JP5831326B2 - Diversity receiver - Google Patents

Description

  The present invention relates to a diversity receiver, and can be applied to a diversity receiver that employs an Orthogonal Frequency Division Multiplexing (OFDM) system, for example.

  FIG. 2 is an internal configuration diagram of a receiver that employs an orthogonal frequency division multiplexing (OFDM) scheme.

  Here, the orthogonal frequency division multiplexing system is a system that digitally modulates and multiplexes a plurality of orthogonal carriers that do not interfere with each other, and is a form of multicarrier modulation system (Non-Patent Document 1). reference). The multi-carrier modulation method is a method for performing modulation by dispersing 1-channel data over a plurality of carrier waves.

  By adopting a multi-carrier modulation method typified by the OFDM method, transmission data can be distributed over multiple carriers, so even if multi-path frequency selective fading occurs, the probability that all data will be lost is reduced. Can do.

  The transmitter uses a multiplexer to divide transmission data into L data series, and then synthesizes and transmits digital modulation signals obtained by modulating each data series with a carrier having a different frequency.

  The receiver 9 includes demodulation units 11 and 21 illustrated in FIG. 3, and the modulated signal in the carrier frequency band captured by the antennas 10 and 20 is received by the band-pass filter 101 of the demodulation units 11 and 21. A digital modulation signal corresponding to the carrier wave is extracted and demodulated.

  The demodulated individual data series is given to DFT (Discrete Fourier Transform) units 13 and 23, and the discrete Fourier transform is performed by the DFT units 13 and 23 to return to the original data series modulated to each carrier frequency. It is.

  The original data series is given to diversity combining sections 30 to 32, and diversity combining sections 30 to 32 perform diversity combining and output one signal to P / S (parallel serial) conversion section 33.

  Here, the diversity technique is a technique for improving reception characteristics (see Non-Patent Document 2). The diversity technique uses the characteristic that multipath fading occurs as the terminal moves and the antenna undergoes completely different fading fluctuations when the antenna of the terminal moves along a different route.

  One type of diversity is space diversity, and the receiver 9 of FIG. 2 illustrates a case where diversity receivers 30 to 32 that have two antennas and perform space diversity combining are provided.

  Examples of diversity synthesis methods include selective synthesis and maximum ratio synthesis.

  Selective combining is a method of selecting a branch having the highest received signal level (received power) among diversity branches. Any one signal is used for demodulation, and the remaining signals are discarded.

  On the other hand, maximum ratio combining is a method in which signals of all branches are made in-phase and then combined with appropriate weights. Maximum ratio combining further improves reception characteristics. When the weighting coefficient of each branch is set equal to the reception envelope level, the combined SNR is maximized.

  Here, a specific calculation method in the conventional diversity combining units 30 to 32 will be described.

In the following, it is assumed that the received signal (demodulated symbol) of the u th carrier at time t of the k th branch is x _k (t, u). Similarly, the transmission signal (symbol before modulation by the transmitter) is d _k (t, u), the propagation path (channel) characteristic is H _k (t, u), and the additional noise is Z _k (t, u). .

  The values of these signals are expressed as complex numbers (complex envelope, equivalent low-frequency signal) for convenience, the absolute value represents the amplitude of the sine wave, and the declination represents the phase of the sine wave.

At this time, the received signal x _k (t, u) is expressed by the following equation (1).

x _k (t, u) = H _k (t, u) d _k (t, u) + Z _k (t, u) (1)
Here, a case where the number of branches is 2 is considered (that is, k is k = 1 or 2).

In the case of selective synthesis, the signal y (t, u) after the synthesis of the u-th carrier at time t is expressed by the following equations (2) and (3).

When | x ₁ (t, u) | ² = | x ₂ (t, u) | ² , y (t, u) = x ₁ (t, u) ₂ (t, u) may be used, but since it is necessary to uniquely determine it, y (t, u) = x ₁ (t, u) is assumed here.

On the other hand, in the case of maximum ratio combining, assuming that the propagation path characteristics H _k (t, u) are known on the receiver side, the combined signal y (t, u) is expressed by the following equations (4) to (6 )become that way. Note that the propagation path characteristic H _k (t, u) is actually a value estimated on the receiver side, but is a content independent of the content of the present invention. Shall not.

y (t, u) = w ₁ (t, u) x ₁ (t, u) + w ₂ (t, u) x ₂ (t, u), (4)
w ₁ (t, u) = H ₁ ^* (t, u) / (| H ₁ (t, u) | ² + | H ₂ (t, u) | ² ), (5)
w ₂ (t, u) = H ₂ ^* (t, u) / (| H ₁ (t, u) | ² + | H ₂ (t, u) | ² ) (6)
Here, w1 (t, u) and w2 (t, u) are weighting coefficients, and H _k ^* (t, u) is a conjugate complex number of H _k (t, u). From the above equation, w ₁ (t, u), w ₂ (t, u) is the same as the phase at the time of transmission (w ₁ (t, u), w ₂ (t, u)) Since the coefficients H ₁ ^* (t, u) and H ₂ ^* (t, u) are complex conjugates of the propagation path characteristics H ₁ (t, u) and H ₂ (t, u), their declination is The values (H ₁ ^* (t, u), H ₂ ^* (t, u)) proportional to the amplitude of the received signal are the same as the propagation path characteristics H. ₁ (t, u) and H ₂ (t, u) are complex conjugate numbers, and the absolute value is the same as the absolute value of the propagation path characteristic).

  From the above, when the number of branches is M (k = 1, 2,..., M), the following formulas (7) and (8) are obtained in the case of selective synthesis.

y (t, u) = x _i (t, u) (7)
However, i is a value satisfying the following.

| x _i (t, u) | ² = max (| x ₁ (t, u) | ² , | x ₂ (t, u) | ² ,…, | x _M (t, u) | ² )・ (8)
_{Here, max (a 1, a 2} , ..., a n) is a _1, a _2, ..., represent the maximum value of a _n.

On the other hand, in the case of maximum ratio synthesis, the following equations (9) and (10) are obtained.

  In FIG. 3, the AGC control unit 106 determines the gain of the AGC amplifier 102 so that the level of the input / output signal of each unit in the receiver 9 is maintained within a predetermined range.

  That is, each circuit element in the receiver 9 has an upper limit value and a lower limit value of a voltage that can be output. If the signal level is too larger than the upper limit value or smaller than the lower limit value, the amplitude of the signal is not proportional to the voltage of each circuit element, and the signal waveform cannot be reproduced correctly. Therefore, since the signal level needs to be maintained at an appropriate value between the upper limit value and the lower limit value, the AGC control unit 106 determines the gain of the AGC amplifier 102.

  That is, the level of the baseband signal from the LPF 105 is observed, and the gain of the AGC amplifier 102 is determined so that this value converges to a predetermined target value. For the observation of the level of the baseband signal, for example, a method of calculating an integral value within a predetermined time can be used to smooth the level of the signal.

  Here, an example of a method for determining the gain of the AGC amplifier 102 by the AGC control unit 106 will be described below.

Assuming that the signal from _LPF 105 to AGC control unit 106 at time t of the k-th branch is x _{LPF, k} (t) and its power is P _{LPF, k} (t), AGC control unit 106 The integral value Q _{LPF, k, r} is calculated as follows.

  Here, T is the time from the start to the end of integration (cycle for performing integration).

If the signal and power of the u th carrier at time t of the k th branch are x _{sub, LPF, k} (t, u), P _{sub, LPF, k} (t, u), respectively, x _{LPF, k} The relationship between (t), P _{LPF, k} (t), and x _{sub, LPF, k} (t, u) and P _{sub, LPF, k} (t, u) is expressed as follows.

On the other hand, the power (input power at the antenna 10) of the input signal to the antenna 10 at time t of the k-th branch is P _{ANT, k} (t), the gain of the AGC amplifier 102 is G _k (t), and the AGC amplifier 102 If the gain of the circuit from the antenna 10 to the AGC control unit 106 other than the above is F _k (t), P _{LPF, k} (t) has the following relationship.

P _{LPF, k} (t) ＝ F _k (t) G _k (t) P _{ANT, k} (t) (26)
Here, the additional noise is assumed to be sufficiently smaller than the desired signal, and is ignored.

If the power of the input signal to the antenna 10 of the u th carrier at the time t of the k th branch (the received power of the u th carrier at the antenna 10) is P _{sub, ANT, k} (t, u). , P _{ANT, k} (t) has the following relationship:

From the equations (21) and (26), the following is derived.

Here, it is assumed that F _k (t) and G _k (t) are constant in the integration interval. That is, equations (29) and (30) are assumed to hold.

F _k (t) = F _{k, r} ((r-1) T ≦ t ≦ rT), (29)
G _k (t) = G _{k, r} ((r-1) T ≦ t ≦ rT), ... (30)
Here, F _{k, r} and G _{k, r} are constants determined by k, r. The following is derived from the equations (28) to (30).

Q _{LPF, k, r} = F _{k, r} G _{k, r} P _{ANT, k, r} T (31)
Here, P _{ANT, k, r} is expressed by Expression (32), and the average power of the signal input to the antenna 10 within the r-th power integration interval of the k th branch (the average received power at the antenna 10). ).

Note that the average power (the average received power of the uth carrier at the antenna 10) input to the antenna 10 of the uth carrier within the rth power integration interval of the kth branch is P _{sub, ANT, If k, r} (u), then P _{sub, ANT, k} (t, u) has the following relationship:

Here, the target value of the power integration of the signal observed by the AGC control unit 106 in the k-th branch is Q ^ _{LPF, k} . In Expression (31), when Q _{LPF, k, r} → Q ^ _{LPF, k} , G _{k, r} → G ^ _{k, r} , the following expression is obtained.

Q ^ _{LPF, k} = F _{k, r} G ^ _{k, r} P _{ANT, k, r} T (34)
Here, since G ^ _{k, r} is a target value of the gain of the AGC amplifier 102, the AGC control unit 106 determines this value as the gain of the AGC amplifier 102 for the next time. That is, if G ^ _{k, r} → _{Gk, r + 1} , the following is derived from equations (31) and (34).

G _{k, r + 1} = E _{LPF, k, r} G _{k, r} , (35)
E _{LPF, k, r} = Q ^ _{LPF, k} / Q _{LPF, k, r} , (36)
Q _{LPF, k, r} = (R _{k, r} T) G _{k, r} , (37)
R _{k, r} = F _{k, r} P _{ANT, k, r} (38)
E _{LPF, k, r} is the ratio of the power integral value Q _{LPF, k, r to} the target value Q ^ _{LPF, k} , R _{k, r} is the disturbance (the factor that disturbs the control of Q _{LPF, k, r} ) Represent. Therefore, the AGC control unit 106 can be configured as a feedback circuit that observes the integrated value of the power of the signal from the LPF 105 and determines the gain of the AGC amplifier 102 so that this value converges to the target value.

  As described above, when the level of the signal input to the antenna 10, that is, the level of the signal input to the demodulation unit 11 is small (that is, the reception power at the antenna 10 is low), the AGC control unit 106 When the level of the signal input to the demodulation unit 11 is large (that is, the reception power at the antenna 10 is high) so as to increase the gain of 102, the value is set so as to decrease the gain of the AGC amplifier 102. decide.

  Here, the signal for calculating the integral value of power (input signal to the AGC control unit 106) is the output signal of the LPF 105, but the output signal of the BPF 101, the output signal of the AGC amplifier 102, or the output signal of the down converter 103 Alternatively, the same method can be used even when the output signal of the LPF 107 or the output signal of the A / D converter 108 is used. Since any output signal does not impair the generality of the present invention, here, only the description in the case where the signal (input signal to the AGC control unit 106) for which the integral value of power is calculated is used as the output signal of the LPF 105 is described. Stay on.

  By the way, when diversity combining is performed by the diversity combining units 30 to 32, it is necessary to correctly reflect the magnitude of the signal level (received power at the antenna of each branch) at the antenna of each branch.

  That is, when the number of branches is 2, the level of the u-th carrier signal input from the DFT unit 13 to the diversity combining units 30 to 32 and the u-th carrier wave input from the DFT unit 23 to the diversity combining units 30 to 32 The signal level of the u th carrier wave input to the antenna 10 (the received power of the u th carrier wave at the antenna 10) and the u th carrier wave signal input to the antenna 20 It is required to be equal to the ratio to the level (the received power of the u th carrier at the antenna 20).

  However, the AGC control units 106 (the AGC control unit 106 in the demodulation unit 11 and the AGC control unit 106 in the demodulation unit 21) of both branches determine the gain of the AGC amplifier 102 so that each has a predetermined signal level. The values may be different from each other.

  The predetermined signal levels targeted by the AGC control units 106 of both branches are normally equal to each other because there is no difference in circuit configuration between both branches.

  Therefore, even when the reception power of the u-th carrier at the antennas of both branches is different, the signal level obtained by adding all the carriers input from the DFT unit 13 to the diversity combining units 30 to 32 and the DFT unit 23 diversity are used. The level of the signal obtained by adding all the carrier waves input to the combining units 30 to 32 becomes equal.

  Here, assuming that the signal levels of all the carrier waves are equal, the level of the u-th carrier signal input from the DFT unit 13 to the diversity combining units 30 to 32 and the DFT unit 23 to the diversity combining units 30 to 32 are assumed. The level of the input u-th carrier signal becomes equal.

  Therefore, the diversity combining sections 30 to 32 do not correctly reflect the magnitude of the received power at the antennas of both branches, and the diversity combining is not performed correctly in this state, so that the reception characteristics are deteriorated.

As measures against the above contents, as a means for the diversity combining units 30 to 32 to correctly reflect the magnitude of the signal level (received power at the antenna of each branch) at the antenna of each branch, The levels of signals (signals from the DFT unit 13 to the diversity combining units 30 to 32 and signals from the DFT unit 23 to the diversity combining units 30 to 32) input from the branches to the diversity combining units 30 to 32 are set in each branch. What is necessary is just to correct | amend with the value of the gain of AGC amplifier 102 (AGC amplifier 102 in the demodulation part 11 and AGC amplifier 102 in the demodulation part 21). For example, the signals input to the diversity combining units 30 to 32 may be divided by the square root of the gain value of the AGC amplifier 102. As a result, the magnitude of the received power at the antenna of each branch is correctly reflected, and diversity combining by the diversity combining units 30 to 32 can be performed correctly.

The above contents will be described using mathematical expressions.

From the equations (26), (29), (30), the following is derived.

P _{LPF, k} (t) ＝ F _{k, r} G _{k, r} P _{ANT, k} (t) (41)
From the equations (24), (27), (41), the following is derived.

  Since all carriers are orthogonal, the power of one carrier does not depend on the power of other carriers. Therefore, the following is derived from the equation (42).

P _{sub, LPF, k} (t, u) = F _{k, r} G _{k, r} P _{sub, ANT, k} (t, u) (43)
Here, since the circuit configurations of the first branch and the second branch are the same, it is assumed that the circuit characteristics F _{k, r} from the antenna 10 other than the AGC amplifier 102 to the AGC control unit 106 are equal in both branches. To do. That is, equation (44) is assumed to hold.

F _{1, r} = F _{2, r} (44)
From the equations (43) and (44), the following is derived.

P _{sub, LPF, 1} (t, u) / {G _{1, r} P _{sub, ANT, 1} (t, u)}
= P _{sub, LPF, 2} (t, u) / {G _{2, r} P _{sub, ANT, 2} (t, u)} (45)
Transforming equation (45) leads to:

P ^ _{sub, LPF, 1} (t, u) / P ^ _{sub, LPF, 2} (t, u) = P _{sub, ANT, 1} (t, u) / P _{sub, ANT, 2} (t, u), ... (46)
P ^ _{sub, LPF, 1} (t, u) = c ² P _{sub, LPF, 1} (t, u) / G _{1, r} , (47)
P ^ _{sub, LPF, 2} (t, u) = c ² P _{sub, LPF, 2} (t, u) / G _{2, r} (48)
Here, c is an arbitrary constant that is not 0. As described above, the signal level P _{sub, LPF, 1} (t, u) of the u-th carrier inputted from the DFT unit 13 to the diversity combining units 30 to 32 and the DFT unit 23 to the diversity combining units 30 to 32 The ratio of the input u-th carrier signal level P _{sub, LPF, 2} (t, u) is the level of the u-th carrier signal input to antenna 10 (received power at antenna 10) P _{sub. , ANT, 1} (t, u) and the level of the u-th carrier signal input to antenna 20 (received power at antenna 20) P _{sub, ANT, 2} (t, u) must be equal to the ratio There is.

However, P _{sub, LPF, 1} (t, u) → P ^ _{sub, LPF, 1} (t, u), P _{sub, LPF, 2} (t, u) → P from equations (46)-(48) If you replace it with ^ _{sub, LPF, 2} (t, u) (by correcting P _{sub, LPF, k} (t, u) to P ^ _{sub, LPF, k} (t, u)) Is possible.

From Expressions (25), (47), and (48), the signal x _{sub, LPF, 1} (t, u) input from the DFT unit 13 to the diversity combining units 30 to 32 and the diversity combining unit 30 to 30 from the DFT unit 23 What is necessary is just to correct | amend the signal _{xsub, LPF, 2} (t, u) input into 32 as follows.

  As described above, it is derived that the amplitude of the signal input to the diversity combining units 30 to 32 may be divided by the square root of the gain value of the AGC amplifier 102.

That is, the input signal of the u-th carrier wave from the demodulating unit DFT units 13 and 23 to the diversity combining units 30 to 32 at the time t of the k-th branch is x _{sub, LPF, k} (t, u), and selective combining is performed. In this case, x _k (t, u) in equations (7) and (8) is _replaced with x ^ _{sub, LPF, k} (t, u) (where k = i and k = 1,2, ..., M) These equations are applied instead of y, and y (t) is used as an output signal of diversity combining sections 30-32. Also, in the case of maximum ratio combining, replace x _k (t, u) in equations (9) and (10) with x ^ _{sub, LPF, k} (t, u) (where k = i). And y (t, u) is used as an output signal of diversity combining sections 30-32.

Japanese Patent Laying-Open No. 2005-026940 Japanese Patent Laid-Open No. 2001-1000079

Itami Makoto, “Intuitive OFDM Technology”, Ohm Co., Ltd., published on November 15, 2005, P.25-P.44 Seiichi Sampei, “From Digital Wireless Transmission Technology to System Design”, Pearson Education, September 1, 2002, P.146-P.154

  However, the conventional receiver needs to perform A / D conversion of the received analog signal to a digital signal for digital signal processing. Quantization noise can occur when performing A / D conversion. The quantization noise level is different for each diversity branch, and is included in the additional noise level.

  In a receiver employing the OFDM scheme, even if the thermal noise level is the same for all branches and all carriers, the quantization noise level is not always the same. May not be able to operate properly. Hereinafter, this problem will be described.

  It has been stated that selective combining is a method of selecting a branch having the highest received signal level among diversity branches of the same carrier wave. However, strictly speaking, this is a method of selecting a branch having the highest SNR (signal to noise ratio) among diversity branches of the same carrier wave. However, if the condition that the added noise levels of all diversity branches of the same carrier wave are equal, this is equivalent to the method of selecting the branch with the highest received signal level.

The above contents will be described using mathematical expressions. First, the power of the received signal of the u-th carrier at time t of the k-th branch is P _k (t, u), the power of the desired signal among the received signals is S _k (t, u), and the power of the additional noise Is N _k (t, u), P _k (t, u) is expressed by the following equation.

P _k (t, u) = S _k (t, u) + N _k (t, u) (201)
At this time, when the SNR of the u-th carrier at time t of the k-th branch is SNR _k (t, u), it is expressed as follows.

SNR _k (t, u) = S _k (t, u) / N _k (t, u) (202)
From the equations (201) and (202), the following is derived.

SNR _k (t, u) = {P _k (t, u) −N _k (t, u)} / N _k (t, u)
= P _k (t, u) / N _k (t, u) −1 (203)
Here, the SNR of the α-th branch is SNR _α (t, u), and the SNR of the β-th branch is SNR _β (t, u). It is assumed that SNR _α (t, u) of the α-th branch is larger than SNR _β (t, u) of the β-th branch, and that the additional noise power of both branches is equal. That is, equations (204) and (205) are assumed to hold.

SNR _α (t, u)> SNR _β (t, u), ... (204)
N _α (t, u) = N _β (t, u) (205)
At this time, the following is derived from the equations (202) to (205).

S _α (t, u)> S _β (t, u) (206)
P _α (t, u)> P _β (t, u) (207)
As described above, it is understood that there is no problem in a method of selecting a branch having the highest desired signal level or a branch having a high received signal level if the levels of the additional noise of all diversity branches of the same carrier are equal. .

  However, if the added noise level of each diversity branch of the same carrier is not equal, the signal power is corrected so that the added noise level is equal in all the diversity branches, and then equations (7) and (8) Need to apply.

The method for correcting the signal power is as follows. Assuming that the number of branches is 2, the SNR of the u th carrier at the time t of the first branch, the power of the desired signal, and the power of the additional noise are respectively SNR ₁ (t, u), S ₁ (t, u), N ₁ (t, u), the SNR of the u th carrier at the time t of the second branch, the power of the desired signal, and the power of the additional noise are SNR ₂ (t, u), S ₂ ( t, u), N ₂ (t, u). At this time, the following holds.

SNR ₁ (t, u) = S ₁ (t, u) / N ₁ (t, u), ... (208)
SNR ₂ (t, u) = S ₂ (t, u) / N ₂ (t, u), ... (209)
The corrected SNR ₁ (t, u), S ₁ (t, u), N ₁ (t, u) are converted to SNR ₁₁ (t, u), S ₁₁ (t, u), N ₁₁ (t, u).

The corrected SNR ₂ (t, u), S ₂ (t, u), N ₂ (t, u) are respectively converted into SNR ₂₂ (t, u), S ₂₂ (t, u), N ₂₂ (t, If u), the following should hold.

SNR ₁₁ (t, u) = S ₁₁ (t, u) / N ₁₁ (t, u), ... (210)
SNR ₂₂ (t, u) = S ₂₂ (t, u) / N ₂₂ (t, u), ... (211)
SNR ₁₁ (t, u) = SNR ₁ (t, u), ... (212)
SNR ₂₂ (t, u) = SNR ₂ (t, u), ... (213)
S ₁₁ (t, u) = {N / N ₁ (t, u)} S ₁ (t, u), (214)
S ₂₂ (t, u) = {N / N ₂ (t, u)} S ₂ (t, u), (215)
N ₁₁ (t, u) = N ₂₂ (t, u) = N (216)
Here, N is an arbitrary constant.

Therefore, the desired signal and additional noise of the u-th carrier at time t of the first branch are s ₁ (t, u) and n ₁ (t, u), respectively, and the u-th at time t of the second branch Let s ₂ (t, u) and n ₂ (t, u) be the desired signal and additional noise of the carrier wave, respectively.

When these corrected signals are s ₁₁ (t, u), n ₁₁ (t, u), s ₂₂ (t, u), and n ₂₂ (t, u), respectively, equations (214) to (214) to ( From 216), the following may be performed.

  The same can be said for the maximum ratio synthesis. That is, equations (9) and (10) are satisfied only when the condition that the added noise levels of all diversity branches of the same carrier are equal is satisfied. Therefore, if the added noise level of each diversity branch of the same carrier is not equal, the signal power is corrected so that the added noise level is equal in all diversity branches in the same manner as in the case of selective combining. It is necessary to carry out the synthesis according to the formulas (9) and (10).

  Here, the SNR in the receiver 9 adopting the OFDM method is calculated.

First, the signal of the carrier wave u at the time t of the k-th branch from the LPF 107 to the A / D converter 108 is assumed to be x _{ADC-IN, k} (t, u). _{x ADC-IN, k (t} , u) s ADC-IN the desired signal _{among, k (t, u),} noise generated by an analog circuit such as an amplifier from the antenna 10 to the A / D converter 108 (hereinafter X _{ADC-IN, k} (t, u) is expressed as follows _{, where} n _{th, k} (t, u) is a thermal noise).

x _{ADC-IN, k} (t, u) = s _{ADC-IN, k} (t, u) + n _{th, k} (t, u) (61)
As described above, the values of these signals are expressed as complex numbers (complex envelope, equivalent low-frequency signal) for convenience, the absolute value represents the amplitude of the sine wave, and the declination represents the phase of the sine wave. That is, a modulated signal x _sig (t) having a carrier frequency at time t as f _c , an amplitude as A (t), and a phase as φ (t) has Re [•] as a real part, and j ² = It is expressed as follows using an imaginary unit j which is −1.

x _sig (t) ＝ A (t) cos {2πf _c t ＋ φ (t)}
= A (t) cos (2πf _c t) cosφ (t) −A (t) sin (2πf _c t) sinφ (t)
= Re [x _comp (t) x _mod (t)], ... (71)
x _comp (t) = A (t) exp {jφ (t)} = A (t) {cosφ (t) + jsinφ (t)}, (72)
x _mod (t) = exp (j2πf _c t) = cos (2πf _c t) + jsin (2πf _c t), (73)
The I phase (in-phase component, corresponding to the real part) of the demodulated signal (complex envelope) represented by the equation (72) is expressed as x _{comp, I} (t), and the Q phase (corresponding to the quadrature component, corresponding to the imaginary part). As x _{comp, Q} (t), the equations (74) and (75) and the equation (72) become the equation (76).

x _{comp, I} (t) ＝ A (t) cosφ (t), ... (74)
x _{comp, Q} (t) ＝ A (t) sinφ (t) (75)
x _comp (t) = x _{comp, I} (t) + j x _{comp, Q} (t) (76)
Similarly, x _{ADC-IN, k} (t, u), s _{ADC-IN, k} (t, u), n _{th, k} (t, u) I phase are respectively represented by x _{ADC-IN, I, k} (t, u), s _{ADC-IN, I, k} (t, u), n _{th, I, k} (t, u), Q phase is x _{ADC-IN, Q, k} (t, u) , s _{ADC-IN, Q, k} (t, u), n _{th, Q, k} (t, u), the following is derived from equations (61) and (76).

x _{ADC-IN, k} (t, u) = x _{ADC-IN, I, k} (t, u) + j x _{ADC-IN, Q, k} (t, u), (81)
x _{ADC-IN, I, k} (t, u) = s _{ADC-IN, I, k} (t, u) + n _{th, I, k} (t, u), ... (82)
x _{ADC-IN, Q, k} (t, u) = s _{ADC-IN, Q, k} (t, u) + n _{th, Q, k} (t, u) (83)
Where s _{ADC-IN, I, k} (t, u) and s _{ADC-IN, Q, k} (t, u) have amplitude A _k (t, u) and phase φ _k (t, u) Then, s _{ADC-IN, I, k} (t, u) and s _{ADC-IN, Q, k} (t, u) can be expressed as follows from equations (74) and (75).

s _{ADC-IN, I, k} (t, u) = A _k (t, u) cosφ _k (t, u), (84)
s _{ADC-IN, Q, k} (t, u) = A _k (t, u) sinφ _k (t, u) (85)
Here, PSK (Phase Shift Keying) or QAM (Quadrature Phase Amplitude Modulation) and FSK (Frequency Shift Keying) used in normal mobile communication are assumed as modulation methods in each carrier wave. That is, the amplitude is assumed to be constant (in reality, it is assumed that the amplitude of the received signal gradually changes due to channel fluctuation (fading) but does not change within a short time). That is, assume that the following holds.

E [A _k (t, u)] = A _{k, u} (86)
Here, E [•] represents an average value with respect to time t, and A _{k, u} is a constant determined by k, u. The phase is random and the distribution is uniform. In this _{case, s ADC-IN, I,} k (t, u) dispersion E of _{[{s ADC-IN, I} , k (t, u)} 2] , the equation (84) and (86), below Guided like

_{Similarly, s ADC-IN, Q,} k (t, u) dispersion E of _{[{s ADC-IN, Q} , k (t, u)} 2] , the equation (85) and (86), It is derived as follows.

_{E [{s ADC-IN,} Q, k (t, u)} 2] = A k, u 2/2 ··· (88)
Also, n _{th, I, k} (t, u), n _{th, Q, k} (t, u) are assumed to be Gaussian noise having an average value of 0 and a variance of σ _{th, k} . Here, σ _{th, k} is a constant determined by k. That is, the following formula is established.

E [{n _{th, I, k} (t, u)} ² ] = E [{n _{th, Q, k} (t, u)} ² ] = σ _{th, k} ² (89)
At this time, the desired signal s _{ADC-IN, k} (t, u) = s _{ADC-IN, I, k} (t, u) + j s _{ADC-IN, Q, k} (t, u) and thermal noise n _{th, k} (t, u) = n _{th, I, k} (t, u) + j n _{th, Q, k} (t, u) are respectively _converted to S _{ADC-IN, k} (t, u) and N _{th, k} Assuming that (t, u), the average values E [S _{ADC-IN, k} (t, u)] and E [N _{th, k} (t, u)] are represented by the equations (88) and (89), respectively. Is expressed as follows.

E [S _{ADC-IN, k} (t, u)] = E [| s _{ADC-IN, k} (t, u) | ² ]
= E [{s _{ADC-IN, I, k} (t, u)} ² ] + E [{s _{ADC-IN, Q, k} (t, u)} ² ] = A _{k, u} ² ,. 90)
E [N _{th, k} (t, u)] ＝ E [| n _{th, k} (t, u) | ² ]
= E [{n _{th, I, k} (t, u)} ² ] + E [{n _{th, Q, k} (t, u)} ² ] = 2σ _{th, k} ² (91)
Therefore, if the power of x _{ADC-IN, k} (t, u) is P _{ADC-IN, k} (t, u), the average value E [P _{ADC-IN, k} (t, u)] From (61), (90), (91), it is expressed as follows.

E [P _{ADC-IN, k} (t, u)] = E [| x _{ADC-IN, k} (t, u) | ² ] = E [| s _{ADC-IN, k} (t, u) | ² ] + E [| n _{th, k} (t, u) | ² ]
= E [S _{ADC-IN, k} (t, u)] + E [N _{th, k} (t, u)] = A _{k, u} ² + 2σ _{th, k} ² (92)
Where s _{ADC-IN, I, k} (t, u), s _{ADC-IN, Q, k} (t, u), n _{th, I, k} (t, u), n _{th, Q, k} ( t, u) are assumed to be independent of each other.

Subsequently, after passing through the A / D conversion unit 108, that is, from the A / D conversion unit 108 to the LPF 105, the signal of the carrier wave u at the time t of the k-th branch is expressed as x _{ADC-OUT, k} (t, u). To do. x _{ADC-OUT, k} (t, u) I phase is x _{ADC-OUT, I, k} (t, u) and Q phase is x _{ADC-OUT, Q, k} (t, u). As in (76), the following is derived.

x _{ADC-OUT, k} (t, u) = x _{ADC-OUT, I, k} (t, u) + j x _{ADC-OUT, Q, k} (t, u) (101)
Here, the A / D converter 108 quantizes the I phase and Q phase of the received signal sampled at the sampling frequency as digital signals. If the quantization step of the A / D converter 108 is q, x _{ADC-IN, I, k} (t, u), x _{ADC-IN, Q, k} (t, u) Can be expressed as follows.

x _{ADC-IN, I, k} (t, u) = qv _{ADC-IN, I, k} (t, u), ... (102)
x _{ADC-IN, Q, k} (t, u) = qv _{ADC-IN, Q, k} (t, u) (103)
At this time, x _{ADC-OUT, I, k} (t, u) and x _{ADC-OUT, Q, k} (t, u) are rounded to an integral multiple of q, so from equations (102) and (103) It is expressed as follows.

x _{ADC-OUT, I, k} (t, u) = x _{ADC-IN, I, k} (t, u) + n _{qu, I, k} (t, u) = qv _{ADC-OUT, I, k} (t, u), ... (104)
x _{ADC-OUT, Q, k} (t, u) = x _{ADC-IN, Q, k} (t, u) + n _{qu, Q, k} (t, u) = qv _{ADC-OUT, Q, k} (t, u), ... (105)
n _{qu, I, k} (t, u) = q {v _{ADC-OUT, I, k} (t, u) −v _{ADC-IN, I, k} (t, u)}, (106)
n _{qu, Q, k} (t, u) = q {v _{ADC-OUT, Q, k} (t, u) −v _{ADC-IN, Q, k} (t, u)}, (107)
v _{ADC-OUT, I, k} (t, u) = round [v _{ADC-IN, I, k} (t, u)], ... (108)
v _{ADC-OUT, Q, k} (t, u) = round [v _{ADC-IN, Q, k} (t, u)], ... (109)
Where n _{qu, I, k} (t, u) and n _{qu, Q, k} (t, u) are respectively x _{ADC-IN, I, k} (t, u) and x _{ADC-IN, Q, k} (t, u) is an error (hereinafter referred to as quantization noise) that occurs when the A / D conversion unit 108 quantizes.

Also, v _{ADC-OUT, I, k} (t, u) and v _{ADC-OUT, Q, k} (t, u) are x _{ADC-OUT, I, k} (t, u) and x _{ADC-OUT, respectively. , Q, k} (t, u) by the A / D converter 108 quantization level (quantized integer after A / D conversion. (Quantization step) × (quantization level) is a true value ). In addition, round [•] indicates that • is rounded off.

Here, n _{qu, I, k} (t, u) and n _{qu, Q, k} (t, u) are uniformly distributed within the quantization step, and assuming that the average value is 0, n _{qu, I, k (t,} u) of the dispersion _{E [{n qu, I,} k (t, u)} 2] is at _{n qu, I, k (t} , u) = a e, as follows It is expressed in

  Here, p (e) is a probability density function and, as described above, is assumed to be uniform within the quantization step, and is expressed as follows.

p (e) = 1 / q (111)
_{Similarly, n qu, Q, k (} t, u) of the dispersion _{E [{n qu, Q,} k (t, u)} 2] is expressed as follows.

_{E [{n qu, Q,} k (t, u)} 2] = q 2/12 ··· (112)
_{Assuming that} the power of the quantization noise n _{qu, k} (t, u) is N _{qu, k} (t, u), the average value E [N _{qu, k} (t, u)] is expressed by equations (110) and ( 112), it is expressed as follows.

E [N _{qu, k} (t, u)] = E [| n _{qu, k} (t, u) | ² ] = E [{n _{qu, I, k} (t, u)} ² ] + E [{n _{qu, Q, k} (t, u)} ² ]
^{= Q 2/6 ··· (113} )
On the other hand, of x _{ADC-OUT, I, k} (t, u), the desired signal is s _{ADC-OUT, I, k} (t, u) and the additional noise is n _{ADC-OUT, I, k} (t, u) And Similarly, of x _{ADC-OUT, Q, k} (t, u), the desired signal is s _{ADC-OUT, Q, k} (t, u) and the additional noise is n _{ADC-OUT, Q, k} (t, u ). At this time, x _{ADC-OUT, I, k} (t, u) and x _{ADC-OUT, Q, k} (t, u) are rewritten as follows.

x _{ADC-OUT, I, k} (t, u) = s _{ADC-OUT, I, k} (t, u) + n _{ADC-OUT, I, k} (t, u), ... (129)
x _{ADC-OUT, Q, k} (t, u) = s _{ADC-OUT, Q, k} (t, u) + n _{ADC-OUT, Q, k} (t, u), ... (130)
From the equations (82), (83), (104), (105), (129) and (130), the following is derived.

s _{ADC-OUT, I, k} (t, u) = s _{ADC-IN, I, k} (t, u), ... (131)
s _{ADC-OUT, Q, k} (t, u) = s _{ADC-IN, Q, k} (t, u), ... (132)
n _{ADC-OUT, I, k} (t, u) = n _{th, I, k} (t, u) + n _{qu, I, k} (t, u), (133)
n _{ADC-OUT, Q, k} (t, u) = n _{th, Q, k} (t, u) + n _{qu, Q, k} (t, u), ... (134)
At this time, the power of the desired signal s _{ADC-OUT, k} (t, u) and additional noise n _{ADC-OUT, k} (t, u) is changed to S _{ADC-OUT, k} (t, u) and N _{ADC-OUT, respectively. , k} (t, u), the average values E [S _{ADC-OUT, k} (t, u)] and E [N _{ADC-OUT, k} (t, u)] , (91), (113), it is expressed as follows.

E [S _{ADC-OUT, k} (t, u)] = E [| s _{ADC-OUT, k} (t, u) | ² ]
= E [{s _{ADC-OUT, I, k} (t, u)} ² ] + E [{s _{ADC-OUT, Q, k} (t, u)} ² ]
= E [{s _{ADC-IN, I, k} (t, u)} ² ] + E [{s _{ADC-IN, Q, k} (t, u)} ² ]
= A _{k, u} ² , ... (135)
E [N _{ADC-OUT, k} (t, u)] = E [| n _{ADC-OUT, k} (t, u) | ² ]
= E [{n _{ADC-OUT, I, k} (t, u)} ² ] + E [{n _{ADC-OUT, Q, k} (t, u)} ² ]
= E [{n _{th, I, k} (t, u)} ² ] + E [{n _{th, Q, k} (t, u)} ² ] + E [{n _{qu, I, k} (t, u)} ² ] + E [{n _{qu, Q, k} (t, u)} ² ]
^{_{= 2σ th, k 2 + q}} 2/6, ··· (136)
Therefore, if the power of x _{ADC-OUT, k} (t, u) is P _{ADC-OUT, k} (t, u), the average value E [P _{ADC-OUT, k} (t, u)] is It is expressed as

E [P _{ADC-OUT, k} (t, u)] = E [| x _{ADC-OUT, k} (t, u) | ² ]
= E [| s _{ADC-OUT, k} (t, u) | ² ] + E [| n _{ADC-OUT, k} (t, u) | ² ]
^{_{= A k, u 2 + 2σ}} th, k 2 + q 2/6 ··· (137)
As described above, after passing through the A / D conversion unit 108, that is, the SNR in the signal x _{ADC-OUT, k} (t, u) from the A / D conversion unit 108 to the LPF 105 is converted to SNR _{ADC-OUT, k} (t, u). Then, the average value E [SNR _{ADC-OUT, k} (t, u)] is expressed as follows from the equations (135) and (136).

E [SNR _{ADC-OUT, k} (t, u)] = E [S _{ADC-OUT, k} (t, u)] / E [N _{ADC-OUT, k} (t, u)]
^{_{= A k, u 2 / {}} 2σ th, k 2 + q 2/6} ··· (138)
Here, as in equations (102) and (103), A _{k, u} , σ _{th, k} can be expressed as follows using q.

A _{k, u} = qv _{A, k, u} , ... (139)
σ _{th, k} = qv _{th, k} (140)
Here, v _{A, k, u} is a constant determined by k, u, and v _{th, k} is a constant determined by k. At this time, the following is derived from the equations (87) to (89), (131), and (132).

E [{s _{ADC-OUT, I, k} (t, u)} ² ] = E [{s _{ADC-OUT, Q, k} (t, u)} ² ] = q ² v _{A, k, u} ² / 2 (88-2)
E [{n _{th, I, k} (t, u)} ² ] = E [{n _{th, Q, k} (t, u)} ² ] = q ² v _{th, k} ² (89-2) )
That is, v _{A, k, u} / 2 ^0.5 is the average value of the absolute values of the quantization levels of s _{ADC-OUT, I, k} (t, u) and s _{ADC-OUT, Q, k} (t, u), v _{th, k} represents an average value of absolute values of quantization levels of n _{th, I, k} (t, u) and n _{th, Q, k} (t, u).

  Similarly, formulas (91) and (135) to (137) can be rewritten as follows.

E [N _{th, k} (t, u)] = 2q ² v _{th, k} ² , (91-2)
E [S _{ADC-OUT, k} (t, u)] = q ² v _{A, k, u} ² , (135-2)
_{E [N ADC-OUT, k} (t, u)] = 2q 2 v th, k 2 + q 2/6 = q 2 (2v th, k 2 +1/6), ··· (136-2)
E [P _{ADC-OUT, k} (t, u)] = q ² (v _{A, k, u} ² + 2v _{th, k} ² +1/6) (137-2)
If v _{ADC-OUT, k} (t, u) = v _{ADC-OUT, I, k} (t, u) + j v _{ADC-OUT, Q, k} (t, u), Equation (104), ( 105) and (137-2) lead to the following.

E [| v _{ADC-OUT, k} (t, u) | ² ] = E [P _{ADC-OUT, k} (t, u)] / q ²
= V _{A, k, u} ² + 2v _{th, k} ² +1/6 (141)
The following is derived from the equations (138) to (140).

E [SNR _{ADC-OUT, k} (t, u)] = v _{A, k, u} ² / (2v _{th, k} ² +1/6) (142)
In addition, an input signal to the antenna 10 of the u-th carrier at time t of the k-th branch is s _{ANT, k} (t, u), and its power (received power at the antenna 10) is S _{ANT, k} (t , u). At this time, the average power E [S _{ANT, k} (t, u)] can be expressed as follows.

E [S _{ANT, k} (t, u)] = G ^ _{k, u} E [S _{ADC-IN, k} (t, u)] (151)
Here, G ^ _{k, u} represents the average gain of the analog circuit from the antenna 10 to the LPF 107, and is a constant determined by k, u.

From the equations (90), (139) and (151), the following is derived.

Here, q _ANT is a quantization step when the circuit from the antenna 10 to the A / D converter 108 is made one equivalent A / D converter. The average power E [N ^ _{qu, k} (t, u)] of the equivalent quantization noise at this time is expressed as follows from the equation (113).

_{E [N ^ qu, k (} t, u)] = q ANT 2/6 ··· (154)
Therefore, E [SNR _{ADC-OUT, k} (t, u)] is rewritten as follows from the equations (91), (152), and (154).

E [SNR _{ADC-OUT, k} (t, u)]
= E [S _{ANT, k} (t, u)] / {E [N _{th, k} (t, u)] + E [N ^ _{qu, k} (t, u)]}, (155)
E [N _{th, k} (t, u)] = 2σ _{th, k} ² , (156)
E [N ^ _{qu, k} (t, u)] = E [S _{ANT, k} (t, u)] / (6 v _{A, k, u} ² ) (157)
From the equations (155) to (157), in the OFDM receiver 9, in addition to the thermal noise due to the analog circuit from the antenna 10 to the LPF 107, the A / D converter 108 averages the absolute value of the quantization level of the desired signal. Value v _{A, k, u} / 2 Quantization noise proportional to the reciprocal of the square of ^0.5 is added.

v _{A, k, u} can be rephrased as a resolution based on its definition. That is, when the quantization step q is large, v _{A, k, u} is small from Equations (102) and (103). This means that the resolution is small. On the other hand, when the quantization step q is small, v _{A, k, u} becomes large. This means that the resolution is large.

That is, from the equation (157), when the resolution is small (v _{A, k, u} is small) with respect to the average power E [S _{ANT, k} (t, u)] of the input signal to the antenna 10, When the average power E [N ^ _{qu, k} (t, u)] of quantization noise is large and the resolution is large (v _{A, k, u} is large), the average power E [N ^ _{qu of} quantization noise _{, k} (t, u)] is reduced.

Therefore, even if E [S _{ANT, k} (t, u)] is sufficiently large with respect to thermal noise, if the resolution is small (v _{A, k, u} is small), the quantization noise is E [S _{ANT, k} (t, u)] will increase to the same order. This is E (N ^ _{qu, k} (t, u)] → 0 when v _{A, k, u} → ∞ in equation (157), but v _{A, k, u} is decreased and v _It can also be seen from the fact that when _{A, k, u} → 1, E [N ^ _{qu, k} (t, u)] → E [S _{ANT, k} (t, u)] / 6.

  This phenomenon is particularly noticeable when frequency selective fading occurs and the received signal power of each carrier is different. This will be described below.

  The AGC control unit 106 performs control so that the level of the input signal to the A / D conversion unit 108 is appropriate. The input signal to the A / D conversion unit 108 is supplied to each of the L pieces of signals by the DFT unit 13. This is a signal before being separated into a baseband signal that has been modulated by the carrier frequency, and is a signal in which all the carriers have been added. Therefore, the AGC control unit 106 performs control so that the total power (the power of the signal obtained by adding all the carrier waves) is constant, but is independent of the power value of the signal of each carrier wave.

  Here, the A / D conversion unit 108 quantizes the signal obtained by adding all the carrier waves. Therefore, when frequency selective fading occurs and the power of the signal of each carrier is different, the resolution of the signal of each carrier is different.

  When the total power is large, the power of the signal of each carrier becomes large (it is sufficiently larger than the power of thermal noise). On the other hand, the quantization noise increases.

  On the other hand, when the total power is small, the power of the signal of each carrier is small (it cannot be said that it is sufficiently larger than the power of thermal noise), but the resolution is large in the carrier of power larger than other carriers. Therefore, the quantization noise power is smaller than the carrier wave power.

  As a result, even if the power of the carrier signal is the same, the power of the quantization noise differs depending on the total power (in other words, the relative power with other branches). That is, in a certain carrier wave, the power of the desired signal is the same, but the power of the quantization noise can be large or small depending on the magnitude of the total power. SNR can be large or small.

  Therefore, as shown in FIG. 4, among all the diversity branches in the same carrier wave, the total power is high in a certain branch, so that the quantization noise power is higher than the power of the desired signal, so that the SNR is deteriorated. In another branch, although the total power is low, the quantization noise power is low relative to the power of the desired signal, so that the SNR may be improved.

As described above, even when the thermal noise level is the same for all branches and all carriers, the quantization noise level is not always equal, and therefore the additional noise is not equal. That is, the assumption of formula (205) does not hold. Therefore, since Equation (206) does not always hold, the average power E [S _{ANT, k} (t, u)] of the desired signal is calculated by Equation (49) with reference to the gain value of the AGC amplifier 102. Even if it is possible, it is impossible to perform diversity by selective synthesis applying Eqs. (7) and (8) with reference to E [S _{ANT, k} (t, u)] only I understand. Similarly, even in the case of the maximum ratio combining, it is impossible to perform diversity by the maximum ratio combining applying the formulas (9) and (10).

  Therefore, in order to solve the above problems, the present invention requires a digital circuit that realizes DFT in the OFDM receiver, and the generation of quantization noise due to the resolution limit in A / D conversion is inevitable. Even when the level of quantization noise differs from diversity branch to diversity branch, a diversity receiver is provided that enables diversity combining to operate properly.

  In order to solve such a problem, the first aspect of the present invention includes a plurality of branch processing means for demodulating received signals modulated by a plurality of carrier frequencies, each having a different radio wave capturing position, In the diversity receiver including the diversity combining unit that selects and outputs one of the received signals from the branch processing unit, each of the plurality of branch processing units includes (1) each separated by discrete Fourier transform, Thermal noise that estimates the average power of thermal noise contained in the baseband received signal that has been modulated by the carrier frequency and corrects the value of the baseband received signal so that it becomes a predetermined value equal to all other branch processing means A correction unit, and (2) a quantization bit width of the baseband received signal corrected by the thermal noise correction unit (the signal is quantized in a predetermined quantization step) The average power of the baseband received signal is estimated so that the predetermined quantization bit width is smaller than the quantization bit width, and the estimated baseband received signal Diversity combining with a quantization noise correction unit that corrects the baseband received signal so that the power falls within the quantized bit width after the reduction when the average power is larger than the reduced quantization bit width. Means is a diversity receiver characterized in that the baseband received signal corrected by the quantization noise correction unit of each branch processing means is diversity combined.

According to a second aspect of the present invention, radio wave capturing positions are different, and a plurality of branch processing means for demodulating a modulated signal modulated by a plurality of carrier frequencies, and a received signal from each branch processing means In the diversity receiver including the diversity combining means for selecting and outputting any of them, each of the plurality of branch processing means is a baseband reception signal modulated by each carrier frequency and separated by discrete Fourier transform. The average power of thermal noise included in the baseband received signal, which was modulated by the same carrier frequency of all other branch processing means , is estimated for each carrier frequency. so that the predetermined value is equal to the power, with a thermal noise correcting unit for correcting the value of the baseband received signals for each carrier frequency, Ibashichi synthesis means is a diversity receiver, characterized in that the baseband reception signal corrected by the thermal noise correction unit of each branch processing means for diversity combining.

  According to a third aspect of the present invention, there are different radio wave capturing positions, a plurality of branch processing means for demodulating modulated signals modulated by a plurality of carrier frequencies, and a received signal from each branch processing means. In a diversity receiver including a diversity combining unit that selects and outputs one of the plurality of branch processing units, each of the plurality of branch processing units is quantized and modulated by each carrier frequency separated by a discrete Fourier transform The average power of the baseband received signal is estimated so that the quantized bit width of the baseband received signal becomes a predetermined quantization bit width smaller than the quantized bit width, and the estimated average power of the baseband received signal is estimated. If the value is larger than the reduced quantization bit width, the baseband is set so that the power is within the reduced quantization bit width. A diversity receiver comprising a quantization noise correction unit for correcting a received signal, wherein diversity combining means diversity combines the baseband received signal corrected by the quantization noise correction unit of each branch processing means .

  According to the present invention, diversity combining can be appropriately performed even when the level of quantization noise differs for each diversity branch.

It is an internal block diagram which shows the internal structure of the receiver of 1st Embodiment. It is an internal block diagram which shows the internal structure of the conventional receiver. It is an internal block diagram which shows the internal structure of a demodulation part. It is explanatory drawing explaining the subject in an OFDM receiver. It is an internal block diagram which shows the internal structure of the receiver of 2nd Embodiment. It is an internal block diagram which shows the internal structure of the receiver of 3rd Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a diversity receiver of the present invention will be described in detail with reference to the drawings.

  In this embodiment, a case where the present invention is applied to a receiver that employs an OFDM system as a modulation system is illustrated.

(A-1) Configuration of First Embodiment (A-1-1) Description of Internal Configuration of Receiver FIG. 1 is an internal configuration diagram illustrating an internal configuration of a receiver of the first embodiment. Note that the receiver 1A according to the first embodiment is a receiver that employs the OFDM method and has a spatial diversity function. Here, a case where the number of diversity branches (hereinafter referred to as branches) is two is illustrated. In addition, a case where the number of subcarriers (the number of dispersed carriers) is L is illustrated.

  As shown in FIG. 1, the receiver 1A of the first embodiment includes an antenna 10, a demodulator 11, an S / P (serial-parallel) converter 12, a DFT unit 13, a thermal noise corrector 14, and a quantization noise correction. 15, antenna 20, demodulator 21, S / P converter 22, DFT unit 23, thermal noise corrector 24, quantization noise corrector 25, diversity combiner 30, diversity combiner 31, diversity combiner 32, P An / S converter 33 and a determination unit 34 are included.

  The receiver 1A in FIG. 1 additionally includes a thermal noise correction unit 14 and a quantization noise correction unit 15 between the output of the DFT unit 13 and the inputs to the diversity combining units 30 to 32, the DFT The configuration of the receiver 9 in FIG. 2 is that a thermal noise correction unit 24 and a quantization noise correction unit 25 are newly added between the output of the unit 23 and the inputs to the diversity combining units 30 to 32. Is different.

  A transmitter (not shown) adopting the OFDM scheme encodes an information bit string in a baseband (baseband) and modulates it to a carrier frequency (high frequency) band. The modulated signal is radiated into the space from an antenna of a transmitter (not shown) and input to the antenna 10 and the antenna 20 of the receiver.

  The carrier frequency band modulation signal input to the antenna 10 is reproduced as a baseband digital signal by the demodulator 11 and input to the S / P converter 12.

  The baseband digital signal input to the S / P converter 12 accumulates L sample values in the time direction by serial-parallel conversion, and is input to the DFT unit 13 in a lump.

  The L baseband signals input to the DFT unit 13 are separated into baseband signals that have been modulated by L carrier frequencies (subcarriers) by performing a discrete Fourier transform, and each of the L carrier waves. The baseband signal that has been modulated by the frequency is supplied to the thermal noise correction unit 14.

  The baseband signal that has been modulated by each of the L carrier frequencies input to the thermal noise correction unit 14 is such that the average power of the thermal noise contained in each carrier signal is the same in all branches. Normalized and input to the quantization noise correction unit 15.

  The average power of the thermal noise included in the signals of the respective carriers modulated by the L carrier frequencies input to the quantization noise correction unit 15 is normalized so as to be equal in all branches. The baseband signal is limited to a predetermined quantization bit width so that the average power of the quantization noise included in each carrier signal is the same value in all branches, and the diversity combining units 30 to 32 receive the baseband signal. Entered.

  Although only three diversity combining units are described here, the number is actually the same as the number of carrier waves, that is, L diversity combining units.

  Diversity combining sections 30 to 32 combine signals from two branches having the same carrier frequency (the same subcarrier number) by a predetermined method and output the combined signals to P / S conversion section 33 as one signal.

Here, selective synthesis or maximum ratio synthesis is assumed as an example of the synthesis method. That is, in the case of selective combining, the input signal from the DFT unit 13 of the u-th carrier at the time t of the k-th branch to the diversity combining units 30 to 32 is x _k (t, u). 8) is applied, and y (t, u) is used as an output signal of diversity combining sections 30-32. In the case of maximum ratio combining, equations (9) and (10) are applied, and y (t, u) is used as the output signal of diversity combining sections 30-32.

  Since the operations of the diversity combining units 31 and 32 are the same as those of the diversity combining unit 30, the description thereof is omitted.

  The baseband signals related to the L carrier waves input to the P / S converter 33 are parallel-serial converted and input to the determination unit 34 as L sample values in the time direction.

  The baseband signal input to the determination unit 34 is reproduced as an original information bit string by performing processing such as decoding.

  On the other hand, the modulated signal in the carrier frequency band input to the antenna 20 is regenerated to a baseband signal by the demodulator 21 and input to the S / P converter 22. The subsequent operations of the S / P conversion unit 22, the DFT unit 23, the thermal noise correction unit 24, and the quantization noise correction unit 25 are the S / P conversion unit 12, the DFT unit 13, the thermal noise correction unit 14, and the quantization noise, respectively. Since it is the same as that of the correction | amendment part 15, description is abbreviate | omitted.

(A-1-2) Description of Internal Configuration of Demodulator FIG. 3 is an internal configuration diagram showing an internal configuration of the demodulator 11. The demodulator 21 has the same configuration as the demodulator 11.

  The demodulator 11 includes a BPF (band pass filter) 101, an AGC (automatic gain control) amplifier 102, a down converter 103, a local oscillator 104, an LPF (low pass filter) 107, and an A / D (analog-digital). The conversion unit 108, the LPF 105, and the AGC control unit 106 are included.

  The signal from the antenna 10 is passed through a predetermined band centered on a desired carrier frequency by the BPF 101, whereby a modulated signal in the carrier frequency band is extracted and input to the AGC amplifier 102.

  The modulation signal input to the AGC amplifier 102 is amplified to a predetermined gain determined by the AGC control unit 106 and input to the down converter 103.

  Here, only one AGC amplifier 102 is provided between the output of the BFP 101 and the input to the down converter 103. However, the AGC amplifier is added to another position to form a plurality of AGC amplifiers. You may do it. For example, the same method is used even when a plurality of configurations are added between the output of the down converter 103 and the input to the LPF 107 or between the output of the LPF 107 and the input to the A / D converter 108. Is possible. In any case, since the generality of the present invention is not impaired, only the description of the case where only one AGC amplifier 102 is provided between the output of the BFP 101 and the input to the down converter 103 will be given here.

  The down converter 103 reproduces a baseband signal by multiplying the modulated signal in the carrier frequency band from the AGC amplifier 102 by the reference signal from the local oscillator 104 and outputs the baseband signal to the LPF 105.

  Here, a case will be described in which the baseband signal is reproduced by assuming direct conversion and by making the frequency of the reference signal the same as the carrier frequency. However, there is a method in which the signal is once converted into a signal in a frequency band intermediate between the carrier frequency band and the baseband and then converted into a baseband signal. Any method does not impair the generality of the present invention, so only a description of direct conversion is assumed here.

  The signal from the down converter 103 is passed through only a band of a predetermined frequency or lower (for example, a band necessary for transmission of a desired signal (a signal component obtained by removing additional noise from the received signal)) by the LPF 107, Unnecessary components are removed and input to the A / D converter 108.

  Since the LPF 107 is an analog filter, it has a pass bandwidth that is usually slightly wider than the bandwidth of the desired signal because it is easier to implement (because it is difficult to achieve the same and steep characteristics as the bandwidth of the desired signal). .

  The A / D converter 108 samples the analog signal from the LPF 107 at a predetermined frequency, converts it into a quantized digital signal, and outputs the digital signal to the LPF 105.

  The signal from the A / D converter 108 is passed through only a band of a predetermined frequency or less by the LPF 105 to remove unnecessary components and is input to the S / P converter 12 and the AGC controller 106.

  The LPF 105 is a digital filter based on FIR (finite impulse response) or the like, and limits a signal having a wider bandwidth than the desired signal that has passed through the LPF 107 to the bandwidth of the desired signal.

  The AGC control unit 106 determines the gain of the AGC amplifier 102 so that the input / output signal level of each unit in the receiver 1A is maintained within a predetermined range. That is, each circuit element in the receiver 1A has an upper limit value and a lower limit value of a voltage that can be output. If the signal level is too larger than the upper limit value or smaller than the lower limit value, the amplitude of the signal is not proportional to the voltage of each circuit element, and the signal waveform cannot be reproduced correctly. Therefore, since the signal level needs to be maintained at an appropriate value between the upper limit value and the lower limit value, the AGC control unit 106 determines the gain of the AGC amplifier 102.

  That is, the level of the baseband signal from the A / D conversion unit 108 is observed, and the gain of the AGC amplifier 102 is determined so that this value converges to a predetermined target value. For the observation of the level of the baseband signal, for example, a method of calculating an integral value within a predetermined time can be used to smooth the level of the signal.

  The configuration of the receiver 1A when the number of branches is 2 has been described above. However, the configuration is the same when the number of branches is 3 or more.

  That is, the antenna 10, the demodulator 11, the S / P converter 12, the DFT converter 13, the thermal noise corrector 14, and the quantization noise corrector 15 exist as many as the number of branches, and the diversity combiners 30 to 32. Receives the signals from the quantization noise correction units of all branches, that is, the quantization noise correction units 15 and 25 and the quantization correction units of other branches. Since other explanations are the same as in the case where the number of branches is two, they are omitted.

(A-2) Operation of the First Embodiment Next, the operation of processing in the receiver 1A of the first embodiment will be described.

  Since the processing operation from the antenna 10 to the DFT unit 13, the processing operation from the antenna 20 to the DFT unit 23, and the processing operation of the determination unit 34 are the same as the conventional processing operation, detailed operations are omitted here.

The baseband signal modulated by the L carrier frequencies from the DFT unit 13 to the thermal noise correction unit 14 is the signal of the u-th carrier at time t of the k-th branch x _{DFT, k} (t, If u) and the I phase and Q phase are x _{DFT, I, k} (t, u) and x _{DFT, Q, k} (t, u), respectively, they are expressed as follows.

x _{DFT, k} (t, u) = x _{DFT, I, k} (t, u) + j x _{DFT, Q, k} (t, u), (301)
x _{DFT, I, k} (t, u) = qv _{DFT, I, k} (t, u) (301-2)
x _{DFT, Q, k} (t, u) = qv _{DFT, Q, k} (t, u) (301-3)
v _{DFT, I, k} (t, u) =-z _{I, B-1} 2 ^B-1 + z _{I, B-2} 2 ^B-2 + ... + z _{I, 0} 2 ⁰ , ... (302)
v _{DFT, Q, k} (t, u) = − z _{Q, B−1} 2 ^B−1 + z _{Q, B−2} 2 ^B−2 +… + z _{Q, 0} 2 ⁰ (303)
Here, q is a quantization step, and v _{DFT, I, k} (t, u), v _{DFT, Q, k} (t, u) are respectively x _{DFT, I, k} (t, u ), x _{DFT, Q, k} (t, u) quantization level. Z _{I, B−1} , z _{I, B−2} ,..., Z _{I, 0} and z _{Q, B−1} , z _{Q, B−2} ,..., Z _{Q, 0} are integers of 0 or 1 is there.

That is, since v _{DFT, I, k} (t, u) and v _{DFT, Q, k} (t, u) are the operation results of the quantized digital signal after passing through the A / D conversion unit 108, An integer expressed as a B-bit binary number, and assuming a two's complement expression as in equations (302) and (303), the minimum value is -2 ^B-1 and the maximum value is 2 ^B-1 An arbitrary integer value is taken within the range of -1. That is, the quantization bit width of x _{DFT, I, k} (t, u), x _{DFT, Q, k} (t, u) is B.

  Here, 2's complement is assumed as the binary representation, but the method described below can be similarly applied to other representations such as unsigned integers and signed absolute values. Hereinafter, only the case of using 2's complement will be described.

On the other hand, when the SNR at x _{DFT, k} (t, u) is SNR _{DFT, k} (t, u), the average value E [SNR _{DFT, k} (t, u)] is expressed as follows: .

E [SNR _{DFT, k} (t, u)]
= E [S _{DFT, k} (t, u)] / {E [N _{DFT, th, k} (t, u)] + E [N _{DFT, qu, k} (t, u)]} (305)
Where S _{DFT, k} (t, u), N _{DFT, th, k} (t, u) and N _{DFT, qu, k} (t, u) are respectively x _{DFT, k} (t, u) It represents the power of the desired signal, the power of thermal noise, and the power of quantization noise. Further, the following is derived in the same manner as in equations (135-2), (91-2), and (113).

E [S _{DFT, k} (t, u)] = q ² v _{DFT, A, k, u} ² , (306)
E [N _{DFT, th, k} (t, u)] = 2q ² v _{DFT, th, k, u} ² , (307)
_{E [N DFT, qu, k} (t, u)] = q 2/6 ··· (308)
Here, v _{DFT, A, k, u} / 2 ^0.5 and v _{DFT, th, k, u} are the absolute values of the desired signal and thermal noise quantization levels in x _{DFT, k} (t, u), respectively. Is the average value.

The thermal noise correction unit 14 estimates the average power of the thermal noise by a predetermined method so that the diversity operation is appropriately performed on the signal x _{DFT, k} (t, u) from the DFT unit 13, and The signal value is corrected so that the average power becomes a predetermined value equal in all branches.

The AGC control unit 106 controls the total power to the A / D conversion unit 108 to be a constant value. Since the total power is the sum of the power of the desired signal and the power of thermal noise, the power of the desired signal Is small and not sufficiently large with respect to the power of thermal noise, the ratio of thermal noise to the total power becomes high. Therefore, the smaller the power of the desired signal, the higher the proportion of thermal noise contained in _{xDFT, k} (t, u). Therefore, it is necessary to correct the influence so as to be constant.

  First, here are two examples of methods for estimating the average power of thermal noise.

  The first example uses a known transmission sequence repetition.

The first and second signals (symbols) of a known sequence of the carrier wave u at time t of the k-th branch of _{xDFT, k} (t, u) are x _{known, 1, k} (t, u ), x _{known, 2, k} (t + δ, u) (where δ> 0, and δ represents the time difference between the first and second symbols), which was the same value at the time of transmission 2 Since the two signals are different due to the addition of thermal noise, the average power E [N _{DFT, th, k} (t, u)] of thermal noise included in x _{DFT, k} (t, u) is estimated. The quantity is calculated as follows:

  For example, in the case of the IEEE802.11a standard, since the long preamble is a repetition of a known transmission sequence, the average power of thermal noise can be estimated by applying the above formula.

In the second example, the value recorded in the memory or the like in advance by actual measurement or the like is used, and the received power at the antenna 10 estimated from the gain of the AGC amplifier 102 and the average power of the thermal noise are used. Thus, the average power of the thermal noise included in x _{DFT, k} (t, u) is estimated. The details will be described below.

In the r-th power integration by the AGC control unit 106, the power of the input signal to the antenna 10 of the u-th carrier at the time t of the k-th branch (received power at the antenna 10) is expressed as P _{ANT, k} (t, u ), A gain value set by the AGC control unit 106 in the AGC amplifier 102 is G _{k, r} , and gains by other circuits are F _{k, r} .

Then, since the power P _{LPF, k} (t) of the input signal to the AGC control unit 106, which is the left side of the equation (41), is constant, the gain value G _{k, If r} is determined, the average power E [P _{ANT, k} (t)] of the input signal to the antenna 10 is uniquely determined. That is, E [P _{ANT, k} (t)] is a function of G _{k, r} and the following holds.

E [P _{ANT, k} (t)] = f (G _{k, r} ) (321)
The function of Equation (321) usually shows almost linearity, but the relationship between E [P _{ANT, k} (t)] and G _{k, r} is measured in advance, and the correspondence is recorded in a memory or the like. Thus, E [P _{ANT, k} (t)] is estimated from G _{k, r} , that is, the received power at the antenna 10 is estimated from the gain value set by the AGC control unit 106 in the AGC amplifier 102. It is possible. The average power E [N _{th, k} (t)] of the thermal noise is recorded in a memory or the like from the design value and the actual measurement value.

The signal whose gain has been adjusted by the AGC amplifier 102 and passed through the A / D converter 108 is maintained at the subsequent stage while maintaining the ratio of E [P _{ANT, k} (t)] and E [N _{th, k} (t)]. The desired signal included in the output signal x _{DFT, k} (t, u) of the DFT unit 13 is s _{DFT, k} (t, u) and the thermal noise is n _{DFT, th, k} (t , u), there is the following relationship.

The relationship between x _{DFT, k} (t, u) and s _{DFT, k} (t, u) and n _{DFT, th, k} (t, u) is expressed as follows.

x _{DFT, k} (t, u) = s _{DFT, k} (t, u) + n _{DFT, th, k} (t, u) (323)
The following is derived from the equation (323).

E [| x _{DFT, k} (t, u) | ² ] = E [| s _{DFT, k} (t, u) | ² ] + E [| n _{DFT, th, k} (t, u) | ² ]・ (324)
Equation (322), E from (324) to erase ^{_{the, E [| 2 | s DFT}} , k (t, u)] [| n DFT, th, k (t, u) | 2] in all carriers Equal (E [| n _{DFT, th, k} (t, 1) | ² ] = E [| n _{DFT, th, k} (t, 2) | ² ] =… = E [| n _{DFT, th, k} ( t, L) | ² ]), the estimated thermal noise average power E [N _{DFT, th, k} (t, u)] contained in x _{DFT, k} (t, u) is To be calculated.

As described above, the average power of the thermal noise included in x _{DFT, k} (t, u) can be estimated by the method according to the first example or the second example.

  In the process of obtaining the estimated value of equation (311) according to the first method example or equation (325) according to the second method example, the quantization bit width is expanded from b by four arithmetic operations, but significant figures In consideration of the above, it is possible to keep the quantization level within a predetermined bit width by rounding the value by rounding off the redundant lower-order bits as appropriate in each calculation process.

From Expressions (302) and (303), the quantization level of the I-phase and Q-phase of x _{DFT, k} (t, u) is an integer expressed as a b-bit binary number. The power quantization level expressed in equations (311) and (325), which is in the same order as the sum of the square of Q and the square of Q phase, is rounded to an integer expressed as a binary number of 2b + 1 bits. And

  In the case of the first method example, even when an interference signal that is regarded as additional noise equivalent to thermal noise is generated for a desired signal due to an external factor of the receiver 1, the average of fluctuating thermal noise is generated. Since power can be estimated, it can be said that it is more advantageous than the second method example when a known repetitive sequence is included in the transmission signal.

Using the average power of the thermal noise estimated as described above, the signal value is corrected so that the average power of the thermal noise becomes equal to a predetermined value in all branches. That is, the predetermined value of the u-th carrier wave at time t is represented by N ^ _{NORM, th} (t, u) (where N ^ _{NORM, th} (t, u)> 0, that is, N ^ _{NORM, th} ( The quantization level of t, u) is a positive integer), and the signal from the thermal noise correction unit 14 to the quantization noise correction unit 15 that is a corrected value is x _{NORM, k} (t, u). If the I phase and Q phase are x _{NORM, I, k} (t, u) and x _{NORM, Q, k} (t, u), respectively, they are expressed as follows.

Here, if N ^ _{NORM, th} (t, u) is too large, x _{NORM in the} branch where E [N _{DFT, th, k} (t, u)] <N ^ _{NORM, th} (t, u) _{, I, k} (t, u) or x _{NORM, Q, k} (t, u) absolute value of the quantization level is x _{DFT, I, k} (t, u) or x _{DFT, Q, k} (t , u) must be larger than the absolute value of the quantization level, but does not fit in the quantization bit width of x _{DFT, I, k} (t, u) or x _{DFT, Q, k} (t, u) The quantization bit width of x _{NORM, I, k} (t, u) and x _{NORM, Q, k} (t, u) is x _{DFT, I, k} (t, u) or x _{DFT, Q, k} ( It is necessary to increase the quantization bit width of t, u) (and this leads to an increase in circuit scale). On the other hand, if N ^ _{NORM, th} (t, u) is too small, the branch where E [N _{DFT, th, k} (t, u)]> N ^ _{NORM, th} (t, u) is x _NORM, The absolute value of the quantization level of _{I, k} (t, u) or x _{NORM, Q, k} (t, u) is expressed as x _{DFT, I, k} (t, u) or x _{DFT, Q, k} (t, u However, if the absolute value of the quantized level after the decrease is too small (if the resolution is too small), the reception characteristics deteriorate. In addition, if N ^ _{NORM, th} (t, u) is too small, the estimation accuracy of E [N _{DFT, th, k} (t, u)] to be compared is increased (N ^ _{NORM, th} (t , u) is small and therefore needs to be accurately estimated to a smaller E [N _{DFT, th, k} (t, u)] value, which requires a sufficiently long time averaging and real-time Impairs sex. Therefore, it is necessary to set an appropriate value.

In the following, three examples are given as methods for determining the value of N ^ _{NORM, th} (t, u). The first example is E [N _{DFT, th, k} (t, u)] of the smallest E [N _{DFT, th, k} (t, u)] branch among all branches of the same carrier. It is a method to do. That is, it is expressed as follows.

N ^ _{NORM, th} (t, u)
= Min (E [N _{DFT, th, k} (t, 1)], E [N _{DFT, th, k} (t, 2)],…, E [N _{DFT, th, k} (t, L)]) ... (333-2)
_{Here, min (a 1, a 2} , ..., a n) is a _1, a _2, ..., it represents the minimum value of a _n. On the other hand, the second example is a method of setting a predetermined constant. That is, it is expressed as follows.

N ^ _{NORM, th} (t, u) = N _{NORM, const, th} (333-3)
Here, N _{NORM, const, th} is a constant that is not 0 (that is, the quantization level of N _{NORM, const, th} is a positive integer). The third example is a combination of the first and second examples, and the most of all branches of the same carrier to prevent N ^ _{NORM, th} (t, u) from becoming too small. E [N _{DFT, th, k} (t, u)] of a small branch of E [N _{DFT, th, k} (t, u)], but this value is smaller than a predetermined value N _{NORM, const, th} When it is small, it is a method of clipping to N _{NORM, const, th} . That is, it is expressed as follows.

N ^ _{NORM, th} (t, u)
= Max (min (E [N _{DFT, th, k} (t, 1)], E [N _{DFT, th, k} (t, 2)], ..., E [N _{DFT, th, k} (t, L) ]), N _{NORM, const, th} ) (333-4) In the expression (333-3) in the second example or the expression (333-4) in the third example, E [N _{DFT, In} a branch _{where th, k} (t, u)] <N ^ _{NORM, th} (t, u), according to equations (332) and (333), x _{NORM, I, k} (t, u), The absolute values of the quantization levels of x _{NORM, Q, k} (t, u) are the quantization levels of x _{DFT, I, k} (t, u), x _{DFT, Q, k} (t, u), respectively. Although the absolute value increases, the quantization noise also increases at this time. The reason is as follows. When the signal quantization levels 1, 2, 3, ... are each multiplied by a (an integer that satisfies a> 1), a, 2a, 3a, ... are obtained. This is equivalent to multiplying the signal quantization step q by a. Since the quantization noise in the signal is proportional to q as shown in equations (106) and (107), the value becomes a times and the quantization noise increases. Therefore, the average power of the quantization noise does not coincide with other branches (branch where E [N _{DFT, th, k} (t, u)] ≧ N ^ _{NORM, th} (t, u)). Therefore, in all branches of the u th carrier, x _{DFT, I, k} (t, t) so that E [N _{DFT, th, k} (t, u)] ≧ N ^ _{NORM, th} (t, u) It is desirable to make the quantization bit width of u), x _{DFT, Q, k} (t, u) sufficiently large (in this case, the expression (333-4) in the third example is the first example) In the first example, E [N _{DFT, th, k} (of the branch with the smallest E [N _{DFT, th, k} (t, u)] is _obtained. (t, u)] is set so that E [N _{DFT, th, k} (t, u)] ≧ N ^ _{NORM, th} (t, u) in all branches). Because the smaller the quantization bit width of x _{DFT, I, k} (t, u), x _{DFT, Q, k} (t, u), x _{DFT, I, k} (t, u), x _{DFT, Q , k} (t, u) needs to have a larger quantization step q, so the quantization level of E [N _{DFT, th, k} (t, u)] is smaller than N _{NORM, const, th} This is because the possibility of becoming smaller (E [N _{DFT, th, k} (t, u)] <q N _{NORM, const, th} ) increases.

As a next best measure, there is a branch where the quantization bit width cannot be sufficiently secured and E [N _{DFT, th, k} (t, u)] <N ^ _{NORM, th} (t, u) In this case, in order to prevent the quantization noise between the branches from becoming equal due to an increase in the quantization noise, the equations (331) to (333) are reviewed and expressed as follows.

If E [N _{DFT, th, k} (t, u)] ≧ N ^ _{NORM, th} (t, u),

If E [N _{DFT, th, k} (t, u)] <N ^ _{NORM, th} (t, u),
x _{NORM, I, k} (t, u) = x _{DFT, I, k} (t, u), ... (335)
x _{NORM, Q, k} (t, u) = x _{DFT, Q, k} (t, u) (336)
The above equation is such that the upper limit value of E [N _{DFT, th, k} (t, u)] for each branch is N ^ _{NORM, th} (t, u) (E [N _{DFT, th, k} ( t, u)]> N ^ _{NORM, th} (t, u) so that E [N _{DFT, th, k} (t, u)] becomes N ^ _{NORM, th} (t, u)) , X _{DFT, I, k} (t, u), x _{DFT, Q, k} (t, u) are corrected.

As an example of the value of N _{NORM, const, th} , if the required SNR (the minimum SNR required for the system that is less than the predetermined error rate) is sufficiently larger than 1, the average power of the quantization noise is also Is larger than N and is the minimum value that can ensure the estimation accuracy (that is, the minimum value of the quantization level by the quantization step q), and the quantization level of N _{NORM, const, th} is set to 1 (N _{NORM, const, (th} = q).

  As described above, the received signal was corrected so that the average power of the thermal noise became equal in all branches while maintaining the SNR.

When the SNR in the signal x _{NORM, k} (t, u) to the quantized noise correction unit 15 after correction is SNR _{NORM, k} (t, u), the average value E [SNR _{NORM, k} (t, u) ] Is expressed as follows from the formulas (305) to (308) and (331) to (333).

E [SNR _{NORM, k} (t, u)]
= E [S _{NORM, k} (t, u)] / {E [N _{NORM, th, k} (t, u)] + E [N _{NORM, qu, k} (t, u)]}, ... (341 )
E [S _{NORM, k} (t, u)] = E [S _{DFT, k} (t, u)] {N ^ _{NORM, th} (t, u) / E [N _{DFT, th, k} (t, u) ]}
= E [SNR _{DFT, k} (t, u)] / N ^ _{NORM, th} (t, u), ... (342)
E [N _{NORM, th, k} (t, u)] = E [N _{DFT, th, k} (t, u)] {N ^ _{NORM, th} (t, u) / E [N _{DFT, th, k} ( t, u)]}
＝ N ^ _{NORM, th} (t, u), ... (343)
_{E [N NORM, qu, k} (t, u)] = E [N DFT, qu, k (t, u)] = q 2/6 ··· (344)
Where S _{NORM, k} (t, u), N _{NORM, th, k} (t, u) and N _{NORM, qu, k} (t, u) are each x _{NORM, k} (t, u) Of the desired signal, thermal noise, and quantization noise. _{Further, x NORM, k (t,} u) quantization noise average value E _{[N NORM, qu, k (} t, u)] _{is, x NORM, k (t,} u) and x _{DFT, k} (t , u) and the quantization step q (both are integers expressed as b-bit binary numbers, the quantization step has not changed), so that x _{DFT, k} (t, u) It becomes equal to the average value of quantization noise E [N _{DFT, qu, k} (t, u)].

  As described above, the thermal noise correction unit 14 estimates the average power of the thermal noise, and corrects the signal value so that the average power of the thermal noise is equal to a predetermined value in all branches.

  Subsequently, the average power of the thermal noise included in the signals of the respective carriers modulated by the L carrier frequencies input to the quantization noise correction unit 15 is equal in all the branches. The normalized baseband signal has a predetermined quantization bit width so that the average power of the quantization noise included in the signal of each carrier is the same value in all branches so that the diversity operation is performed appropriately. And input to diversity combining sections 30-32.

Here, the input signal to the quantization noise correction unit 15 is the output signal x _{NORM, k} (t, u) of the thermal noise correction unit 14, and the quantization level is expressed as a b-bit binary number. It is an integer.

On the other hand, diversity combining units 30 to 32, which are subsequent parts, process integers expressed as B-bit binary numbers as quantization levels. That is, the output signal of the quantization noise correction section _{15 x CLIP, k (t,} u) When, x _CLIP, quantization levels _k (t, u) is an integer represented as a binary number of B bits There is a need.

If the quantization bit width b of the output signal of the DFT unit 13, which is the front part of the thermal noise correction unit 14, and the input signals of the diversity combining units 30 to 32, which are the subsequent part of the quantization noise correction unit 15. X _{CLIP, k} (t, u) may be set as follows if the quantization bit width B is equal to (ie, if b = B).

x _{CLIP, k} (t, u) = x _{NORM, k} (t, u) (351)
However, as described above, when frequency selective fading occurs and the power of the signal of each carrier is different, the resolution of the signal of each carrier is different.

  Therefore, when the total power is large, the power of the signal of each carrier is large (it is sufficiently larger than the power of thermal noise), but the resolution of the carrier with power smaller than the other carriers is small, so Quantization noise increases with power.

  In order to solve the above problem, the quantization bit width of the signal passing through the A / D conversion unit 108 and the part subsequent to the A / D conversion unit 108 is increased, that is, the values of b and B are increased. However, it is not desirable because it leads to an increase in circuit scale and an increase in manufacturing cost.

  Therefore, in order to increase the circuit scale to a minimum, only the quantization bit width of the signal passing through the part from the A / D conversion unit 108 to the DFT part 13 is increased, and the signal passing through the subsequent part is increased. It is desirable to suppress the quantization bit width. That is, although the quantization bit width b of the output signal of the DFT unit 13 which is a front part of the thermal noise correction unit 14 is increased, the inputs of the diversity combining units 30 to 32 which are subsequent parts of the thermal noise correction unit 14 It is desirable to maintain the quantization bit width B of the signal.

However, in this case, since b> B, it is necessary to limit the number of significant digits of the quantization level of the output signal of the DFT unit 13. That is, when the quantization level represented by the bit width b exceeds the range that can be expressed by B bits (when 2's complement expression is assumed, it is smaller than −2 ^B−1 which is the minimum value, or By increasing the quantization step and decreasing the absolute value of the quantization level (in order to maintain the true value, the quantization step is multiplied by a (if greater than the maximum value 2 ^B−1 −1)) a> 1) and by multiplying the quantization level by 1 / a), the signal is converted into an integer represented by B bits, and the quantization level of the signal from the quantization noise correcting unit 15 to the diversity combining units 30 to 32 is There is a need to. However, if the quantization step is different for each branch of the same carrier at this time, the average power of the quantization noise of each branch differs from Equation (113). Therefore, in order to perform diversity appropriately, in order for the average power of quantization noise of all branches to be equal, it is necessary to apply the same quantization step value to all branches of the same carrier. Therefore, the quantization step is changed so that the quantization level of the signal of the branch with the highest average power of the signal is within B bits, and the same quantization step is applied to the other branches (the highest average power of the signal is (If the quantization level of a branch signal falls within B bits, the quantization level of the other branch signals to which the same quantization step is applied always falls within B bits).

  In realizing the above, when b> B, processing is performed as follows.

  As described above, PSK (Phase Shift Keying) or QAM (Quadrature Phase Amplitude Modulation) and FSK (Frequency Shift Keying) used in normal mobile communication are assumed as the modulation method in each carrier wave.

That is, the amplitude is assumed to be constant (in reality, it is assumed that the amplitude of the received signal gradually changes due to channel fluctuation (fading) but does not change within a short time). Therefore, if the value of power that is the square of the amplitude, that is, the desired signal of the output signal x _{NORM, k} (t, u) of the thermal noise correction unit 14 is s _{NORM, k} (t, u), the power S _It is assumed that the average value E [S _{NORM, k} (t, u)] of _{NORM, k} (t, u) hardly changes.

Hereinafter, a method for estimating E [x _{NORM, k} (t, u)] will be described. Here are two examples.

In the first method example, E [S _{NORM, k} (t, u)] is first estimated. This process is performed by thermal noise included in x _{DFT, k} (t, u) in the thermal noise correction unit 14. E [N _{DFT, th, k} (t, u)] is similar to the estimation method, and some signals use the same signal, so the thermal noise correction unit 14 uses E [S _{NORM, k} (t, u)] is estimated, and the method is described below. The estimation method of E [S _{NORM, k} (t, u)] uses a repetition of a known transmission sequence. Usually, such a known transmission sequence is included in order to estimate the fluctuation of the received signal due to the communication path. Moreover, in order to remove the influence of thermal noise, it is often repeated. The first method example and the second method example of the received signal (symbol) of the known sequence of the carrier wave u at time t of the k-th branch of _{xDFT, k} (t, u) are described above. Like x _{known, 1, k} (t, u), x _{known, 2, k} (t + δ, u).

Then, these signals, which were the same symbol at the time of transmission, differ due to the addition of thermal noise, but the effect can be reduced by averaging, so x _{DFT, k} (t, u) The estimated value of the average power E [S _{DFT, k} (t, u)] of the desired signal included in is calculated as follows.

For example, in the case of the IEEE802.11a standard, since the long preamble is a repetition of a known transmission sequence, the average power of the desired signal can be estimated by applying the above formula. The estimated value of the average power E [S _{NORM, k} (t, u)] of the desired signal included in x _{NORM, k} (t, u) is E [S _{DFT, k} (t , u)] and E [N _{DFT, th, k} (t, u)] estimated by Equation (311) according to the first method or Equation (325) according to the second method. From the equations (342) to (343), the following is obtained.

E [| x _{NORM, k} (t, u) | ² ] = E [S _{NORM, k} (t, u)] + E [N _{NORM, th, k} (t, u)], ... (361-2 )
E [S _{NORM, k} (t, u)]
= E [S _{DFT, k} (t, u)] {N ^ _{NORM, th} (t, u) / E [N _{DFT, th, k} (t, u)]}, ... (361-3)
E [N _{NORM, th, k} (t, u)] = N ^ _{NORM, th} (t, u) (361-4)
Here, E [N _{NORM, th, k} (t, u)] is an average power of thermal noise included in x _{NORM, k} (t, u).

As described above, E [ _{NDFT, th, k} (t, u)] is an integer of 2b + 1 bits. Similarly, E [S _{DFT, k} (t, u)] is also rounded to zero by the most significant bit of the decimal part, and the calculation result is rounded to obtain a 2b + 1 bit integer, and E [| x _NORM, The calculation result of _k (t, u) | ² ] is also rounded to an integer of 2b + 1 bits.

The second method example is a method of estimating the average power E [| x _{NORM, k} (t, u) | ² ] of x _{NORM, k} (t, u) by integrating the signal for a predetermined time. It is. That is, it is calculated as follows.

Here, r is the number of signal samples in the integration interval, and T _S is the signal sampling period. As in the first method example, the calculation result is rounded to an integer of 2b + 1 bits.

To correct the quantization level of x _{NORM, k} (t, u) consisting of b bits to the quantization level of x _{CLIP, k} (t, u) consisting of B bits under the condition of b> B, 2b + 1 Bit average x _{NORM, k} (t, u) power E [| x _{NORM, k} (t, u) | ² ] is the average of 2B + 1 bit x _{CLIP, k} (t, u) The value E [| x _{CLIP, k} (t, u) | ² ] may be included. Accordingly, it is considered that the quantization level of E [| x _{NORM, k} (t, u) | ² ] is reduced so as to be within 2B + 1 bits.

Actually, since there is a slight fluctuation depending on the transmission path, it is accommodated with a little margin. Therefore, here, it is assumed that the quantization level of E [| x _{NORM, k} (t, u) | ² ] composed of 2b + 1 bits is reduced so as to be within 2B + 1-D bits. Here, D is an integer greater than or equal to 0 and less than 2B + 1 determined by the characteristics of the transmission path.

The quantization level of E [| x _{NORM, k} (t, u) | ² ] consisting of 2b + 1 bits, and the quantization level of E [| x _{CLIP, k} (t, u) | ² ] consisting of 2B + 1-D bits In order to reduce it to the above, the following calculation is performed.

(Quantization level of E [| x _{NORM, i} (t, u) | ² ])> 2 ^{2B + 1-D} ,
(Quantization level of E [| x _{CLIP, k} (t, u) | ² ])
= (Quantization level of E [| x _{NORM, k} (t, u) | ² ])
× {P ^ _CLIP (t, u) / (Quantization level of E [| x _{NORM, i} (t, u) | ² ])} (373)
(Quantization level of E [| x _{NORM, i} (t, u) | ² ]) ≦ 2 ^{2B + 1−D} ,
(Quantization level of E [| x _{CLIP, k} (t, u) | ² ])
= (Quantization level of E [| x _{NORM, k} (t, u) | ² ]) (374)
P ^ _CLIP (t, u) = 2 ^{2B + 1-D} (375)
E [| x _{NORM, i} (t, u) | ² ]
= Max (E [| x _{NORM, 1} (t, u) | ² ], E [| x _{NORM, 2} (t, u) | ² ],…, E [| x _{NORM, M} (t, u) | ² ]) ... (376)
Here, the result of the formula (373) is rounded by rounding or the like so as to be an integer. Note that here, (E [| x _{NORM, i} (t, u) | ² ] quantization level)> 2 ^{2B + 1−D} and (E [| x _{NORM, i} (t, u) | ² ] Quantization level) ≦ 2 ^{2B + 1−D} , the formula (373) and the formula (374) are used, but all [| x _{NORM, i} (t, u) Only the expression (373) may be applied to the range of the quantization level in [ ² ] (however, since errors due to rounding to an integer value are accumulated, cases are divided here).

In addition, formula _{(375) P ^ CLIP (t} , u) of the most _{E [| x NORM, k (} t, u) | 2] is a large branch of _{E [| x NORM, k (} t, u) | ² ], (quantization level of E [| x _{NORM, i} (t, u) | ² ]) = 2 ^{2B + 1-D} , but it only needs to be within 2B + 1-D bits Priority may be given to the ease of calculation, and a smaller value may be used. That is, from the quantization levels of x _{NORM, I, k} (t, u) and x _{NORM, Q, k} (t, u), x _{CLIP, I, k} (t, u) and x _{CLIP, Q, k} (t , u) can be reduced to the quantization level by determining P _CLIP (t, u) so that a bit shift operation can be performed. That is, (E [| x _{CLIP, k} (t, u) | ² ] quantization level) = (1/2 ^2ε ) (E [| x _{NORM, k} (t, u) | ² ] quantization level ) (Ε is an integer), (x _{CLIP, I, k} (t, u) quantization level) = (1/2 ^ε ) (x _{NORM, I, k} (t, u) quantization level ) And (quantization level of x _{CLIP, Q, k} (t, u)) = (1/2 ^ε ) (quantization level of x _{NORM, Q, k} (t, u)), so only ε bits It is only necessary to shift right (shift in the lower bit direction). As an example, P ^ _CLIP (t, u) may be determined as in the following equation.

P ^ _CLIP (t, u)
^{= (2 2B + 1-D} / 2 ε) (E [| x NORM, i (t, u) | 2] quantization levels) (375-2)
Here, ε is a maximum integer satisfying the following.

^{(2 2B + 1-D /} 2 ε) (E [| x NORM, i (t, u) | 2] quantization ^{levels) ≦ 2 2B + 1-D} ··· (375-3)
Equation (373) - (376) _{is, E [| x NORM, k} (t, u) | 2] is the largest branch _{E [| x NORM, k (} t, u) | 2] quantization levels , E [| x _{CLIP, k} (t, u) | ² ] is clipped to 2 ^{2B + 1−D} which is the maximum quantization level of E [| x _{CLIP, k} (t, u) | ² ] The quantization bit width is set to 2 ^{2B + 1-D} or less.

Therefore, in order to reduce the quantization level of x _{NORM, k} (t, u) consisting of b bits to the quantization level of x _{CLIP, k} (t, u) consisting of B bits, equations (373) to ( The following calculation may be performed using the quantization level of E [| x _{CLIP, k} (t, u) | ² ] calculated in 376).

Here, E [| x _{NORM, k} (t, u) | ² ] in the equations (373) to (376) is the same as the equations (361-2), (361-3) and (361 -4) or equation (362). Note that the update timing of the estimated value of E [| x _{NORM, k} (t, u) | ² ] depends on the format of the received frame, but as an example, it is the same from the start to the end of reception of one frame A method of using a value, that is, a method of estimating and updating E [| x _{NORM, k} (t, u) | ² ] at the beginning of the received frame and using the same value until reception of this frame is considered. This is because, normally, all signals in the same frame have equal weight, and therefore it is desirable that the quantization steps of all signals in the same frame are equal.

It is assumed that the calculation result (quantization level) of Expression (381) is rounded to an integer represented as a B-bit binary number by rounding off to zero or the like in the calculation process, as described above. Also, when the range that can be expressed by B bits is exceeded, that is, when 2's complement expression is assumed, it is smaller than −2 ^B−1 which is the minimum value, or 2 ^B−1 − which is the maximum value. If it is greater than 1, it is clipped to the minimum and maximum values, respectively.

  As described above, the average power of the quantization noise included in each carrier signal is limited to a predetermined quantization bit width so that all branches have the same value.

When the SNR in the signal x _{CLIP, k} (t, u) to the diversity combining sections 30 to 32 after correction is SNR _{CLIP, k} (t, u), the average value E [SNR _{CLIP, k} (t, u) ] Is expressed as follows from the formulas (373) to (376) and (341) to (344).

E [SNR _{CLIP, k} (t, u)] = E [S _{CLIP, k} (t, u)] / {E [N _{CLIP, th, k} (t, u)] + E [N _{CLIP, qu, k} ( t, u)]},
... (391)
E [S _{CLIP, k} (t, u)] = E [S _{NORM, k} (t, u)], ... (392)
E [N _{CLIP, th, k} (t, u)] = E [N _{NORM, th, k} (t, u)], ... (393)
E [| x _{NORM, i} (t, u) | ² ] / q> 2 ^{2B + 1-D}
_{E [N CLIP, qu, k} (t, u)] = q clip 2/6, ··· (394)
q _clip = {E [| x _{NORM, i} (t, u) | ² ] / P ^ _CLIP (t, u)} q, ... (395)
E [| x _{NORM, i} (t, u) | ² ] / q ≦ 2 ^{2B + 1−D}
E [N _{CLIP, qu, k} (t, u)] = E [N _{NORM, qu, k} (t, u)] = q / 6 (396)
Where S _{CLIP, k} (t, u), N _{CLIP, th, k} (t, u), N _{CLIP, qu, k} (t, u) are each x _{CLIP, k} (t, u) Of the desired signal, thermal noise, and quantization noise. P ^ _CLIP (t, u), E [| _{xNORM, i} (t, u) | ² ] are as shown in the equations (375) and (376) already described.

  As shown in the equations (391) to (396), the quantization bit width of the signals of the diversity combining units 30 to 32, which is the subsequent stage of the thermal noise correction unit 14, is limited to B, and the quantum By increasing the quantization step so that the quantization bit width is within B only for signals whose quantization level cannot be expressed in B-bit binary numbers (not within the range of B-bit binary numbers) The quantization noise is clipped to the lower limit when the quantization bit width is B, and the absolute value of the quantization level is decreased by the increase of the quantization step, so that there is no change in the power of the desired signal and the thermal noise. (By multiplying the quantization step by a and multiplying the quantization level by 1 / a, the true values of the desired signal and thermal noise power, which are the products, do not change.)

  As described above, the quantization noise correction unit 15 restricts the average power of the quantization noise included in each carrier signal to a predetermined quantization bit width so that all the branches have the same value. It was.

  Subsequently, the diversity combining unit 30 combines signals from two branches having the same carrier frequency (the same subcarrier number) by a predetermined method and outputs the combined signals to the P / S conversion unit 33 as one signal.

Here, the input signal to the diversity combining unit 30 is the output signal x _{CLIP, k} (t, u) of the quantization noise correction units 15 and 25, and the quantization level is represented as a B-bit binary number. It is an integer. x _{CLIP, k} (t, u) is selected by the diversity combining unit 30 because the average power of the thermal noise and the average power of the quantization noise are equalized in all diversity branches from the above procedure. Is replaced by x _{CLIP, k} (t, u) → x _k (t, u), equations (7) and (8) are applied, and y (t, u) is converted to diversity combining unit 30. To the P / S conversion unit 33.

On the other hand, when the diversity combining unit 30 performs maximum ratio combining, the equations (9) and (10) are similarly applied by replacing x _{CLIP, k} (t, u) → x _k (t, u). Then, y (t, u) may be output from the diversity combining unit 30 to the P / S conversion unit 33.

  Since the operation of the thermal noise correction unit 24 is the same as that of the thermal noise correction unit 14, the description thereof is omitted. Further, the operation of the quantization noise correction unit 25 is the same as that of the quantization noise correction unit 15, and thus the description thereof is omitted.

  Since the operations of the diversity combining units 31 and 32 are the same as those of the diversity combining unit 30, the description thereof is omitted.

(A-3) Effect of First Embodiment As described above, according to the first embodiment, the receiver 1A adopting the OFDM scheme requires a digital circuit that realizes DFT, and performs A / D conversion. The generation of quantization noise due to the resolution limit is inevitable, but even if the level of quantization noise varies from diversity branch to diversity branch, diversity combining (here, selective combining or maximum ratio combining method) can operate properly. The effect that it is is acquired.

(B) Second Embodiment Next, a second embodiment of the diversity receiver of the present invention will be described in detail with reference to the drawings.

(B-1) Configuration of Second Embodiment FIG. 5 is an internal configuration diagram illustrating an internal configuration of a receiver according to the second embodiment. In FIG. 5, the receiver 1B of the second embodiment includes an antenna 10, a demodulator 11, an S / P converter 12, a DFT unit 13, a thermal noise corrector 14, an antenna 20, a demodulator 21, and an S / P converter. Unit 22, DFT unit 23, thermal noise correction unit 24, diversity combining unit 30, diversity combining unit 31, diversity combining unit 32, P / S conversion unit 33, and determination unit 34.

  The receiver 1B of the second embodiment is obtained by partially changing the internal configuration of the receiver 9 of FIG. Further, the demodulation units 11 and 21 have the internal configuration of FIG.

  Specifically, the thermal noise correction unit 14 is newly added between the output of the DFT unit 13 and the input to the diversity combining units 30 to 32, and the output of the DFT unit 23 and the diversity combining units 30 to 30 are added. 2 is different from FIG. 2 in that a thermal noise correction unit 24 is newly added between the input to the input 32.

  The thermal noise correction unit 14 receives from the DFT unit 13 the baseband signal that has been modulated by each of the L carrier frequencies, and the average power of the thermal noise included in each carrier signal is equal in all branches. It normalizes to become. Further, the thermal noise correction unit 14 provides baseband signals normalized to the same value in all branches to the diversity combining units 30 to 32.

  Since the thermal noise correction unit 24 is the same as the thermal noise correction unit 14, a description thereof is omitted here.

  The other components are the same as those in FIG.

(B-2) Operation of Second Embodiment Since the description of the operation of the thermal noise correction unit 14 is the same as that of the first embodiment, the description is omitted.

  In the second embodiment, the quantization bit width b of the signal after the output of the DFT unit 13, which is the front part of the thermal noise correction unit 14, and the diversity combining unit, which is the subsequent part of the thermal noise correction unit 14 This is applicable when the quantization bit width B of signals 30 to 32 matches.

The input signal to the diversity combining unit 30 is the output signal x _{NORM, k} (t, u) of the thermal noise correction units 14 and 24, and the quantization level is expressed as a binary number of B = b bits. It is an integer.

Here, x _{NORM, k} (t, u) is the same value for all the diversity branches from the above procedure, and the average power of the quantization noise is already equal to all the diversity powers. When the diversity combining unit 30 performs selective combining because the branches are set to the same value, x _{NORM, k} (t, u) → x _k (t, u) is replaced with the expression (7), Applying (8), y (t, u) may be output from the diversity combining unit 30 to the P / S conversion unit 33.

On the other hand, when the diversity combining unit 30 performs the maximum ratio combining, the equations (9) and (10) are similarly applied by replacing x _{NORM, k} (t, u) → x _k (t, u). Then, y (t, u) may be output from the diversity combining unit 30 to the P / S conversion unit 33.

  Since the operation of the thermal noise correction unit 24 is the same as that of the thermal noise correction unit 14, the description thereof is omitted. In addition, the operations of the diversity combining units 31 and 32 are the same as those of the diversity combining unit 30, and thus description thereof is omitted.

  Other operations are also the same as in the conventional case, and thus the description thereof is omitted.

(B-3) Effect of Second Embodiment As described above, according to the second embodiment, a receiver employing the OFDM scheme requires a digital circuit that realizes DFT, and in A / D conversion Generation of quantization noise due to resolution limitations is inevitable, but even if the level of quantization noise varies from diversity branch to diversity branch, diversity combining (here, selective combining or maximum ratio combining method) can operate properly. The effect that there is.

  In the second embodiment, the quantization bit width b of the signal after the output of the DFT unit 13, which is the front part of the thermal noise correction unit 14, and the diversity combining unit, which is the subsequent part of the thermal noise correction unit 14 It is necessary to match the quantization bit width B of signals 30 to 32.

  When frequency selective fading occurs and the power of the signal of each carrier is different, the resolution of the signal of each carrier is different, but when the quantization bit width of the signal of each carrier is increased, the first As in the embodiment, since sufficient resolution can be obtained, it is possible to suppress the power of the quantization noise to a sufficiently small value. However, the circuit scale increases as compared with the first embodiment.

  On the other hand, when the quantization bit widths of these signals are reduced, the circuit scale can be suppressed as in the first embodiment, but frequency selective fading occurs, and the power of the signal of each carrier wave is reduced. However, since the resolution of the signal of each carrier is different, sufficient resolution cannot be obtained, so that the power of the quantization noise is increased as compared with the first embodiment.

(C) Third Embodiment Next, a third embodiment of the diversity receiver of the present invention will be described in detail with reference to the drawings.

(C-1) Configuration of Third Embodiment FIG. 6 is an internal configuration diagram illustrating an internal configuration of a receiver according to the third embodiment. In FIG. 6, the receiver 1C of the third embodiment includes an antenna 10, a demodulator 11, an S / P converter 12, a DFT unit 13, a quantization noise corrector 15, an antenna 20, a demodulator 21, and an S / P. A conversion unit 22, a DFT unit 23, a quantization noise correction unit 25, a diversity combining unit 30, a diversity combining unit 31, a diversity combining unit 32, a P / S conversion unit 33, and a determination unit 34 are included.

  The receiver 1C of the third embodiment is obtained by partially changing the internal configuration of the receiver 9 of FIG. Further, the demodulation units 11 and 21 have the internal configuration of FIG.

  Specifically, the quantization noise correction unit 15 is newly added between the output of the DFT unit 13 and the inputs to the diversity combining units 30 to 32. The output of the DFT unit 23 and the diversity combining unit are The point that the quantization noise correction | amendment part 25 is newly added between the inputs to 30-32 differs from FIG.

  The quantization noise correction unit 15 receives from the DFT unit 13 the baseband signal that has been modulated by each of the L carrier frequencies, and calculates the average power of the quantization noise included in the signal of each carrier in all branches. This is limited to a predetermined quantization bit width so as to be equal, and is given to diversity combining sections 30-32.

  Since the quantization noise correction unit 25 is the same as the quantization noise correction unit 15, the description thereof is omitted.

  Other configurations are the same as those in FIG.

(C-2) Operation of Third Embodiment The description of the operation of the quantization noise correction unit 15 is the same as that of the first embodiment, and thus the description thereof is omitted.

An input signal to the quantization noise correction unit 15 is a signal x _{DFT, k} (t, u) from the DFT unit 13. In addition, the quantization noise correction unit _{15, x DFT, k (t,} u) the average value E of the power of _{[| x DFT, k (t} , u) | 2] is a function for estimating a first It is assumed that it is provided instead of the thermal noise correction unit 14 in the embodiment.

_{x DFT, k (t, u} ) of the electric power mean value E _{[| x DFT, k (t} , u) | 2] with respect to an example of a method of estimating a is by thermal noise correction unit 14 of the first embodiment Similar to the two example methods.

  Since the operation of the quantization noise correction unit 25 is the same as that of the quantization noise correction unit 15, the description thereof is omitted.

  Since other operations are the same as those in the conventional case, the description thereof is omitted.

(C-3) Effect of Third Embodiment As described above, according to the third embodiment, a receiver adopting the OFDM scheme requires a digital circuit that realizes DFT, and in A / D conversion Generation of quantization noise due to resolution limitations is inevitable, but even if the level of quantization noise varies from diversity branch to diversity branch, diversity combining (here, selective combining or maximum ratio combining method) can operate properly. The effect that there is.

  In the third embodiment, since the average power of thermal noise is not equalized in all diversity branches from the above procedure, the received power at the antenna is sufficiently larger than the average power of thermal noise. When the average power of the thermal noise is negligible (the average power of the thermal noise is sufficiently smaller than the upper limit value clipped in the first and second embodiments, the thermal noise correcting units 14 and 24 cause the average power of the thermal noise to be ignored. Is effective only when it is not necessary to make the same value in all diversity branches.

(D) Other Embodiments In the first to third embodiments described above, an example of a demodulating unit in which an analog circuit and a digital circuit are mixed is shown. -It is not limited to the thing of 3rd Embodiment.

  For example, the majority may be constituted by a digital circuit, and conversely, the majority may be constituted by an analog circuit. Further, not only the demodulator but also most of the diversity receivers may be configured with digital circuits, and conversely, may be configured with analog circuits. Furthermore, the part constituted by the digital circuit may be realized by a CPU and a program executed by the CPU.

1A-1C, 9 ... receiver,
DESCRIPTION OF SYMBOLS 10 ... Antenna, 11 ... Demodulation part, 12 ... S / P (serial parallel) conversion part, 13 ... DFT (discrete Fourier transform) part, 14 ... Thermal noise correction part, 15 ... Quantization noise correction part,
DESCRIPTION OF SYMBOLS 20 ... Antenna, 21 ... Demodulation part, 22 ... S / P conversion part, 23 ... DFT part, 24 ... Thermal noise correction part, 25 ... Quantization noise correction part,
DESCRIPTION OF SYMBOLS 30 ... Diversity combining part, 31 ... Diversity combining part, 32 ... Diversity combining part, 33 ... P / S (parallel serial) conversion part, 34 ... Determination part

Claims (10)

The radio wave acquisition position is different, and a plurality of branch processing means for demodulating a reception signal modulated by a plurality of carrier frequencies and one of the reception signals from each of the branch processing means are selected. Diversity receiver equipped with diversity combining means for outputting
Each of the plurality of branch processing means is
The average power of the thermal noise included in the baseband received signal modulated by each carrier frequency separated by the discrete Fourier transform is estimated, and the average power of the thermal noise is equal to all the other branch processing means. A thermal noise correction unit for correcting the value of the baseband reception signal so as to be a predetermined value;
The average power of the baseband reception signal is estimated so that the quantization bit width of the baseband reception signal corrected by the thermal noise correction unit becomes a predetermined quantization bit width smaller than the quantization bit width. When the estimated average power of the baseband received signal is larger than the quantized bit width after reduction, a quantum that corrects the baseband received signal so that the power falls within the quantized bit width. And a noise correction unit,
The diversity receiver characterized in that the diversity combining means performs diversity combining of the baseband received signal corrected by the quantization noise correcting unit of each branch processing means. The radio wave acquisition position is different, and a plurality of branch processing means for demodulating a modulated signal modulated by a plurality of carrier frequencies and a received signal from each of the branch processing means are selected. Diversity receiver equipped with diversity combining means for outputting
Each of the plurality of branch processing means is
The average power of the thermal noise included in the baseband received signal modulated by each carrier frequency separated by the discrete Fourier transform is estimated for each carrier frequency, and the average power of the thermal noise is calculated for all the other branches. A thermal noise correction unit that corrects the value of the baseband received signal for each carrier frequency so as to have a predetermined value equal to the average power of the thermal noise included in the baseband received signal modulated by the same carrier frequency of the processing means. With
The diversity receiver, wherein the diversity combining means performs diversity combining of the baseband received signal corrected by the thermal noise correction unit of each branch processing means. The radio wave acquisition position is different, and a plurality of branch processing means for demodulating a modulated signal modulated by a plurality of carrier frequencies and a received signal from each of the branch processing means are selected. Diversity receiver equipped with diversity combining means for outputting
Each of the plurality of branch processing means is
The quantization bit width of the quantized baseband reception signal modulated by each carrier frequency separated by the discrete Fourier transform is set to a predetermined quantization bit width smaller than the quantization bit width. When the average power of the baseband received signal is estimated, and the estimated average power of the baseband received signal is larger than the quantization bit width, the power is within the quantized bit width after reduction. Comprises a quantization noise correction unit for correcting the baseband received signal,
The diversity receiver characterized in that the diversity combining means performs diversity combining of the baseband received signal corrected by the quantization noise correcting unit of each branch processing means.   The thermal noise correction unit obtains a variance value of the repeated signal sequence based on repetition of a known signal sequence included in the baseband received signal, thereby separating each of the above-described discrete Fourier transforms. The diversity receiver according to claim 1, wherein an average power of thermal noise included in a quantized baseband reception signal modulated by a carrier frequency is estimated.   The thermal noise correction unit estimates the power of the received signal at the antenna based on the gain value set in the amplifying means for amplifying the received signal of the predetermined part received by the antenna, and the estimated antenna Based on the ratio between the power of the received signal and the known average power of the thermal noise of the diversity receiver as a whole, it is quantized and modulated by each carrier frequency, separated by a discrete Fourier transform. The diversity receiver according to claim 1, wherein an average power of thermal noise included in the baseband received signal is estimated.   The quantization noise correction unit obtains an average value of the repeated signal series based on repetition of a known signal series included in the baseband received signal, and thereby separates each discrete Fourier transform. The diversity receiver according to claim 1, wherein an average power of a desired signal included in a quantized baseband reception signal modulated by a carrier frequency is estimated.   The said quantization noise correction | amendment part estimates the average electric power of the said baseband received signal by integrating the said baseband received signal for predetermined time, and calculating | requiring an average value. Diversity receiver. The predetermined value equal to all of the other branch processing means corrected by the thermal noise correction unit is equal to 1 which is the minimum unit of quantized values. The diversity receiver according to any one of 4, 4 and 5.   9. The diversity receiver according to claim 1, wherein the diversity combining unit performs diversity combining by selective combining.   9. The diversity receiver according to claim 1, wherein the diversity combining unit performs diversity combining by maximum ratio combining.

Images