM, quantization is reflected.

は、正規化(normalization)がソフト出力の生成後に遂行され、量子化されたＬＬＲが適切な範囲(range)と解像度(resolution）を有するように設定される。

The normalization is performed after the soft output is generated, and the quantized LLR is set to have an appropriate range and resolution.

At that time, the metric to be calculated by the demapper 270 can be simplified as shown in Expression (10), and the function of SMG _i () is a simple linear operation realized only by a shift operation and an adder.

である。 Where a _k is

It is.

Equation (10) is used to calculate the soft metric associated with the in-phase component. SMG _i (Q _k , a _k ) is used to calculate a soft metric associated with the orthogonal component as shown in equation (7).

As shown in the table below, the 16QAM metric obtained from the function of SMG _i () is a domain (domain) to which the in-phase signal component I _k and the quadrature signal component Q _k calculated from the received symbol and the channel fading coefficient belong. ). In order to determine the soft metric, only I _k , Q _k , and a _k are considered.

Table 1 shows 16QAM metrics generated from I _k , and Table 2 shows 16 QAM metrics generated from Q _k . In the same way, soft bit metrics of Λ (S _{k, 5} ), Λ (S _{k, 4} ), and Λ (S _{k, 3} ) related to 64QAM are calculated as shown in Table 3 below. Can be done. Also, Λ (S _{k, 2} ), Λ (S _{k, 1} ), and Λ (S _{k, 0} ) can be calculated from Q _k . Next, a description will be given soft output associated with I _k.

<Table 3> shows the soft metric of 64QAM generated from _{I k.} In this way, QPSK, 16QAM, and 64QAM soft outputs can be calculated.

が削除されることによって計算される。 However, this soft output value itself is derived from equation (7) to represent the input LLR of the decoder.

Is calculated by deleting.

  In a hardware embodiment, the dynamic range of the decoder's input metric may increase excessively or degrade performance.

が正規化に反映される。 Therefore,

Is reflected in normalization.

  FIG. 5 shows an operation configuration of an input metric normalizer according to the second embodiment of the present invention.

の値を反映するためのメトリック正規化器の例を示す。 FIG.

An example of a metric normalizer to reflect the value of

  Since the “c” value is stored according to the modulation schemes of QPSK, 16QAM, and 64QAM, the normalizer 275 sets the “c” value when receiving the modulation order or the modulation information mod_order mapped thereto. be able to.

を計算するために、雑音分散推定器(図２の参照番号２６５)が必要である。 Noise variance corresponding to the variance of the sum of noise and interference

To calculate a noise variance estimator (reference number 265 in FIG. 2).

の雑音分散値を推定することができる。 The noise variance estimator 265 uses a variance algorithm.

Can be estimated.

を受信する。 In the normalizer 275, the multiplier 520 is calculated by converting the variance value using the conversion table 510 reflecting the division.

Receive.

によってデマッパ２７０からのメトリックΛ(ｎ)を乗算してＬＬＲの正規化を遂行する。 Multiplier 520

Is multiplied by the metric Λ (n) from the demapper 270 to perform LLR normalization.

  After the LLR has been normalized, a rounding / clipping section 530 inputs LLR Λ ′ (n) having the desired range and number of bits to the decoder. Depending on the modulation order or coding rate supported by the system, the number of bits M of the input metric is approximately 24-26, and the number of normalized output bits is 6-8.

  In FIG. 5, the noise variance can be estimated by various methods. For example, it is disclosed in Reference Document 2 (TASummers and SGWilson, “SNR mismatch and online estimation in turbo decoding”, IEEE Trans. Commun. Vol. 46, no. 4, Apr. 1998) adopted for reference. Methods can be used. In addition, the variance associated with noise and interference, ie the noise variance, can be estimated from the pilot channel of the CDMA system or the pilot tone of the OFDM system.

  FIG. 6 shows another embodiment of the operation configuration of the input metric normalizer according to the first embodiment of the present invention.

  FIG. 6 shows an example of realizing the normalizer of FIG. Normalization is realized by two shifters 630 and 640 and one adder 650. This normalization configuration can perform proper normalization while minimizing power consumption.

  In FIG. 6, the modulation order (mod-order) and the noise variance are input to a normalization index calculation unit 610, and a normalization index (norm_index) is calculated.

  Next, an example of the normalization method will be described in detail. The normalization index calculation unit 610 has temp_norm_index that is mapped to an estimated value that can be received from the noise variance estimator 265. Since division by noise variance must be reflected, temp_norm_index that is inversely proportional to the noise variance value must be selected.

となるように選択されなければならない。 For example, temp_norm_index is

Must be chosen to be

  However, “a” is a constant defined in relation to the operating range regarding noise and data channel values. Gain and noise estimates are expressed in dB scale taking log functions. In addition, [.] Means conversion to an integer closest to the input. In order to reflect the multiplication of the constant c, which is changed by the modulation order, the norm_index value is obtained using the following calculation.

norm_index = temp_norm_index, (QPSK)
norm_index = temp_norm_index-2, (16QAM)
norm_index = temp_norm_index-4, (64QAM)

  In the normalization table 620, the norm_index value is converted into a normalization gain value multiplied by a normalization coefficient as shown in Table 4. An LLR normalization adjustment of about 3 dB is possible in one step of Table 4. Only when finer adjustments are possible and the number of LLR bits has to be reduced, the normalization factor in Table 4 can be subdivided into finer steps and multiple adders can be used. .

  Thereafter, a value calculated by multiplying the norm_index value by a normalization coefficient is input to the shifters 630 and 640 and used to perform a shift operation on the metric Λ (n) from the demapper 270. The shifted value is added to the adder 650 to calculate the LLR. The normalized LLR is input to rounding / clipping unit 660. The desired range and the number of bits of LLR Λ ′ (n) are output from rounding / clipping unit 660.

  The normalization method described above is an example for realizing normalization in a channel decoder of a system using QPSK, 16QAM, and 64QAM. Of course, the present invention includes all possible ways of normalizing the SMG output LLR using the noise estimate and the modulation order.

<Second Embodiment>
Unlike the first embodiment, it may be difficult to calculate an accurate noise variance value in the communication system. When the channel code is close to the Shannon limit of the channel capacity without an error such as a turbo code and LDPC (Low Density Parity Check) code, a predetermined signal-to-noise ratio (SNR) With a noise threshold, error-free transmission with higher SNR is possible. That is, when the modulation order, the coding rate, and the frame size are set in a communication system using various modulations and coding rates, the SNR of the operation region that achieves the FER (Frame Error Rate) required by the system is obtained. Determined. Once this SNR is predetermined in the system, it can be used for LLR normalization.

  In this embodiment, once the modulation and coding rate are determined in the system, the desired value can be obtained through simulation of the system.

  FIG. 7 shows additive white Gaussian noise (hereinafter referred to as “AWGN”) of QPSK and 1/2 coding, QPSK and 3/4 coding, and 16QAM and 1/2 coding in an IEEE 802.16e system. ) Shows the FER performance of the channel.

  Referring to FIG. 7, when the RER required by the system is about 1%, a CINR (Carrier to Interference and Noise Ratio) region for static operation in the case of QPSK and 1/2 coding Is about 2-3 dB. Even if the LLR normalization is not optimal, since the CINR is sufficiently large in the SNR region of 2 to 3 dB or more, the FER is sufficiently low and does not affect the overall performance of the system. When CINR is lower than this, FER has a value close to “1” regardless of the LLR normalization.

  Therefore, the performance of LLR normalization does not substantially deteriorate even when a value predetermined in the system is used without using the actually measured noise variance. Basically, if the signal power obtained by automatic gain control is known, the SNR is defined and the noise variance value can also be detected. In the case of this QPSK and 1/2 coding, it is assumed that the basic operation region has 3 dB. Assuming that the signal power P is constant when the automatic gain loop is applied, the noise variance mapped to the signal power P and the 3 dB CINR has a relationship as shown in the following equation (11).

  That is, the noise variance is defined as shown in the following equation (12).

  When the noise variance calculated in this way is stored in the receiver in advance, the stored noise variance value is used even if the actual noise variance value is not calculated every time in the case of QPSK and 1/2 coding. Thus, optimal performance is obtained when performing LLR normalization.

  In the second embodiment of the present invention, CINR is fixed at about 3 dB based on AWGN. In an actual case, the QAM symbols constituting one frame undergo almost independent fading due to interleaving or the like. The 1% FER required by the system is realized with a higher CINR compared to AWGN. Therefore, the noise variance value stored in advance in the system should be set in consideration of FER. Although an example of QPSK and 1/2 coding has been described, it goes without saying that the same method can be applied even when other modulation orders and other coding rates are selected.

  In the case of the above configuration, the automatic gain controller (AGC) of the system operates normally and does not change greatly from an ideal value.

  FIG. 8 shows a configuration of a wireless communication transceiver to which the metric normalizer according to the second embodiment of the present invention is applied.

  Referring to FIG. 8, the transmitted binary data i (n) is encoded by a channel encoder 810 in the transmitter 800. The channel encoder 810 generates a series of binary code symbols c (n). The mapper 820 generates several code symbol blocks among the generated code symbols, maps them to one point on the signal point arrangement, and converts them into complex-valued modulation symbols x (n). This sequence x (n) is applied to the modulator 830. Modulator 830 generates a continuous-time wave by a symbol in a CDMA system or an orthogonal frequency division multiplexing (hereinafter referred to as “OFDM”) system, and transmits the generated continuous-time wave to receiver 850 via channel 840.

  In the receiver 850, a demodulator / channel estimator 860 performs a baseband demodulation and channel estimation process on the signal passing through the channel 840. The demodulator can be realized by a technique applied to the baseband. For example, the demodulator can be a CDMA rake receiver or an OFDM demodulator implemented with an IFFT process and a channel estimator.

  Channel estimation values and received symbols obtained after baseband modulation are output from demodulator / channel estimator 860 to demapper 870. The demapper 870 receives the received symbol y (n) and the channel estimation value c (n) modulated by QAM or PSK from the demodulator / channel estimator 860, and outputs a metric for each bit through demapping. The demapper 870 can obtain the metric using various algorithms. The demapping algorithm described with reference to FIG. 2 can be used.

  In the IEEE802.16 system receiver 850, the convolutional turbo decoder, which is the channel decoder 880, receives and decodes the soft metric corresponding to the reliability information of each bit, so that the received signal distorted before the channel decoder 880 is received. The process of calculating the soft metric from Such a process is performed by a demapper 870 in the receiver 850.

  Therefore, the LLR normalizer 875 uses the metric Λ (n) output from the demapper 870 and the above-described modulation and coding rate AMC (Adaptive Modulation and Coding) information from the controller 865 to determine a predetermined noise. Receive the variance value and perform normalization. The channel decoder 880 receives the normalized value Λ ′ (n) and outputs i (n).

  FIG. 9 shows an operation configuration of the input metric normalizer according to the second embodiment of the present invention.

  In FIG. 9, a predetermined noise variance table is used according to AMC information of modulation and coding rate. Referring to FIG. 9, the noise variance table 910 of the normalizer 875 stores “c” values according to QPSK, 16QAM, and 64QAM modulation schemes. Upon receiving AMC information such as the modulation order, coding rate, and frame size, the normalizer 875 can set the reference “c” value according to the noise value and modulation order predetermined by the AMC information.

を受信する。 In the normalizer 875, the multiplier 930 is calculated by converting the noise variance value and the reference “c” value using the conversion table 920 reflecting the division.

Receive.

を乗算すると、ＬＬＲが正規化される。 Multiplier 930 takes the metric Λ (n) from demapper 870

Is multiplied to normalize the LLR.

  After LLR normalization, rounding / clipping unit 940 inputs LLRΛ '(n) having the desired range and number of bits to the decoder.

  FIG. 10 shows another operation configuration of the input metric normalizer according to the second embodiment of the present invention.

  In FIG. 10, the normalizer predetermines a set of normalization coefficients instead of the conversion table 920 in FIG. 9, receives only the normalization index, and sets the normalization coefficients.

  The normalization index calculation unit 1010 receives information such as the modulation order, coding rate, and frame size, determines a normalization index that can reflect the “c” value and the noise variance value, and outputs the normalization index to the normalization table 1020. The normalized index value can be determined using a predetermined table.

  Upon receiving a defined normalization index, a set of possible normalization counts can be predetermined in the normalization table 1020. When this normalization index is received, a normalization coefficient is set. Multiplier 1030 multiplies metric Λ (n) from demapper 870 by a normalization factor, and the LLR is normalized. After LLR normalization, rounding / clipping unit 1040 inputs LLR Λ ′ (n) having the desired range and number of bits to the decoder.

  FIG. 11 shows still another operation configuration of the input metric normalizer according to the second embodiment of the present invention.

  FIG. 11 shows a configuration obtained by simplifying the configuration of FIG. Normalization is realized by two shifters 1130 and 1140 and one adder 1150. This normalization configuration can perform proper normalization while minimizing power consumption.

  Information on the modulation and coding rate (or FEC code type) is input to the normalization index calculator 1110 to calculate a normalization index (norm_index). In the normalization table 1120, the calculated norm_index value is converted into a normalization gain value multiplied by a normalization coefficient as shown in <Table 4>. An LLR normalization adjustment of about 3 dB is possible in one step of Table 4. Only when finer adjustments are possible and the number of LLR bits has to be reduced, the normalization factor in Table 4 can be subdivided into finer steps and multiple adders can be used. .

  Thereafter, a value calculated by multiplying the norm_index value by the normalization coefficient is input to the shifters 1130 and 1140 and used to perform a shift operation on the metric Λ (n) from the demapper 870. The shifted value is added to the adder 1150 to calculate the LLR. The normalized LLR is input to rounding / clipping unit 1160. LLR Λ ′ (n) having the desired range and number of bits is output.

  An example of a normalization method using <Table 4> is as follows.

  This example is used in an IEEE 802.16e system where accurate noise variance measurement is difficult. At this time, the fact that a code having the same modulation method and the same FER generally achieves 1% FER with almost the same SNR is used. For each modulation scheme, the SNR for 1% FER is calculated and a norm_index reflecting the hypothetical noise index is provided. The IEEE 802.16e system has the following modulation code codes for general data bursts to which convolutional turbo codes are applied. In this embodiment, norm_index_basic is applied for the actual norm_index. Table 5 below shows an example of normalization of IEEE 802.16e when the configuration of FIG. 8 is realized.

  In Table 5, norm_index_basic is used to reflect burst boosting or zone boosting as defined by the IEEE 802.16e system. The IEEE 802.16e system supports -12 dB to 9 dB boosting with a burst power control concept. A region boosting of 4.77 dB is supported when the frequency reuse factor is 1/3. In this case, since the LLR value is influenced by boosting, this value can be compensated for and the effective operation area of the LLR can be reduced. For example, norm_index can be calculated as shown in the following equation (13).

  In equation (13), the boosting unit is dB, and [a] means rounding off. Furthermore, norm_index has a value in the given range [0 24]. Using this method, a more general LLR normalization is possible.

  The implementation method of the present invention described above is an example of a method for normalizing an LLR corresponding to an input metric for a decoder using a normalization coefficient and AMC information. The present invention includes all implementations that apply normalization to the output of a soft output generator functioning as a demapper and perform normalization using AMC information.

  12 and 13 perform floating point operations when using a 6-bit or 8-bit soft input metric (denoted as “fading, 6 bits” or “AWGN, 6 bits” and “fading, 8 bits”). The performance of the convolutional turbo code defined in the IEEE 802.16e system in the case of (indicated by “fading, Ft” and “AWGN, Ft”) is shown. The turbo decoder uses the max-log-MAP method. It can be seen that embodiments of the present invention using 6-bit or 8-bit normalized LLR have little performance difference from the floating point operation case.

  As mentioned above, in the detailed description of the present invention, specific embodiments have been described. However, it is understood by those skilled in the art that various modifications can be made without departing from the scope of the claims. Is clear. Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be determined based on the description of the scope of claims and equivalents thereof.

It is a figure which shows the structure of the transmitter / receiver in the conventional radio | wireless communications system. It is a figure which shows the structure of the transmitter / receiver to which the input metric normalizer by the 1st Embodiment of this invention is applied. It is a figure which shows the signal point arrangement | positioning and mapping of QPSK. It is a figure which shows the signal point arrangement | positioning and mapping of 16-QAM. It is a figure which shows the signal point arrangement | positioning and mapping of 64-QAM. It is a figure which shows an example which calculates a soft metric. It is a figure which shows the operation | movement structure of the input metric normalizer by the 1st Embodiment of this invention. It is a figure which shows the other operation | movement structure of the input metric normalizer by the 1st Embodiment of this invention. It is a figure which shows the FER performance of an AWGN channel. It is a figure which shows the structure of the transmitter / receiver to which the input metric normalizer by the 2nd Embodiment of this invention is applied. It is a figure which shows the operation | movement structure of the input metric normalizer by the 2nd Embodiment of this invention. It is a figure which shows the other operation | movement structure of the input metric normalizer by the 2nd Embodiment of this invention. It is a figure which shows the further another operation | movement structure of the input metric normalizer by the 2nd Embodiment of this invention. It is a figure which shows the performance of the convolution turbo decoder with respect to the 6-bit input metric applied to the metric normalizer by the 1st and 2nd embodiment of this invention. FIG. 7 is a diagram illustrating the performance of a convolutional turbo decoder applied to a metric normalizer according to the first and second embodiments of the present invention and for a 6-bit input metric.

Explanation of symbols

100 transmitter 110 channel encoder 120 mapper 130 modulator 140 channel 150 receiver 160 demodulator / channel estimator 170 demapper 180 channel decoder 200 transmitter 210 channel encoder 220 mapper 230 modulator 240 channel 250 receiver 260 demodulator / channel estimation 265 Noise variance estimator 270 Demapper 275 LLR normalizer 280 Channel decoder 510 Conversion table 520 Multiplier 530 Rounding / clipping unit 610 Normalization index calculator 620 Normalization table 630, 640 Shifter 650 Adder 660 Rounding / clipping 800 transmitter 810 channel encoder 820 mapper 830 modulator 840 channel 850 860 demodulator / channel estimator 865 controller 870 demapper 875 LLR normalizer 880 channel decoder 910 noise variance table 920 conversion table 930 multiplier 940 rounding / clipping unit 1010 normalization index calculator 1020 normalization table 1030 multiplication Unit 1040 rounding / clipping unit 1110 normalization index calculation unit 1120 normalization table 1130, 1140 shifter 1150 adder 1160 rounding / clipping unit

Claims (28)

A device for normalizing a soft metric input to a channel decoder in a wireless communication system,
A demapper that generates a soft metric per bit for the received modulation symbols ;
The soft metric is received, a normalized LLR (Log Likelihood Ratio) is obtained using a modulation order and a noise variance value of the received modulation symbol, and the normalized LLR is converted into a predetermined number of bits. A normalizer that outputs an input LLR of the channel decoder;
The apparatus characterized by including. The demapper is a constant value determined by an in-phase component (X _k ) and a quadrature component (Y _k ) of a received modulation symbol (R _k ), a channel fading coefficient (g _k ), and a modulation order of the received modulation symbol. The apparatus of claim 1, wherein the soft metric is generated using at least one of (c). The normalizer uses the soft metric

To perform normalization, where:

2. The apparatus of claim 1, wherein is the noise variance value determined from a channel estimate of received modulation symbols. The apparatus of claim 3, wherein the soft metric is generated according to the following table.

Here, I _k , Q _k , and a _k are | g _k | ² X _k , | g _k | ² Y _k , and | g _k | ² c, respectively. The normalizer is
A conversion table for receiving the calculated said noise variance value from the channel estimation values of the received modulation symbols, and outputs the ratio of the constant value for the noise variance value,
A multiplier that multiplies the soft metric output from the demapper by a ratio of a constant value to the noise variance value to output the normalized LLR;
A rounding / clipping unit that converts the normalized LLR into a predetermined number of bits and outputs an input LLR of the channel decoder;
The apparatus of claim 1, comprising: The normalizer is
Receiving the noise variance value calculated from the channel estimation value of the received modulation symbol, selecting a temporary normalization index corresponding to division by the noise variance value, and modulating the modulation from the selected temporary normalization index A normalization index calculation unit that generates a normalization index by subtracting a predetermined value according to the order;
A normalization table that multiplies the normalization index by a normalization factor to convert it into a normalization gain value;
A shifter that shifts in-phase and quadrature components of the received modulation symbol by the normalized gain value;
An adder that adds the shifted values and outputs a normalized LLR;
A rounding / clipping unit that converts the normalized LLR into a predetermined number of bits and outputs an input LLR of the channel decoder;
The apparatus according to claim 2, comprising: A device for normalizing a soft metric input to a channel decoder in a wireless communication system,
A demapper that generates a soft metric per bit for the received modulation symbols ;
The soft metric is received, the soft metric is multiplied by a normalization coefficient obtained from adaptive modulation and coding (AMC) information to obtain a normalized LLR (Log Likelihood Ratio), and the normalized LLR A normalizer that converts the channel number into a predetermined number of bits and outputs the input LLR of the channel decoder;
The apparatus characterized by including. The demapper is a constant value determined by an in-phase component (X _k ) and a quadrature component (Y _k ) of a received modulation symbol (R _k ), a channel fading coefficient (g _k ), and a modulation order of the received modulation symbol. 8. The apparatus of claim 7, wherein the soft metric is generated using at least one of (c).   The normalizer performs normalization by multiplying the soft metric by a normalization coefficient, wherein the normalization coefficient is set by a normalization index calculated from AMC information. 8. The apparatus according to 7. The apparatus of claim 9, wherein the soft metric is generated according to the following table:

Here, I _k and a _k are | g _k | ² X _k and | g _k | ² c , respectively. The normalizer is
A noise dispersion table for setting and outputting a noise dispersion value and a constant value according to at least one information inputted from a reception controller storing the AMC and boosting information;
A conversion table that receives the noise variance value and the constant value and outputs a ratio of the constant value to the noise variance value;
A multiplier for multiplying the soft metric output from the demapper by a ratio of the constant value to a noise variance value to output the normalized LLR;
A rounding / clipping unit that outputs the input LLR of the channel decoder by converting the normalized LLR into a predetermined number of bits ;
The device of claim 7, comprising: The normalizer is
A normalization index calculator for generating a normalization index in which a constant value and a noise variance value are reflected by at least one information input from a reception controller storing the AMC information and boosting information;
A normalization table for setting a normalization coefficient mapped to the normalization index;
A multiplier for multiplying the soft metric output from the demapper by the set normalization coefficient and outputting the normalized LLR;
A rounding / clipping unit that outputs the input LLR of the channel decoder by converting the normalized LLR into a predetermined number of bits ;
The device of claim 7, comprising: The normalizer is
A normalization index calculator for generating a normalization index in which a constant value and a noise variance value are reflected by at least one information input from a reception controller storing the AMC information and boosting information;
A normalization table that multiplies the normalization index by a normalization factor to convert it into a normalization gain value;
A shifter that shifts in-phase and quadrature components of the received modulation symbol by the normalized gain value;
An adder that adds the shifted values and outputs a normalized LLR;
A rounding / clipping unit that outputs the input LLR of the channel decoder by converting the normalized LLR into a predetermined number of bits ;
The device of claim 7, comprising:   The apparatus of claim 7, wherein the AMC information includes at least one of a modulation order, a coding rate, and a frame size. A method for normalizing a soft metric input to a channel decoder in a wireless communication system, comprising:
Generating a bit-by-bit soft metric for the received modulation symbols ;
Receiving the soft metric, obtaining a normalized LLR (Log Likelihood Ratio) using a modulation order and a noise variance value of the received modulation symbol ;
Converting the normalized LLR into a predetermined number of bits ;
Outputting an input LLR of the channel decoder;
A method characterized by comprising: Generating the soft metric comprises:
A constant value (c) determined by the in-phase component (X _k ) and quadrature component (Y _k ) of the received modulation symbol (R _k ), the channel fading coefficient (g _k ), and the modulation order of the received modulation symbol. The method of claim 15, wherein the soft metric is generated using at least one of them. The step of obtaining the normalized LLR includes:
To the soft metric

To perform normalization, where:

16. The method of claim 15, wherein is the noise variance value determined from a channel estimate of received modulation symbols. The method of claim 17, wherein generating the soft metric generates the soft metric according to the following table.

Here, I _k and a _k are | g _k | ² X _k and | g _k | ² c , respectively. The step of obtaining the normalized LLR includes:
Receiving the noise variance value calculated from a channel estimate of the received modulation symbol and outputting a ratio of the noise variance value to the constant value;
Multiplying the soft metric by a ratio of a noise variance value to the constant value to output the normalized LLR ;
The method of claim 15, characterized in that it comprises a. Outputting the input LLR of the channel decoder comprises:
Receiving the noise variance value calculated from a channel estimate of the received modulation symbol and selecting a temporary normalization index corresponding to division by the noise variance value;
Subtracting a modulation order value from the selected temporary normalized index to generate a normalized index;
Converting the normalized index to a normalized gain value by multiplying by a normalization factor;
Shifting in-phase and quadrature components of the received modulation symbol by the normalized gain value;
Adding the shifted values to output a normalized LLR;
Converting the normalized LLR into a predetermined number of bits and outputting as an input LLR of the channel decoder;
The method of claim 15, comprising: A method for normalizing a soft metric input to a channel decoder in a wireless communication system, comprising:
Generating a bit-by-bit soft metric for the received modulation symbols ;
Receiving the soft metric and determining a normalized LLR by multiplying the normalization factor determined by adaptive modulation and coding (AMC) information with the soft metric;
Converting the normalized LLR into a predetermined number of bits and outputting an input LLR of a channel decoder;
A method characterized by comprising: Generating the soft metric comprises:
A constant value (c) determined by the in-phase component (X _k ) and quadrature component (Y _k ) of the received modulation symbol (R _k ), the channel fading coefficient (g _k ), and the modulation order of the received modulation symbol. The method of claim 21, wherein the soft metric is generated using at least one of them. The step of obtaining the normalized LLR includes:
Normalization is performed by multiplying the soft metric by a normalization coefficient, wherein the normalization coefficient calculates a normalization index according to AMC information, and obtains the normalized LLR by the calculated normalization index. The method according to claim 21. The method of claim 23, wherein generating the soft metric generates the soft metric according to the following table.

Here, I _k and a _k are | g _k | ² X _k and | g _k | ² c , respectively. Outputting the input LLR of the channel decoder comprises:
Setting a noise variance value and a constant value according to at least one information input from a reception controller storing the AMC and boosting information;
Receiving the noise variance value and the constant value and outputting a ratio of the constant value to the noise variance value;
Multiplying the soft metric by a ratio of a constant value to the noise variance value to output the normalized LLR;
Converting the normalized LLR into a predetermined number of bits and outputting as an input LLR of the channel decoder;
The method of claim 21, comprising: Outputting the input LLR of the channel decoder comprises:
Generating a normalization index in which a constant value and a noise variance value are reflected by at least one information input from a reception controller storing the AMC and boosting information;
Setting a normalization factor mapped to the normalization index;
Multiplying the soft metric by the set normalization factor and outputting the normalized LLR;
Converting the normalized LLR into a predetermined number of bits and outputting an input LLR of the channel decoder;
The method of claim 21, comprising: Outputting the input LLR of the channel decoder comprises:
Generating a normalization index in which a constant value and a noise variance value are reflected by at least one information input from a reception controller storing the AMC and boosting information;
Converting the normalized index to a normalized gain value by multiplying by a normalization factor;
Shifting in-phase and quadrature components of the received modulation symbol by the normalized gain value;
Adding the shifted values to output a normalized LLR;
Converting the normalized LLR into a predetermined number of bits and outputting an input LLR of the channel decoder;
The method of claim 21, comprising:   The method of claim 21, wherein the AMC information includes at least one of a modulation order, a coding rate, and a frame size.

Images