JP2002111626A - Ofdm demodulator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce computational complexity required for decoding in a Viterbi decoder, and to obtain excellent decoding characteristics. SOLUTION: An input to the Viterbi decoder is controlled using a preamble for estimating a propagation path attached to an OFDM packet as an index. That is, an effective bit section is selected fluidly and properly from a value computed for the input to the Viterbi decoder by the power of the preamble for estimating the propagation path.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an OFDM (Orthogonal Frequency D) used for a digital radio communication system.
The present invention relates to a demodulator for an ivision multiplexing (orthogonal frequency division multiplexing) signal, and more particularly to a technique for controlling an input to a Viterbi decoder for decoding a convolutional code in an OFDM demodulator.

[0002]

2. Description of the Related Art OFDM modulation uses a plurality of subcarriers in the frequency domain, each of which has 16 QAM (Quad).
(rature Amplitude Modulation) A method of transmitting a time-domain signal waveform obtained by performing Fourier transform on a signal subjected to modulation or the like. An OFDM signal can be considered as a composite wave of a plurality of modulated waves, and the ratio of the peak amplitude to the average amplitude increases as the number of subcarriers increases.

[0003] In a wireless system such as a wireless LAN (Local Area Network) or the like, the receiver is usually assumed to move, so that the distance between the transmitter and the receiver fluctuates arbitrarily. Therefore, the reception level at the receiver differs depending on whether the distance from the transmitter is long or short. This reception level is used for the A / D converter (Analog-Digit
al Converter) must be within the dynamic range, and an AGC (Auto Gain Control) circuit is used to equalize the reception level of each packet.

In an AGC circuit, the amplitude of a received signal is A
To adjust within the dynamic range of the / D converter, a preamble signal for the AGC circuit is transmitted at the head of the packet as shown in FIG. The AGC circuit performs amplification gain control based on the reception level of the preamble signal, and equalizes the reception level of each packet.

[0005] In the OFDM modulated radio system, the above-mentioned two features, namely, 1) the ratio between the peak amplitude and the average amplitude is increased due to the use of a plurality of subcarriers.
2) Since the reception level fluctuates because the distance between the transmitter and the receiver fluctuates, amplitude fluctuation generally increases, and a deviation occurs in equalization of the reception level of each packet.

On the other hand, the OF convolutionally coded by the transmitter
The DM signal is decoded at the receiver using a Viterbi decoder. In a wireless system, the propagation path between the transmitter and the receiver is not uniquely determined, and the receiver receives multi-path fading because it receives a superposition of signals passing through several different propagation paths. For this reason, each subcarrier constituting the OFDM signal receives different propagation path characteristics, and the received power differs depending on each subcarrier.

In the decoding process of an OFDM signal using a Viterbi decoder, the OFDM signal of each subcarrier is normalized by a propagation path estimation preamble attached to a received packet,
The signal decoded from the constellation (signal arrangement) as shown in FIG. 2 is weighted by the power of the corresponding subcarrier of the preamble for propagation path estimation and input to the Viterbi decoder. The input to the Viterbi decoder is defined as the likelihood of the weighted signal magnitude, and the soft decision for giving flexibility to Viterbi decoding is provided with a certain threshold value, and is clearly defined as “0” or “1”. This is more effective than the hard decision of decoding after distinguishing. Generally, in Viterbi decoding by soft decision, the more the number of quantization bits used for soft decision, the closer to ideal decoding can be, and when weighting by subcarrier power as in the present system, the effect is remarkable. In maximum likelihood decoding represented by Viterbi decoding, several types of expected sequences are prepared as decoding sequence candidates, the likelihood of each sequence is calculated, and then the one with the highest likelihood is decoded. Determined as a series.

In the demodulation circuit, the received OFDM signal is converted into a digital signal by an A / D converter at a stage subsequent to the AGC circuit, and thereafter, a value is shown because an addition operation, a multiplication operation and the like are performed digitally. Use more bits. Therefore, every time these operations are performed, it is necessary to cut off the signal bits by rounding the value, etc., and to pass only the specified bits to the next stage circuit. In particular, the input to the Viterbi decoder is required. Is important because it directly affects the signal error rate of the system. That is, it is necessary to effectively select an appropriate signal bit from a signal whose number of bits has expanded by various demodulation processing operations and input the signal bit to the Viterbi decoder.

FIG. 3 shows a configuration example of a weighting and bit selection circuit for a conventional Viterbi decoder. The input to the weighting and bit selection circuit 120 is
After converting the signal a101 whose reception level is equalized by the function of the AGC circuit 101 into a digital signal a102 by the A / D converter 102, the processing circuit 103 performs operations such as frequency offset compensation, propagation path compensation, and phase noise removal. Are the propagation path estimation preamble signal al06 and the OFDM data signal al03.

The information of each subcarrier included in the OFDM data signal al03 is normalized by the amplitude of the corresponding subcarrier of the propagation path estimation preamble signal al06 in the normalization circuit 104, and the normalized signal al
04 is decoded from the constellation by the constellation decomposition circuit 105. On the other hand, in the propagation path estimation preamble signal al06, the power value signal al07 of each subcarrier is calculated by the squaring circuit al06. The signal al05 decoded from the constellation is weighted by the power al07 of the corresponding subcarrier of the propagation path estimation preamble in the weighting circuit 108, and the weighted signal al08 is output from the weighting circuit 108 to the lower bit cutout circuit 109. Is done. When the input bit to the Viterbi decoder 110 is N ₀ , the lower bit cutout circuit 109 outputs the weighted signal al0
Only _{N 0} bits from the MSB (Most Significant Bit) of the eight inputs to a Viterbi decoder 110 as valid data signals AL09, truncating the remaining lower bits. This selection of valid bits remains fixed no matter what signal is input.

[0011]

In a wireless communication system, it is usually assumed that a receiver operates.
The distance between the transmitter and the receiver varies arbitrarily. Therefore, the reception level of each user is different depending on whether the distance from the transmitter is long or short. It is necessary to keep this reception level within the dynamic range of the A / D converter used, and an AGC circuit for performing this control adjustment is incorporated in the wireless system. However, in the OFDM modulation, which is a modulation method in which amplitude fluctuations are drastic, it is particularly difficult to keep the reception level of different packets constant, and a deviation occurs, so that a function for compensating the performance of AGC is required.

On the other hand, OF signals convolutionally encoded by the transmitter
In the decoding process in the Viterbi decoder that decodes the DM signal, the information placed on each subcarrier of the OFDM data signal is normalized by the amplitude of the corresponding subcarrier of the propagation path estimation preamble, and then decoded from the constellation. Then, weighting is performed with the reception power of the corresponding subcarrier of the propagation path estimation preamble. Using this weighting as a likelihood, soft decision for giving flexibility to decoding is effective, and is also applied to Viterbi decoding. As the number of quantization bits allocated to this likelihood increases, the weighting becomes more continuous. It can be given as a value, and decoding close to the ideal is possible.

However, in maximum likelihood decoding, several types of expected sequences are prepared as decoding sequence candidates, and the likelihood of each sequence is calculated. Is determined. In these processes, as the number of quantization bits increases, the amount of calculation becomes enormous, and there is a problem that the scale of hardware is enlarged. For this reason,
It is necessary to limit the number of quantization bits so that the decoding performance does not suddenly deteriorate, and it is necessary to effectively select a bit at an appropriate position. When an appropriate bit is not selected, that is, the weighted signal a
The number of bits determined from the MSB in l08, for example, 6
Even if the bits are fixed and output to the Viterbi decoder, depending on the size of the signal al08, only 3 bits out of 6 bits may actually contribute to Viterbi decoding effectively. The problem is that they are not used effectively. Therefore, it is necessary to appropriately select and extract effective bits from the weighted signal al08 as appropriate based on some signal.

The present invention provides a method for solving these problems. An object of the present invention is to reduce the amount of calculation required for decoding by a Viterbi decoder and obtain good decoding characteristics. Therefore, an object of the present invention is to provide an effective selection method of input bits to a Viterbi decoder. Another object of the present invention is to perform equivalent reception level gain control by using the proposed method to assist the function of the provided AGC circuit.

[0015]

To achieve the above object, the present invention uses the following means. The OFDM signal packet, as shown in FIG.
A preamble used for M signal detection and AGC and diversity selection and coarse frequency offset estimation and time synchronization (referred to herein as an AGC preamble), and used for propagation path estimation and fine frequency offset estimation Preamble (this is referred to as a propagation path estimation preamble in this specification)
And a data part.

The OFDM signal is supplied to the AGC circuit by the AGC circuit.
The gain control is performed so that the output signal level becomes constant according to the reception level of the C preamble signal, and the A / D
The analog signal is converted into a digital signal by a converter, frequency offset is compensated by preamble information, converted to a frequency domain signal by fast Fourier transform, propagation path compensated by propagation path estimation preamble information, and phase noise is removed by a pilot signal. You.

The OFDM signal that has undergone the above operation is decoded from a constellation according to the modulation scheme, and the signal decoded from the constellation of each subcarrier is input to a Viterbi decoder. According to the assumption that the path fading characteristic is constant, the propagation path characteristics received by the preamble section and the data section are equivalently equal, and only the amplitude of each corresponding subcarrier of the propagation path estimation preamble indicating the propagation path characteristic is calculated. Multiply and equivalently weight the data portion with the received power of each subcarrier. Next, the power average value of all the subcarriers of the propagation path estimation preamble is approximately calculated, and the absolute value of the signal weighted by the power is rounded by the power average value of all the subcarriers of the propagation path estimation preamble. In this manner, the number of quantization bits is reduced, and the number of quantization bits is further reduced by appropriately discarding lower bits. The power average value of the propagation path estimation preamble is valid only while the packet including the preamble is being processed by the demodulation circuit, and is updated accordingly when a new preamble is input.

[0018]

Embodiments of the present invention will be described below in detail with reference to the drawings. According to the present invention, an OFDM signal whose gain is controlled by an AGC circuit in a receiver is converted into a Viterbi decoder for OFDM data by using the average power of all subcarriers of a propagation path estimation preamble added to the head of an OFDM packet as an index. Control the input bits to

A weighting and bit selection circuit for a Viterbi decoder according to one embodiment of the present invention will be described with reference to FIG.

The input to the weighting and bit selection circuit 220 is to convert the signal a201, whose reception level has been equalized by the function of the AGC circuit 201, into the A / D converter 20
After the conversion into a digital signal a202 by the processing circuit 203, the processing circuit 203 performs operations such as frequency offset compensation, propagation path compensation, and phase noise removal, and the propagation path estimation preamble signal a206 and the OFDM data signal a2
03.

The information of each subcarrier included in the OFDM data signal a203 is normalized by the amplitude of the corresponding subcarrier of the propagation path estimation preamble signal a206 in the normalization circuit 204, and the normalized signal a2
04 is decoded from the constellation by the constellation decomposition circuit 205. On the other hand, for the propagation path estimation preamble signal a206, the power value of each subcarrier is calculated by the squaring circuit 206 and output as a signal a207. Signal a205 decoded from constellation
Information of each subcarrier is weighted by the weighting circuit 207 with a207 indicating the power value of the corresponding subcarrier of the propagation path estimation preamble, and the weighted signal a210 is output to the comparator and rounding circuit 20.
9 is output. The a207 indicating the power value of each subcarrier of the preamble for estimating the propagation path is also input to the averaging circuit 208, and the averaging circuit 208 calculates the average power value for one subcarrier, and obtains the average power signal a2.
09 is output to the comparator and rounding circuit 209. The comparator and rounding circuit 209 rounds the weighted signal a210 with the average power signal a209.
If it is larger, the value of the signal a209 is output and the signal a209 is output.
When the size of the signal 210 is smaller than the signal a209, the value of the signal a210 is output as it is.

A signal a211 rounded by the power average value a209 is input to the upper bit cutout circuit 210, and a bit at the same position as the upper bit that is unnecessary for expressing the power average value a209 is signaled. a211 and output as signal a212, with the proviso that
When electric power mean value a209 is expressed in the number of input bits _{N 0} to the Viterbi decoder 212, of the signal a 211 L
Viterbi decoder 21 from SB (Least Significant Bit)
The number of bits equal to the number of input bits N ₀ to 2 is extracted and output as a signal a212. In the lower bit cut circuit 211, so that the number of input bits N ₀ to the Viterbi decoder 212 has been determined, it does truncating the lower bits, the Viterbi decoder the signal a213 21
Enter 2

The weighting and bit selection circuit 220 will be described with reference to specific examples, including numerical values and formats, and the flow of demodulation in a transmitter and a receiver. The number of subcarriers of the OFDM signal N_sc = 52 (including four pilot carriers), the FFT (FastFourier Transform; Fast Fourier Transform) length N_fft = 64, and each subcarrier is:
16QAM modulation is performed according to the constellation of FIG. 2, and the signal is represented in two's complement notation hereinafter. As shown in FIG. 1, an OFDM packet to be transmitted is prepended with an AGC preamble signal and a propagation path estimation preamble signal, and the remaining portion is composed of an OFDM data signal. These signal sequences are convolutionally coded and transmitted. I do.

The number of OFDM symbols to be assigned to the preamble for AGC and the preamble for propagation path estimation are N_sp = 1 and N_lp = 1, respectively. The preamble for propagation path estimation is SC- _{26... 26} = ｛l, 1 at the transmitter. , 1,1,1, -1,1, -1, -1, -1,1,1, -1,1,
-1,1, -1,1,1, -1, -1,1,1, -1,1,1,1,0,1, -1, -1,1,1, -1,1,
-1, -1,1,1, -1,1, -1, -1,1,1, -1, -1, -1, -1, -1,1, -1,1,1,1｝ For example, The known BPSK information that is proportional to the value of the subcarrier is placed on each subcarrier.

In the data portion, each subcarrier is 16QAM-modulated, and information is carried. FFT is performed for each OFDM symbol, and an OFC including an AGC preamble, a propagation path estimation preamble, and OFDM data
Construct and transmit M packets.

The OFDM signal undergoes different amplitude modulation and phase rotation for each subcarrier in the propagation path. The propagation path characteristic is represented by α (i) (i = 1, 2,..., 52; subcarrier number). And In addition, when the receiver moves, the reception level at the receiver differs depending on time and packet.

The AGC circuit 201 equalizes the reception level of each packet. This operation is performed based on the reception level of the AGC preamble added to the head of each packet, and the A / D conversion is performed. The signal is converted into a digital signal by the device 202. The processing circuit 203 performs frequency offset compensation using the AGC preamble information and the propagation path estimation preamble, converts the signal into a frequency domain signal using FFT, compensates the propagation path using the propagation path estimation preamble information, and removes phase noise using the pilot signal. .

The weighting and bit selection circuit 220 receives the OFDM packet preamble signal a206 of the received OFDM packet and the OFDM data signal a203 compensated by the AGC preamble and the propagation path estimation preamble separately. . The propagation path estimation preamble is used as a signal representing the propagation path amplitude characteristic | α (i) | of each subcarrier of the OFDM packet including the same.

The information of each subcarrier included in the OFDM data signal a203 is normalized in the normalization circuit 204 by the amplitude | α (i) | of the corresponding subcarrier of the propagation path estimation preamble signal a206, and is normalized. The constellation decomposition circuit 20
At 5 it is decoded from the constellation of FIG. In the case of this example, a signal subjected to 16QAM modulation is placed on each subcarrier, and the I component (In-phase component) and the Q component (Quadrature-phase component) of one subcarrier included in the OFDM data signal a203. The values are x,
When y is set, a constellation decomposition circuit 205 uses one subcarrier to calculate a205 (bl) = x a205 (b2) = x−2 a205 (b3) = y a205 (b4) = y−2 a205 (bl) ) To a205 (b4) are decoded and output as signal a205.

On the other hand, the propagation path estimation preamble signal a2
06, the power value | α (i) | ² of each subcarrier is calculated by the squaring circuit 206 and output as a signal a207. Signal a2 decoded from constellation
05 is weighted by the power a207 of the corresponding subcarrier of the propagation path estimation preamble in the weighting circuit 207, and a210 (bl) = | α | ² × a210 (b2) = | α | ² (x-2) a210 (b3 ) = | α | 2 y a210 (b4) = | α | 2 (y-2) in the above a210 (bl) ~a210 (b4) is calculated, weighted signal a210 Is output as The weighted signal a210 is weighted by the power characteristic of the propagation path, and the greater the received power of the corresponding subcarrier, the heavier the weight, and the higher the signal reliability.

Here, the signal a209 and the signal a210
Is 16 bits.
If the weighted OFDM data signal a210 is input to the Viterbi decoder 212 as it is with 16 bits, the hardware configuring the Viterbi decoder becomes large and the time required for Viterbi decoding becomes enormous. In the conventional method, 1
Of the 6 bits, only the number of bits determined from the MSB, 6 bits in the example described above, were fixed and selected and input to the Viterbi decoder 212. Four examples are shown below as the value of a210. Although the upper 6 bits of the signal a210 are transferred to the Viterbi decoder 212, it can be said that the second to fourth bits from the upper 6 bits among the 6 bits are not effectively acting. That is, in the conventional method, when a210 = “0000001000000000” (16 bits), “0”
00000 ”(6 bits) is changed to“ 0 ”when a210 =“ 0000100000000000 ”(16 bits).
00010 ”(6 bits) is changed to“ 1 ”when a210 =“ 1111111000000000 ”(16 bits).
11111 ”(6 bits) is changed to“ 1 ”when a210 =“ 1111100000000000 ”(16 bits).
11110 ″ (6 bits) are input to the Viterbi decoder 212, respectively.

Accordingly, in the present invention, in order to effectively use the number of input bits allocated to the Viterbi decoder 212,
The input bits to the Viterbi decoder 212 are selected using the power a209 per subcarrier of the propagation path estimation preamble as an index. The averaging circuit 208 receives the signal a207 indicating the power value of each subcarrier of the preamble for estimating the propagation path as an input, calculates the average power value per subcarrier, and outputs the averaged value as the signal a209. The signal a 209 is a preamble 52 for the propagation path estimation.
Approximation is performed by adding the subcarrier signals and dividing by an FFT length of 64, which is obtained by calculating the sum of 52 subcarriers and truncating the lower 6 bits. Also,
Since this power average value is used as an index of the bit truncation of the input to the Viterbi decoder 212, the averaging circuit 208 truncates the lower 5 bits and outputs the average power of 2.
In some cases, double and lower 4 bits are truncated to output four times the average power.

The comparator and rounding circuit 209 compares the power average value a209 of the preamble for estimating the propagation path with the OFDM data signal a210 decoded from the constellation weighted by the power characteristic of the propagation path. Is smaller than the power average value a209, the data signal a210 is directly used as the signal a21.
When the absolute value of the data signal a210 is greater than the power average value a209, the sign of the data signal a210 is multiplied by the power average value a209, and the result is output as a signal a211. For example, a209 = “0000010000000000”
In the case of (1024) (16 bits), when a210 = “0000001000000000” (16 bits), a
When 211 = “0000001000000000” (16 bits), a210 = “0000100000000000” (16 bits), a
When 211 = “0000010000000000” (16 bits) a210 = “1111111000000000” (16 bits), a
When 211 = "1111111000000000" (16 bits) a210 = When "1111100000000000" (16 bits) a
211 = “1111110000000000” (16 bits) By this operation, the data signal a21
If the absolute value of 0 is larger than the power average value a209, the value is rounded by the power average value a209.

Power average value a of propagation path estimation preamble
209 = “000001” as shown in the above example.
In the case of "0000000000" (16 bits), the upper 1 bit is a sign bit and is necessary. However, since the total 4 bits from the upper 2 bits to the upper 5 bits have no meaning as a value, they can be cut. At 210, an appropriate upper bit of the signal a211 is removed, and in this case, a total of 4 bits from the upper 2 bits to the upper 5 bits are removed and output as a signal a212. When “0000001000000000” (16bit), a
212 = “001000000000” (12 bits), a211 = “0000010000000000” (16 bits), a
When 212 = “010000000000” (12 bits), a211 = “1111111000000000” (16 bits), a
When 212 = “111000000000” (12 bits), a211 = “1111110000000000” (16 bits), a
212 = “110000000000” (12 bits).

When the number of input bits to the Viterbi decoder 212 is 6
If the bits are specified, it is necessary to cut out the lower bits, respectively,
Perform in 1. In the case of this example, the lower 1 bit to 6 bits are unnecessary, and are cut off. That is, when a212 = “001000000000” (12 bits),
3 = “001000” (6 bits),. When a212 = “010000000000” (12 bits), a21
When 3 = “010000” (6 bits) and a212 = “111000000000” (12 bits), a21
When 3 = “111000” (6 bits) and a212 = “110000000000” (12 bits), a21
3 = "110000" (6 bits), respectively.

When the power average value a209 of the propagation path estimation preamble can be represented by 6 bits or less while the input to the Viterbi decoder 212 is 6 bits, for example, the power average value a209 = “0000000000001000” (16 bits In the case of ()), it can be represented by only 5 bits such as “01000”. In this case, the processing is performed as follows. First, rounding of the value by the power average value a209 is performed by using the power average value a209.
Is larger than the input bit to the Viterbi decoder 212. In other words, the comparator and the rounding circuit 209 calculate a when a210 = “0000000000000100” (16 bits)
When 211 = “0000000000000100” (16 bits), a210 = “0000000000010000” (16 bits), a
When 211 = “0000000000001000” (16 bits), a210 = “1111111111111100” (16 bits), a
When 211 = “1111111111111100” (16 bits), a210 = “1111111111110000” (16 bits), a
211 = “1111111111111000” (16 bits).

However, the upper bit cutting circuit 210
Then, when cutting out the upper unnecessary bits of the signal a211,
The output signal a212 has 5 bits, and the Viterbi decoder 21
2 is smaller than the number of input bits. In such a case, the upper bit cutout circuit 210 outputs 6 bits from the lower 1 bit of the signal a211 as the signal a212. That is, when a211 = “0000000000000100” (16 bits),
When 212 = "000100" (6 bits), a211 = "0000000000001000" (16 bits), a
When 212 = "001000" (6 bits), a211 = "1111111111111100" (16 bits), a
When 212 = "111100" (6 bits), a211 = "1111111111111000" (16 bits), a
212 = “111000” (6 bits).

Therefore, the lower bit cutting circuit 21
In 1, if the signal a212 is exactly 6 bits, it is output to a213 without any operation. In other words, the lower bit cutout circuit 211 determines that when a212 = “000100” (6 bits), a213 = “00”
0100 ”(6 bits), when a212 =“ 001000 ”(6 bits), a213 =“ 00 ”
When "1000" (6 bits) and a212 = "111100" (6 bits), a213 = "11"
When 1100 ”(6 bits) and a212 =“ 111000 ”(6 bits), a213 =“ 11
Output as 1000 ”(6bit), respectively.

The signal a213 processed as described above is input to the Viterbi decoder 212, and the convolutional code is decoded.

In the above method, the power average value of all the subcarriers of the propagation path estimation preamble is calculated for the signal weighted by the power after the demapping from the constellation, and the upper bits of the decoded signal are calculated. The number of quantization bits was reduced by cutting off the power average value as an index and further appropriately cutting off lower bits. It has been described that the reduction in the number of quantization bits makes it possible to reduce the load on the Viterbi decoder and reduce the circuit size of the Viterbi decoder. However, the method does not mention the number of bits to be allocated to the signal value to be handled in a circuit located before the Viterbi decoder. In the above method, the data input to the constellation decoding circuit originally has a different signal level for each packet, and an empty bit is generated in an upper bit in a calculation result of an operation such as multiplication. For example, it is necessary to keep the increasing bit number width due to the occurrence of, for example. Increasing the width of the number of bits allocated to the signal data increases the computational burden and increases the circuit size as it is, thus increasing the size of the circuit located before the Viterbi decoder.

In order to reduce the scale of not only the Viterbi decoder but also the circuits located before the Viterbi decoder, operations such as addition, subtraction, multiplication, and division are performed in each circuit, resulting in a bit width. It is necessary to select an effective part effectively from the data in which the number of bits has increased and reduce the bit number width. As a result, the number of operation bits in each circuit can be reduced, the width of the number of bits of data transferred from the circuit to the circuit is reduced, and the scale of each circuit can be reduced.

As means for solving this problem, as shown in FIG. 5, the FFT circuit 203-3 and the propagation path compensation circuit 20 are used.
3-4. The amplitude adjusting circuit 230 is arranged behind each of the phase noise removing circuits 203-5. FFT circuit,
In a propagation path compensation circuit, a phase noise elimination circuit, and the like, a multiplication operation or the like is internally performed, so that the bit width of data becomes large and upper bits may become empty. The basic function of the amplitude adjustment circuit 230 is to receive the data having a large number of bits, detect the maximum absolute value from the data placed on each subcarrier of the preamble for estimating the propagation path, and determine the maximum absolute value of the data. That is, the bit shift amount is calculated, and the amplitude of all data included in the same OFDM packet is adjusted according to the calculated value. This degree of amplitude adjustment (amplitude adjustment value) is valid only while data included in the packet is being demodulated, and when the next OFDM packet arrives, the degree of amplitude adjustment is updated by the same procedure. Shall be. 5, the AGC circuit 201, the A / D converter 202, the frequency synchronizing circuit 203-1, the frequency offset compensating circuit 203-2, the FFT circuit 203-3, the propagation path compensating circuit 203-4, the phase noise removing circuit 203-5, The decoding circuit 205 from the constellation and the Viterbi decoder 212 have the same functions as those described with reference to FIGS.

Decoding circuit 205 from constellation
The bit selection circuit 240 arranged after the
, And is provided with a high-order bit cutting and rounding circuit 241 and a low-order bit cutting circuit 242. The bit selection circuit 240 adds a bit shift by a predetermined amount without referring to the propagation path estimation preamble, performs a rounding process when necessary, and performs lower-order processing so as to match the number of input bits to the Viterbi decoder 212. This is a circuit that cuts off bits. The reason why the averaging process as in the above-described solution method can be omitted from the bit selection circuit 240 that determines the input bits to the Viterbi decoder 212 is that each OFDM packet is provided by the function of the amplitude adjustment circuit 230 arranged earlier. Can be regarded as equivalently averaged.

Next, a detailed processing procedure of the amplitude adjustment circuit 230 will be described with reference to FIG. FIG. 6 is a diagram showing the configuration of the amplitude adjustment circuit 230. In FIG. 6, reference numeral 231 denotes a preamble separation circuit for estimating a propagation path, 232 denotes a format conversion circuit, 233 denotes a logical sum circuit, 234 denotes a bit shift amount determination circuit, and 235 denotes a bit shift amount determination circuit. The upper bit cutting and rounding circuit 236 is a lower bit cutting circuit.

The amplitude adjustment circuit 230 is arranged behind a circuit that includes an operation such as addition, subtraction, multiplication, division, etc., in which the bit width of data becomes large and the upper bits may become empty. The input signal is an OFDM packet composed of an OFDM data symbol to which a preamble for propagation path estimation is added at the head. First, a preamble for estimating the propagation path is extracted from this OFDM packet, the format of the value is changed, and it is converted to an absolute value with a sign bit added. Let N_ms be the maximum possible bit shift amount arbitrarily determined from the limitation of hardware and the basic principle of operation. For the data of each subcarrier included in the preamble, the upper N_ms bits excluding the sign bit are extracted, and the logical sum of them is calculated. The calculation result of this logical sum is a value of N_ms bits, and the position of the bit at which “1” first appears from the MSB is the MSSB of the maximum absolute value of the subcarrier value included in the preamble for propagation path estimation.
(Most Significant Set Bit), and the position of the bit whose maximum absolute value included in the preamble is used effectively. When this position is the N_l bit from the high order of the logical sum value, N_l-1 is determined as the bit shift amount N_s. However, when "1" is not added to the logical sum, the bit shift amount N_s is determined to be N_ms. Next, the OFDM
Bit shift is performed in the MSB direction by the determined bit shift amount N_s for all data included in the packet. However, if data overflows as a result of the bit shift, the value is rounded to saturate the value. The lower bits are complemented by “0” by the shifted amount. Further, the lower bits are truncated so as to match the input bit width of the circuit arranged at the rear.

The function of the amplitude adjusting circuit 230 will be described using a specific example. The preamble for estimating the propagation path is transmitted as known BPSK information from the transmitter side. However, since the preamble is returned to a digital signal again after undergoing amplitude distortion in the propagation path, the absolute value of each subcarrier is not a fixed value, Distorted. It is assumed that the output from the circuit arranged immediately before is a 16-bit two's complement number, the maximum bit shift amount N_ms = 4 bits, and 12-bit data is passed to the circuit arranged immediately after.

Assume that the number of subcarriers is four and the number of data OFDM symbols included in the packet is two. The data values of each subcarrier are 0001100001010111... Preamble (subcarrier 1) 0000101001011111. (Subcarrier 2) 1111100010010000 ... preamble (subcarrier 3) 0000000000111000 ... preamble (subcarrier 4) 0001101101010111 ... data 1 (subcarrier 1) 1100101001011111 ... ... data 1 (subcarrier 2) 1111100010010000 ... data 1 (sub) Carrier 3) 0000111000111000 data 1 (subcarrier 4) 0000000001001111 data 2 (subcarrier 1) 0010100001011111 data 2 (subcarrier 2) 0001100010010000 ... data 2 (subcarrier 3) 0000000000111000 data 2 (subcarrier) Carrier 4) above Let's say something like

First, only the preamble is extracted and 2
Is changed to a combination of a sign bit and an absolute value.

0001100001010111... Preamble (subcarrier 1) 0000101001011111... Preamble (subcarrier 2) 1000011101110000... Preamble (subcarrier 3) 0000000000111000... Preamble (subcarrier 4) N_ms = 4; The bits are taken out and the logical sum is calculated. Here, 0011or0001or0000or0000 = 0011. In this calculation result, since “1” is set for the first bit in the third bit as viewed from the high order, N_l = 3 and the bit shift amount is determined as N_l−1 = 2.

Therefore, all values included in the packet are bit-shifted by two bits, and the lower two bits are complemented by "0".

0110000101011100 ... preamble (subcarrier 1) 0010100101111100 ... preamble (subcarrier 2) 1110001001000000 ... preamble (subcarrier 3) 0000000011100000 ... preamble (subcarrier 4) 0110110101011100 ... data 1 (subcarrier 1) 1000000000000000 ... Data 1 (subcarrier 2) 1110001001000000 ... Data 1 (subcarrier 3) 0011100011100000 ... Data 1 (subcarrier 4) 0000000100111100 ... Data 2 (subcarrier 1) 0111111111111100 ... Data 2 (subcarrier 2) 0110001001000000 ... Data 2 (subcarrier 3) 0000000011100000... Data 2 (subcarrier 4) As a special case, subcarrier 2 of data 1 is a negative number, but is replaced with a negative maximum value due to overflow. Further, the subcarrier 2 of the data 2 is a positive number, which is replaced with a positive maximum value because overflow occurs.

Since the output requires 12 bits,
Since the lower 4 bits are unnecessary, they are rounded down. 011000010101... Preamble (subcarrier 1) 001010010111... Preamble (subcarrier 2) 111000100100... Preamble (subcarrier 3) 000000001110... Preamble (subcarrier 4) …… Data 1 (subcarrier 1) 100000000000 …… Data 1 (subcarrier 2) 111000100100 …… Data 1 (subcarrier 3) 001110001110 …… Data 1 (subcarrier 4) 000000010011 …… Data 2 (subcarrier 1) 011111111111 ... Data 2 (subcarrier 2) 011000100100... Data 2 (subcarrier 3) 000000001110... Data 2 (subcarrier 4) Process as described above.

By passing the data having undergone the above operations to the circuit arranged immediately thereafter, it is possible to express a numerical value by effectively utilizing the allocated bit number width.

[0054]

When the AGC circuit functions ideally,
The power average value of the preamble becomes equal for each packet, but actually causes a deviation. According to the present invention, the digital circuit after A / D conversion performs processing such as compensation and decoding from a constellation, and performs propagation path estimation at each processing stage until input to a Viterbi decoder for decoding a convolutional code. The AGC function can be assisted by the digital circuit by appropriately selecting valid and appropriate bits from the signal bits using the power average value and amplitude adjustment value of the preamble as an index. Further, by applying the present invention, it is possible to select valid bits from a data signal whose number of bits has been increased by a digital signal operation. For example, conventionally, even when 12 bits are allocated to each processing circuit input, In practice, it is possible to solve the problem that only 8 bits can be effectively used, and it is possible to effectively use the allocated number of bits.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a configuration of an OFDM signal packet.

FIG. 2 is an explanatory diagram showing a constellation of subcarriers of an OFDM signal subjected to 16QAM modulation.

FIG. 3 is a block diagram showing a configuration of a weighting and bit selection circuit for a Viterbi decoder according to the related art.

FIG. 4 is a block diagram illustrating a configuration of a weighting and bit selection circuit for a Viterbi decoder according to an embodiment of the present invention.

FIG. 5 is a block diagram showing a main configuration of an OFDM decoding device having a function of controlling the number of operation bits according to another embodiment of the present invention.

FIG. 6 is a block diagram illustrating a configuration of an amplitude adjustment circuit in FIG. 5;

FIG. 7 is a block diagram showing a configuration of a bit selection circuit in FIG. 5;

[Explanation of symbols]

 201 AGC circuit 202 A / D converter 203 Processing circuit 203-1 Frequency synchronization circuit 203-2 Frequency offset compensation circuit 203-3 FFT circuit 203-4 Propagation path compensation circuit 203-5 Phase noise removal circuit 204 Normalization circuit 205 Constellation Decomposition circuit 206 Square circuit 207 Weighting circuit 208 Average circuit 209 Comparator and rounding circuit 210 Upper bit cutout circuit 211 Lower bit cutout circuit 212 Viterbi decoder 220 Weighting and bit selection circuit a203 OFDM data signal a204 Normalized signal a205 Constellation A206 A preamble signal for propagation path estimation a207 A signal indicating the power value of each subcarrier a209 A signal indicating the average power value a210 Weighted Signal a211 Signal rounded by power average value a212 Output signal of upper bit cutout circuit a213 Output signal of lower bit cutout circuit 230 Amplitude adjustment circuit 231 Preamble separation circuit for propagation path estimation 232 Format conversion circuit 233 Logical OR circuit 234 Bit shift amount Decision circuit 235 Upper bit cutting and rounding circuit 236 Lower bit cutting circuit 240 Bit selection circuit 241 Upper bit cutting and rounding circuit 242 Lower bit cutting circuit

Continued on the front page F term (reference) 5J065 AC02 AD10 AH04 AH12 AH13 AH23 5K014 AA01 BA10 BA11 EA01 HA00 5K022 DD01 DD18 DD33 DD34 5K046 AA05 DD02 DD15 EE19 EE42 EE56

Claims (10)

[Claims] 1. An OFDM signal demodulation device having a Viterbi decoder for decoding a convolutional code, comprising: an averaging circuit for calculating an average power value of a subcarrier; and an OFDM signal weighted based on an output of the averaging circuit. An OFDM demodulator, comprising: a rounding circuit for rounding a data signal; and a bit cutting circuit for cutting off unnecessary data, wherein an effective bit of the OFDM data signal is appropriately selected according to a power average value. 2. The power average value according to claim 1, wherein the power average value is obtained by adding each subcarrier signal power of the propagation path estimation preamble to an FFT (Fast Fourier Trajectory).
nsform; an OFDM demodulator characterized by being obtained by dividing by a length of a fast Fourier transform. 3. The OFDM apparatus according to claim 1, wherein the power average value is valid only while a packet including a propagation path estimation preamble from which the power average value is calculated is being demodulated. Demodulator. 4. The OFDM demodulator according to claim 1, wherein the bit cutout circuit includes an upper bit cutout circuit and a lower bit cutout circuit. 5. The OFDM demodulator according to claim 4, wherein the upper bit cutoff circuit cuts off a bit at the same position as an upper bit unnecessary for expressing the power average value. 6. An OFDM demodulator comprising a Viterbi decoder for decoding a convolutional code, comprising an amplitude adjusting circuit for controlling the number of quantization bits. 7. The amplitude adjustment circuit according to claim 6, wherein the amplitude adjustment circuit is configured to determine an MSSB (Most Signi? Er) of a maximum absolute value of a subcarrier included in a propagation path estimation preamble.
An OFDM demodulator characterized in that a bit shift amount is determined from a position of a ficant set bit). 8. The amplitude adjustment value according to claim 6, wherein the amplitude adjustment value by the amplitude adjustment circuit is valid only while a packet including a propagation path estimation preamble from which the calculation is performed is demodulated. An OFDM demodulator characterized by the above-mentioned. 9. The logic circuit according to claim 6, wherein the amplitude adjustment circuit calculates a logical sum of each subcarrier data included in the preamble, and calculates a logical sum from a calculation result of the logical sum. An OFDM demodulator, comprising: a bit shift determining circuit for determining a bit shift amount; and a bit cutting circuit for cutting off unnecessary data, and appropriately selecting valid bits of an OFDM data signal. 10. The OFDM demodulator according to claim 9, wherein said bit cutout circuit comprises an upper bit cutout and rounding circuit and a lower bit cutout circuit.