JP2019087967A - Frame error rate prediction device, wireless communication device and wireless communication system using the same - Google Patents

Abstract

To provide a frame error rate prediction device of a communication system in which a pairwise error rate differs depending on a position on a trellis in a convolutional code.SOLUTION: A frame error rate prediction device 1110 of a communication system communicating by a convolutional coding OFDM system includes a FER prediction unit 3030 that calculates a pairwise error rate of an error path in decoding processing for convolutional coding at each position of a section of a trellis diagram corresponding to the size of a frame for each of a coding rate and a modulation scheme stored in storage means as a pair on the basis of propagation path characteristic information representing attenuation of propagation path power and noise power estimated in a communication path on a basis of information representing the attenuation of power by the communication path and the noise power to predict a frame error rate as the sum of the pairwise error rates calculated at respective time points.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a frame error rate prediction device, a wireless communication device using the same, and a wireless communication system.

  In recent years, offloading of mobile traffic has progressed in the ISM (industrial, scientific, and medical radio) band, and it is desired to improve the frequency utilization efficiency of a wireless local area network (LAN). In order to realize high frequency utilization efficiency, it is necessary to be able to achieve a frame error rate (FER) sufficiently low for practical use and to use a transmission rate as high as possible. In general, it is conceivable to control MCS (modulation and coding scheme) according to the propagation condition so that FER is about 1% to 10%. Here, MCS refers to a table of predetermined combinations of modulation schemes and channel coding rates (see, for example, Patent Document 1). For example, when the reception condition of the receiver is poor, or when transmission data that requires communication with a low error rate uses MCS with a low transmission rate, conversely, when the reception condition of the receiver is good, a relatively high error Transmission data that allows rates is adaptively controlled using a decision method such as using a high transmission rate MCS.

As a method of controlling to a suitable MCS in a wireless LAN, there is known a method of adjusting the MCS according to the transmission success rate and the number of retransmissions, and selecting the MCS suitable for the propagation situation (see Non-Patent Document 1) . However, in this method, since frame transmission is performed with MCS that does not match the propagation condition until an optimal MCS is selected, throughput may be reduced due to retransmission and low rate transmission. Therefore, to improve transmission efficiency, it is desirable to predict FER (Frame Error Rate) in advance and to determine MCS based on the result. However, to achieve that, high-accuracy FER prediction is required.
It is known that FER can be accurately predicted by theoretical calculation of FER value based on pairwise error probability (PEP) in a region where FER is 1% to 10% (Non-Patent Document 2) .

  In the case of determining the error rate by PEP in the currently popular IEEE 802.11a or later wireless LAN, it is necessary to determine the PEP in convolutional coded OFDM (COFDM: coded orthogonal frequency division multiplexing). The PEP in COFDM has been studied in Non-Patent Document 3 and so on.

  In Non-Patent Document 3, analysis of PEP is performed when bit errors after interleaving can be regarded as random. When it can be regarded as a random error, an error path in which the Hamming distance is separated by the minimum free distance becomes dominant, and the frequency response of the propagation path can also be regarded as random. Non-Patent Document 3 presents a method of deriving PEP using this property.

  However, assuming a practical wireless LAN operation, the communication path is not necessarily an environment that can be regarded as a random error. In a wireless LAN such as IEEE 802.11a, since the bandwidth and interleaving size are not sufficiently wide and only delay dispersion of several hundreds (ns) is considered, the occurrence of bit errors can be sufficiently dispersed. It may not be possible. Therefore, the frequency response can not be regarded as random equivalently, and the above-mentioned approximate expression of Non-Patent Document 3 does not hold, which does not depend on the frequency response of the propagation path. In addition, the premise that the PEP of the error path of the minimum free distance dominates does not hold, and it is necessary to consider the PEP of the error path having the Hamming distance greater than the minimum free distance. Therefore, the PEP derivation method presented in Non-Patent Document 3 can not accurately approximate PEP in a wireless LAN environment.

  In order to solve such a problem, in the technology disclosed in Non-Patent Document 4, in a communication system that performs communication according to the convolutional coding OFDM method, a transmitter and a receiver are received as bit error rate (BER) prediction. The PEP is derived based on the noise power and the information representing the power attenuation by the communication path estimated in the machine or both. In the case of OFDM communication in a multipath environment, the influence of the propagation path differs depending on the frequency response of subcarriers. Furthermore, multi-level modulation is also used in a wireless LAN. Thus, each codeword bit is affected differently and PEP is different at each position in the trellis. In the technique disclosed in Non-Patent Document 4, PEP is calculated at each time point in the trellis section corresponding to the interleaver size in order to take into consideration the difference in PEP due to the position of the trellis, and bit errors are obtained from the obtained PEP Calculate the expected value of the number of bits to be and perform BER prediction. As FER prediction, it is assumed that the error rate of each information bit is equal, and it is assumed that FER is predicted based on the predicted BER.

Unexamined-Japanese-Patent No. 2010-41074 specification

Unexamined-Japanese-Patent No. 2011-211433 specification

Unexamined-Japanese-Patent No. 2013-18756 specification

S. Biaz and S. Wu, “Rate adaptation algorithms for IEEE 802.11 networks: A survey and comparison,” 2008 IEEE Symposium on Computers and Communications, pp. 130-136, July 2008.

A. Martinez, A. Guillen i Fabregas, and G. Caire, "Error probability analysis of bit-interleaved coded modulation," IEEE Transactions on Information Theory, vol. 52, no. 1, pp. 262-271, Jan. 2006.

Y. Hori and H. Ochiai, "Performance analysis and interleaver structure optimization for short-frame BICM-OFDM systems," IEEE Transactions on Wireless Communications, vol. 15, no. 1, pp. 651-662, Jan. 2016.

菅 et al., “Bit Error Probability Prediction of Convolutionally Coded OFDM Transmission Suitable for IEEE 802.11 Wireless LAN,” IEICE Technical Report IT 2016-125, Mar. 2017.

  In the prior art as described above, the error rate of each information bit is different by using different SNRs and multilevel modulation for each subcarrier (that is, although the PEP is different at each position of the trellis), they are averaged. The calculated BER is calculated. However, since frame errors occur when any one of the information bits in a frame is a bit error, it is necessary to consider the difference in error rate of each information bit in order to predict FER with high accuracy. Therefore, as in the conventional method, when calculating FER from an averaged value such as BER, it is not possible to consider the difference in error rate of each bit in a frame, so FER is calculated based on BER. What to do may not always be appropriate.

  The present invention has been made to solve the above problems, and its object is to predict FER in a communication system in which the PEP differs depending on the position on the trellis in the convolutional code. It is an object of the present invention to provide a FER prediction device, and a wireless communication device and a wireless communication system using the FER prediction device.

  According to one aspect of the present invention, there is provided a FER predictor for a communication system that communicates in a convolutional coding scheme that can be decoded by a trellis diagram, and the same interleaving process is repeatedly performed on transmission bits in a frame. Storage means for storing information on a set of a coding rate of a convolutional coding and a modulation scheme which are preset to be adopted in a coding scheme, and an attenuation of propagation path power estimated on a communication path Based on channel characteristic information and noise power, for each of the coding rate and modulation scheme set stored in the storage means, the PEP of the error path in the decoding process for convolutional coding is on the trellis corresponding to the size of the frame And FER prediction means for predicting FER as the sum of PEPs calculated at each position of That.

  Preferably, the FER prediction means 1 divides the transmission bit into 1 if the channel variation for the coded sequence after the de-interleaving process is cyclically considered to be cyclic in the frame. The block error rate is calculated as the sum of PEPs in the trellis diagram section corresponding to the block, and FER is predicted from the block error rate and the number of blocks contained in the frame, assuming that each block error rate is equal to each other.

  Preferably, the communication system performs communication by orthogonal frequency division multiplexing, and the block is one symbol in orthogonal frequency division multiplexing.

  Preferably, the propagation path characteristic information is a power delay profile estimated in the communication path.

  Preferably, the propagation path characteristic information is an impulse response of the propagation path.

  According to another aspect of the present invention, there is provided a wireless communication apparatus that communicates by a convolutional coding method, wherein feedback of propagation path characteristic information and noise power estimated from the reception apparatus and representing attenuation of propagation path power is performed. Receiving means for receiving as information, storage means for storing information of a combination of a coding rate of a convolutional coding and a modulation method set in advance as adopted in the convolutional coding method, and feedback information And selecting means for selecting a group that maximizes throughput among the sets of the coding rate and the modulation scheme stored in the storing means, the selecting means storing the memory means based on the channel characteristic information and the noise power. For each of the coding rate and modulation scheme set stored in the frame, the PEP of the error path in the decoding process for convolutional coding is Is calculated at each position on the trellis corresponding to Z, and the sum of PEPs calculated at each position is a predetermined FER based on frame error rate prediction means for predicting a frame error rate and the predicted frame error rate. And modulation scheme selecting means for selecting a set that provides the maximum throughput, and performing convolutional coding and modulation processing on transmission data according to the set selected by the modulation scheme It further comprises transmission means for

  Preferably, the frame error rate prediction means is configured to divide the transmission bits into block lengths set such that propagation path variation for the de-interleaved coded sequence can be regarded as periodic in a frame. The block error rate is calculated as the sum of PEPs in the trellis diagram section corresponding to one block, and FER is predicted from the block error rate and the number of blocks included in the frame, assuming that each block error rate is equal to one another.

  Preferably, the communication system performs communication by orthogonal frequency division multiplexing, and the block is one symbol in orthogonal frequency division multiplexing.

  Preferably, the propagation path characteristic information is a power delay profile estimated in the communication path.

  Preferably, the propagation path characteristic information is an impulse response of the propagation path.

According to still another aspect of the present invention, there is provided a wireless communication system for performing communication according to a convolutional coding OFDM method, comprising: a receiver; and the receiver including channel characteristic information representing attenuation of channel power in the communication path. And communication path state estimation means for estimating noise power, and first transmission means for transmitting the estimated channel characteristic information and noise power, further comprising a transmission apparatus, the transmission apparatus comprising: Receiving means for receiving, as feedback information, the transmitted channel characteristic information and the estimated value of noise power; and a coding rate and modulation scheme of convolutional coding preset as employed in the convolutional coding OFDM system. Storage means for storing the set of information, and based on the feedback information, among the sets of the coding rate and the modulation scheme stored in the storage means Selecting means for selecting a set maximizing the throughput, wherein the selecting means is based on the propagation path characteristic information and the noise power, for each of the coding rate and modulation scheme sets stored in the storage means, a convolutional code FER prediction means for calculating the PEP of the error path in the decoding process for the coding at each position on the trellis corresponding to the size of the frame, and predicting the FER as the sum of the PEP calculated at each position; On the basis of the selected FER, the modulation scheme selecting means for selecting the set that provides the maximum throughput in the predetermined FER, and the convolutional coding is performed on the transmission data according to the set selected by the modulation scheme selection unit. And second transmitting means for performing and transmitting modulation processing.
(Definition of terms)
In the present specification, the meaning of terms is as follows.

"Pairwise error rate (PEP)" and, when the code word C _L of the length L from the transmitter and is transmitted, the receiver side of this sequence, when decoding contain errors and C _L ', the code word C _L refers to the probability determining mistakenly codeword C _L '.

  The “minimum free distance” means that, in the convolutional code, when the code words for two information sequences u and v are c (u) and c (v), two code sequences of all error paths and true paths are used. The minimum value of the Hamming distance.

  The “power delay profile” indicates how received power is distributed in the delay time domain, assuming that a propagation path is composed of multiple paths with different delay times. That is, in the propagation path of the implementation, the impulse response has a time spread because the path length is different for each path. Therefore, the "power delay profile" is represented by the squared average value of the impulse response.

  The "diversity order" refers to the number of subcarriers for which the frequency response of each subcarrier to which different bits are mapped between the codeword corresponding to a certain error path and the transmission codeword can be regarded as random (uncorrelated). In an environment where channel errors can be regarded as random errors, the diversity order is maximum (the frequency response of the channel through which different bits have passed between the codeword corresponding to the error path and the transmission codeword is uncorrelated). If the diversity order is lower, burst errors are likely to occur and the BER is degraded.

  The "error path" refers to a path in which a bit error occurs when the path is selected by maximum likelihood decoding for a convolutional code.

  The “error pattern” refers to a code word of an error path corresponding to a section different between the true path and the error path. In a predetermined convolutional code, since the number of error patterns that can occur is limited according to the generator polynomial, they are numbered and referred to as, for example, “p th error pattern”.

  The “Hamming distance” refers to the number of different bits in corresponding positions in two bit strings.

  According to the present invention, it is possible to predict FER in a communication system in which the PEP differs depending on the position on the trellis in the convolutional code.

FIG. 1 is a block diagram for describing a configuration of a wireless communication system according to a first embodiment. It is a conceptual diagram for demonstrating typically the transmission and reception process by the OFDM system shown in FIG. FIG. ₁₈ is a diagram showing an error path at dH = 5 of a convolutional code having a constraint length of 3 and a generator polynomial (5, 7) ₈ ; It is a block diagram for demonstrating the structure of the adaptive rate control part 1110 shown in FIG. 5 is a flowchart for describing adaptive rate control according to the first embodiment. It is a conceptual diagram for demonstrating the problem when PEP differs according to the position on a trellis. It is a conceptual diagram which shows the concept of Formula (4). It is a conceptual diagram for demonstrating the pattern of PEP when the same interleaving process is repeatedly implemented by each block in the flame | frame. It is a conceptual diagram for demonstrating the procedure which predicts FER by the operation by a block unit.

  Hereinafter, configurations of a wireless communication system and a wireless communication apparatus according to an embodiment of the present invention will be described. In the following embodiments, components and processing steps denoted by the same reference numerals are the same or equivalent, and the description thereof will not be repeated if not required.

  As described below, in this embodiment, as in the wireless LAN environment, an error in the communication path is not necessarily regarded as a random error, and the PEP differs depending on the position in the trellis of the convolutional code. In such a case, a method of accurately predicting FER will be described.

  In the following, in order to make the technical significance of the present embodiment easy to understand, the calculation method of FER will be described in the following order.

  (1) First, calculation of FER assuming a system in which the PEP is the same at any position on the trellis will be described.

  (2) Next, as a method of calculating FER of the present embodiment, a method of calculating PEP and predicting FER in a trellis section corresponding to a frame size will be described.

(3) Furthermore, in (2), in the communication system in which the same interleaving is repeatedly performed, such as a wireless LAN, when it can be considered that there is no channel fluctuation in a frame, the channel received by the deinterleaved coded sequence The block is set so that the fluctuation is periodic with the block length as a cycle, and the PEP calculation interval is limited to the trellis interval equivalent to one block by using equal block error rates in the block period. A method of predicting the FER will be described.
First Embodiment
FIG. 1 is a block diagram for explaining the configuration of the radio communication system according to the first embodiment.

  Referring to FIG. 1, transmitting apparatus 1000 performs error correction coding section 1002 for performing error correction coding processing using convolutional code on data of a transmission sequence, and data for error correction coding. Data obtained by dividing the data string into the number of subcarriers based on MCS selected as described later, serial-parallel conversion is performed on the interleaving unit 1004 that performs interleaving processing and the data string after interleaving Inverse Fourier transform processing and guard interval addition processing are performed on the digital signal of the modulation unit 1010 for performing subcarrier modulation and the modulation unit 1010 output to generate an OFDM symbol, and digital analog conversion processing is performed. Orthogonal Frequency Division Multiplexing (OFDM) Modulating Unit 1012 , A high frequency processing unit (RF unit) 1014 that performs orthogonal modulation processing, up conversion processing, power amplification processing and the like on the signal after OFDM modulation, and an antenna 1020 for transmitting a high frequency signal of the RF unit 1014 Including.

  Note that the RF unit 1014 executes low noise amplification processing, down conversion processing, quadrature demodulation processing, and the like on the signal received by the antenna 1020.

  Also, the modulation scheme of subcarrier modulation is not particularly limited, and it is assumed that there are, for example, types such as BPSK, QPSK, 16 QAM, 64 QAM, and the like.

  Further, transmitting apparatus 1000 receives a power delay profile and an estimated value of noise power from receiving apparatus 2000 side via RF section 1014, and receives processing section 1100 for executing demodulation and decoding processing, and the received power. Adaptive rate control unit that controls the coding rate of the error correction coding unit 1002 and the modulation scheme of the modulation unit 1010 by performing control to change MCS adaptively based on the delay profile and the estimated value of noise power. And 1110.

  The communication scheme of the power delay profile and the noise power estimated value from the reception apparatus 2000 may be the same communication scheme as that of data transmission, or another communication scheme may be adopted.

  The receiving apparatus 2000 includes an antenna 2002, an RF unit 2010 that performs low noise amplification processing, down conversion processing, quadrature demodulation processing and the like of the signal of the antenna 2002, and analog-to-digital conversion processing on the signal from the RF unit 2010. A received data sequence is generated by the inverse processing of modulation section 1010 on the signal from OFDM demodulation section 2012 for performing OFDM demodulation processing such as guard interval removal processing, Fourier transform processing, etc. and OFDM demodulation section 2012. A demodulation unit 2014 includes a de-interleaving unit 2016 for performing de-interleaving processing on the output of the demodulation unit 2014, and an error correction unit 2018 for performing error correction processing by decoding on a convolutional code.

  Also in the reception apparatus 2000, the RF unit 2010 executes quadrature modulation processing, up-conversion processing, power amplification processing and the like for transmitting the power delay profile and the estimated value of the noise power.

  Receiving apparatus 2000 further calculates a power delay profile / noise power estimation for calculating an estimated value of a power delay profile and an estimated value of noise power by, for example, a pilot signal or the like in a signal received via RF unit 2010. Unit 2100, and a transmission processing unit 2110 for converting the estimated value from the power delay profile / noise power estimation unit 2100 into a transmission signal and transmitting it from the antenna 2002 to the transmission apparatus 1000 via the RF unit 2010. Including.

  FIG. 2 is a conceptual diagram for schematically explaining transmission and reception processing in the OFDM system shown in FIG.

  As shown in FIG. 2, the convolutionally encoded and interleaved transmission signal is mapped for each subcarrier as a complex signal by a predetermined modulation scheme, and is subjected to digital-to-analog conversion after inverse Fourier transformation to perform quadrature modulation etc. It is subjected to frequency conversion processing including and sent to the transmission line.

  The signal received from the transmission path is subjected to frequency inverse conversion including quadrature detection, analog-to-digital converted, and Fourier-transformed, and the demapped received signal is subjected to de-interleave processing and error correction processing using a convolutional code. Be done.

Consider a trellis path (hereinafter, true path) that generates a codeword with all 0 configuration bits and a path (hereinafter, error path) that generates a codeword separated by Hamming distance d _H (true because convolutional code is a linear code) If you set the path and error path as above, there is no loss of generality).

FIG. 3 is a diagram showing an error path at d _H = 5 of a convolutional code having a constraint length of 3 and a generator polynomial (5, 7) ₈ .

  In this case, the true path corresponds to the code word ..., 0, 0, 0, 0, 0, 0, ..., and the error path corresponds to the code word ..., 1, 1, 0, 1, 1, 1, ... Do. Since the error path leaves the true path at a certain position on the trellis (t = 2 in FIG. 3) and then merges, it has a pattern of bits different only in this section. In this embodiment, this different bit pattern is called an error pattern. For example, in FIG. 3, the error pattern is [1, 1, 0, 1, 1, 1].

Furthermore, as described below, FER is predicted based on such error patterns that may occur for a given convolutional code.
[Configuration of prediction processing of FER]
FIG. 4 is a block diagram for explaining the configuration of adaptive rate control section 1110 shown in FIG.

  Referring to FIG. 4, adaptive rate control section 1110 includes MCS storage section 3020 for storing information of a set of MCS set in advance, an estimated value of a power delay profile sent from the side of receiving apparatus 2000, and For each MCS stored in MCS storage unit 3020, FER prediction unit 3030 predicts FER for each MCS based on the information of the frame size used for communication And a MCS selection unit 3040 for selecting an MCS achieving the maximum throughput among MCSs below the FER preset and required in the system based on the value of FER.

  The operation of the FER prediction unit 3030 will be described in more detail later.

  FIG. 5 is a flowchart for explaining the adaptive rate control according to the first embodiment.

  Referring to FIG. 5, first, on the receiving device 2000 side, the power delay profile / noise power estimation unit 2100 executes estimation of the power delay profile and the noise power (S100).

  The estimation of the power delay profile and the estimation of the noise power are not particularly limited, and for example, the methods disclosed in the following documents can be used.

Known Literature 1: T. Cui and C. Tellambura, “Power delay profile and noise variance estimation for OFDM,” IEEE Communications Letters, vol. 10, no. 1, pp. 25-27, Jan 2006
The estimated value is transmitted from the receiving device 2000 (S102), received by the transmitting device 1000 (S104), and the FER predicting unit 3030 of the transmitting device 1000 detects the error path at each position in the trellis section corresponding to the frame size. By calculating the PEP and calculating the total sum in the frame, the predicted value of FER is calculated for each MCS (S108), and the MCS selection unit 3040 selects a predetermined MCS based on the calculated FER. Then, in the range in which the defined FER is achieved, the MCS with the largest throughput is selected (S110).

  According to the selected MCS, error correction coding section 1002 executes convolutional coding at the selected coding rate, and modulation section 1010 performs subcarrier modulation in the selected modulation scheme (S112). .

  The modulated data is transmitted from transmitting apparatus 1000 (S114) and received by receiving apparatus 2000 (S116).

With such processing, the receiver side estimates the power delay profile and the noise power, and when it feeds back to the transmitter side, the coding rate and the modulation scheme can be adaptively changed. Therefore, by accurately predicting FER, it is possible to reduce the number of frame transmissions in MCS that do not match the propagation situation and improve throughput.
[Operation Performed by FER Predictor 3030]
Below, the operation | movement which the FER estimation part 3030 performs is demonstrated according to numerical formula.

1. Characteristics of the Method of the Present Embodiment As will be apparent from the following description, the prediction of FER of the present embodiment has the following characteristics.

  i) Calculate the achievable diversity order from the impulse response length of the propagation path and the minimum free distance, and enable prediction even in a propagation environment where only low diversity order can be achieved.

  ii) Calculate the error rate taking into account error paths longer than the minimum free distance.

  iii) FER is predicted by calculating and summing PEP for each position in the trellis interval corresponding to the frame size.

  A more detailed description will be given below.

2. System Model 2.1 System Model In the following, SISO (single-input single-output) -OFDM is assumed for simplicity of explanation. Then, let K _{data be} the number of data subcarriers and K _fft be the size of FFT (fast Fourier trans-form). Also, assuming that the channel is a frequency selective fading channel, it is assumed that there is no fluctuation in the OFDM symbol. Assuming that the propagation path is a finite impulse response filter, each tap coefficient is expressed by the following equation.

Here, h = [h ₁ , ..., h _L ] ^T (L is the impulse response length) is a vector which is a random variable in which each element follows a complex normal distribution with an average of 0 and a variance of 1 independent of each other. It is a diagonal matrix that represents a power delay profile. However, (...) ^T represents transposition of a vector or a matrix, and the element of P satisfies the following formula.

Also, L does not exceed the guard interval length. After guard interval removal, the received signal Y _k of the kth subcarrier subjected to FFT is given as follows.

Here, X _k is a transmission symbol, and Z _k is an FFT result of complex additive white noise that has an average of 0 and a variance of σ ² . H _k is the frequency response of the propagation path in the kth subcarrier, and is given by the following equation.

Here, g ^H _Kfft (k) is a vector obtained by extracting the first element to the L th element of the k-th row of the FFT matrix, and (...) ^H represents Hermitian transposition. Also, the following relationship holds.

(1) Process according to the prior art to be compared with the operation of the FER prediction unit of the present embodiment Based on the technology disclosed in the following known documents, first, a system in which the PEP becomes the same at any position on the trellis The calculation method of FER at the time of assuming is demonstrated.

  That is, the interleaver in which the influence of the propagation path received by the codeword error bit is large enough to be regarded as uncorrelated, and the codeword error bit is equally mapped to each bit position in the modulation symbol at any position on the trellis The calculation method of FER at the time of assuming the system which uses is explained.

Literature 2: M. Pursley and D. Taipale, "Error Probabilities for Spread-Spectrum Packet Radio with Convolutional Codes and Viterbi Decoding," in IEEE Transactions on Communications, vol. 35, no. 1, pp. 1-12, Jan 1987.
Publication 3: CY Lou and B. Daneshrad, "PER prediction for convolutionally coded MIMO OFDM systems-An analytical approach," MILCOM 2012-2012 IEEE Military Communications Conference, Orlando, FL, 2012, pp. 1-6.
First, since FER is the probability that any error event occurs at any position in the frame, in the convolutional code, the error event that the pth error pattern occurs at time t of the trellis is et _, Assuming _p (under bar), FER is expressed as the following equation. Here, "x (under bar)" indicates that an under bar is added below the letter "x", and indicates that the letter is a vector.

  Here, it is further considered to apply the following union bound equation, which holds for the upper limit of the probability of the sum event, to the above FER equation (3).

  In that case, FER is expressed as follows.

  Furthermore, according to known reference 2, when PEP can be regarded as the same at any position on the trellis, FER is as the following equation.

Here, L _frame is the number of bits in a frame. Also, PEPEP (e _p (under bar)) indicates the probability that an error event will occur at a certain location not specified on the trellis.

2.2 Method of calculating PEP in a trellis section corresponding to a frame size and predicting FER The processing according to the prior art to be compared with the operation of the FER prediction unit of the present embodiment as described above is as follows. There are some problems.

  That is, when the PEP differs depending on the position on the trellis, it is not appropriate to calculate FER as in the above equation (3-2), and as a result, the prediction accuracy is degraded.

  FIG. 6 is a conceptual diagram for explaining problems in the case where PEPs differ depending on the position on the trellis.

  First, as shown in FIG. 6, after interleaving processing is performed on a convolutionally encoded transmission signal by an interleaver, processing for performing multi-level modulation mapping for each subcarrier will be considered.

  In FIG. 6, for example, 16 QAM is assumed as multi-level modulation, and only mapping in the real axis direction is extracted and shown for simplicity of description.

  First, when transmitting signals are divided and mapped for each subcarrier, the SNR is different depending on the mapped subcarriers, so on the receiving side, the PEP is different depending on the position of the trellis at the time of decoding a convolutional code. It can occur.

  Further, in the case of multi-level modulation, it is likely that there are hard-to-error bits and hard-to-error bits, and the PEP may differ depending on the mapped bit position.

  For example, considering 2-bit signals "00", "01", "11" and "10", if the mapped quadrant is "0" according to the value of the first bit, the left quadrant (figure In the case of “1”, it is divided into right quadrants (dotted lines in the figure). As a result, in the demapping process on the receiving side, an error of the first bit is less likely to occur.

On the other hand, for the second bit, in the case of “1”, there are two signal points adjacent to each other across the origin, and in the case of “0”, the above “a” is located across the origin. It becomes two signal points facing each other across two signal points in the case of 1 ′ ′. Therefore, in the space to be mapped (constellation), the distance between signal points due to bit inversion is smaller in the second bit than in the first bit, so the second bit is the first bit. Errors are more likely to occur.
That is, for example, in a communication system such as a wireless LAN, the SNR for each subcarrier is different, and multilevel modulation is used (the error rate is different at bit positions in a mapped symbol), so PEP differs depending on the position on the trellis. .

  When PEP differs depending on the position on the trellis, calculation of PEP is performed by taking into consideration the position on the trellis in the trellis section corresponding to the frame size to obtain FER as shown in the following equation. Predict the FER by summing the calculated PEPs.

  Therefore, specifically, FER is calculated by the following calculation.

Here, in the equation (4), L _frame means a section corresponding to the frame size. Therefore, equation (4) means that FER is calculated by summing the PEPs calculated for each position of each frame in the frame. However, PEP in the formula (4) The calculation of (e _{t, p} (underscore)), for example, the PEP in consideration to the position where the error pattern occurs on a trellis, such as Non-Patent Document 3 or 4 Calculation Use. Moreover, when applying the calculation method of PEP which can not consider the position on a trellis (for example, the following publicly known document 4), the same PEP is substituted into Formula (4) for all t for every error pattern. Thus, FER can be calculated. Therefore, Formula (4) is not limited by the calculation method of PEP.

Known Literature 4: G. Caire, G. Taricco, and E. Biglieri, "Bit-interleaved coded modulation," IEEE Trans. Inf. Theory, vol. 44, no. 3, pp. 927-946, May 1998.
FIG. 7 is a conceptual diagram showing the concept of equation (4).

  In FIG. 7, in the trellis section corresponding to the frame size, different error events which may occur at different positions are shown with different thicknesses.

  In practice, for example, there are a plurality of error events starting from the position of t = 1.

  Therefore, up to which range of error events to consider is set in advance. Although not particularly limited, for example, an event to be considered is selected by considering only an error event whose Hamming distance is a certain value or less.

In the following, in this way, only error events whose true path and error path Hamming distances are equal to or less than the distance d _H are considered.

  For these error events, PEP is calculated for each position in the frame and summed to determine the probability that an error event will occur.

  In the following, the calculation of PEP when PEPs differ according to the position of the trellis will be described.

First, for example, in the case described in FIG. 3, the error pattern is [1, 1, 0, 1, 1, 1], which indicates an error path at d _H = 5 of the convolutional code.

Hereinafter, it will be representative of the PEP error paths p-th error pattern starting at position t in the trellis PEP and (e _{t, p} (underscore)).

According to Non-Patent Document 4, conditional PEPs of a trellis path (hereinafter, true path) which generates a code word of which configuration bits are all 0 and a path (hereinafter, error path) which generates a code word separated by Hamming distance d _H , Is represented by the following equation.

Error path e _{t, p} a (underscore) interleaves, the sub-carrier number w th error bit error pattern is mapped k _w, the intra-symbol bit positions for transmitting in the sub-carrier and i _w. Further, the distance represented by Δ _w (i _w ) is a square Euclidean distance between the signal point when there is no error and the signal point when there is an error bit, for the symbol to which the error bit is assigned. Also, H _k represents the frequency response of the k th subcarrier, and H _kw is the frequency response of the subcarrier to which the w th error bit is mapped. Furthermore, H (under bar) and Δ (under bar) are vectors having H _kw and Δ _w (i _w ) as elements, respectively.

  In this case, PEP is given by the following equation by obtaining expected values for H (under bar) and Δ (under bar).

Here, E {...} _a represents an expected value operation for a. In equation (5), assuming that the propagation path changes for each transmission frame, in order to derive post-decoding FER in the case of transmitting a plurality of frames, an expected value is obtained for the frequency response H _kw of the subcarrier. Also, Δ _w (i _w ) can be determined from the true path and error path, and the interleaver structure, but in practice the true path depends on the input bit of the encoder and Δ _w (i _w ) is a random variable Become. Therefore, in Equation (5), the expected value is also obtained for Δ _w (i _w ). Furthermore, the conditional PEP of equation (5) is given by the following equation.

According to the equation (5), in order to derive the PEP, an expected value of H _kw and Δ _w (i _w ) may be obtained for the equation (6). First, the molecule of formula (6) is modified as follows.

  Next, the eigenvalue decomposition of PA′P is as follows.

Here, λ _l is a nonzero eigenvalue of PA′P having the l-th largest Euclidean norm, v _l is the l-th element of v and its amplitude follows the Rayleigh distribution of the probability density function

  r is the diversity order. However, rank (...) represents the rank of the matrix, and min (...) represents the minimum value.

  Here, PEP can be said as follows.

  The transmitted bit string can be represented by one path on the trellis diagram, and when a path other than the transmitted bit string is selected on the receiving side, a post-decoding bit error occurs. On the receiving side, the metric of each path is calculated, and the path with the largest metric is selected. “PEP” is the probability that the metric of a certain error path exceeds the metric of the true path, and in the low FER region, the bit string corresponding to the error path asymptotically approaches the probability of obtaining a decoding result.

  The equation (12) is substituted into the equation (6) to obtain the following equation.

Similar to Non-Patent Documents 3 and 4, the following equation is obtained by obtaining an expected value for v _l .

Further, an expected value is obtained with respect to Δ _w (i _w ).
Δ _w (i _w ) is a finite discrete random variable having a probability density different depending on the bit position i _w in the symbol, and λ _l is an eigenvalue of PA _{P P,} so an error path e _{t, p} (under bar) If is given, it can be uniquely determined from Δ _w (i _w ).

Since Δ _w (i _w ) is a finite discrete random variable, λ _l is also a finite discrete random variable. Therefore, it becomes as follows.

Here, assuming that m _iw possible values of the discrete random variable Δ _w (i _w ), the following relationship is established.

In regions where the FER is low, it is expected that symbol errors will be dominated by errors in adjacent signal points because the SNR is sufficiently high (the magnitude of noise relative to the distance between symbols is small).
If the minimum Euclidean distance between the signal points and _{d min, w = 1, ...} , d H relative to Δ _w (i _w) = a d _min ² become term is dominant in the J term expectations Be done. Also, P _j of this dominant term is denoted as P _min . That is, P _min represents the probability of Δ _w (i _w ) = d _min ² for w = 1,..., D _H , and is determined from the modulation scheme, error path, and interleaver structure.

Further, λ _{l, min} is an eigenvalue of the matrix PA _{min ′} P calculated by the following equation.

  In this embodiment, the interleaver structure is known in advance on the transmitter side and the receiver side. Therefore, when the calculation of equation (15) is executed, on the transmitter side, information of power delay profile (required for calculation of matrix PA'P), noise power (required for calculation of .sigma.) And configuration of encoder configuration If there is (ie, information on MCS), it can be executed.

The total power part of equation (15) decreases as the diversity order r increases, and the PEP of the error path with a large diversity order becomes lower than the PEP of the error path with a small diversity order. However, assuming a wireless LAN, there is a possibility that d _H LL because the delay dispersion is small. At this time, the diversity orders of all error paths are equal to L, and there is a possibility that the PEPs of paths separated from the minimum free distance can not be ignored even in the region where the FER is low. Therefore, PEP of an error path separated from the true path by the Hamming distance d _f + φ is considered. Here, for the value of φ, an appropriate value is set in advance experimentally.

  If the PEP can be predicted by the above-described procedure, the predicted value of FER by the FER prediction unit 3030 can be executed by the following equation.

However, the summation with respect to p is performed when the Hamming distance of the error path is d _f + φ or less.

According to the configuration of the present embodiment, in a communication system using a convolutional code, it is possible to realize an FER prediction device capable of accurately predicting FER even in a situation where an error rate differs depending on the position of trellis. Further, by using such a FER prediction device, a wireless communication device and a wireless communication device capable of adaptively selecting an appropriate MCS and performing communication even when errors in the communication path can not be regarded as random errors. A communication system can be realized.
[Modification of Embodiment 1]
In the above description, in prediction of the frame error rate (FER), PEP is calculated at each position on the trellis corresponding to a frame.

  However, for example, in a communication system such as a wireless LAN, the same interleaving is repeatedly performed in the same frame.

  In such a communication system, when it is possible to consider that channel fluctuation is constant in a frame, channel fluctuation received by the deinterleaved coded sequence becomes periodic with an interleaving block length as a cycle.

  FIG. 8 is a conceptual diagram for explaining the PEP pattern when the same interleaving process is repeatedly performed on each block in the frame as described above.

As shown in FIG. 8, when the same interleaving process is repeatedly performed on each block in a frame, the influence of the propagation path on which each codeword error bit is received is periodic. As a result, block error rates become equal in the interleaving block cycle.
Using this, it is possible to reduce the operation cost and predict FER by limiting the section for calculating PEP to the section of the trellis corresponding to one block.

  FIG. 9 is a conceptual diagram for explaining the procedure of predicting FER by the operation in block units as described above.

  When the block is set so that the influence of the propagation path to which the coded sequence after de-interleaving is received becomes periodic with an interleave block length as a cycle, PEPs that are equal in the block cycle will be repeated.

_Assuming that a block period is L _block , the number of blocks in a frame is _Nitl, and PEP in the block is _Pitl , FER is expressed as the following equation.

That is, in calculating the prediction of FER, it is sufficient to _obtain the pairwise error rate _Pitl of one block, and the prediction can be performed with less calculation cost than calculating the PEP in the section for the frame size.

  For example, in the case of a wireless LAN, propagation path fluctuation in a frame is small, and interleaving is performed in units of OFDM symbols, so PEPs can be considered to be equal in a cycle corresponding to OFDM symbols. Therefore, the probability that an error occurs in each OFDM symbol can be regarded as the same, and it becomes possible to perform prediction of FER using the above equation.

For example, if the probability P _sym of occurrence of one or more error events in an OFDM symbol is calculated, FER of a system to be communicated by OFDM is expressed by the following equation according to the following equation.

L _frame is one frame in length, L _sym is the number of coding bits in an OFDM 1 symbol (= interleaving size), and R represents a coding rate. The right handed index of () represents the number of OFDM symbols for transmitting one frame of information bits.

However, more generally, the size of the interleaving block is not limited to the OFDM symbol unit, and if the same interleaving process is repeated over a plurality of interleaving blocks in one frame, such The PEP may be calculated for the block.
Second Embodiment
In the first embodiment, the PEP is calculated by the power delay profile. However, depending on the conditions of the propagation path, it is possible to calculate PEP from the frequency response of the subcarrier (equivalent to impulse response) as follows. In the second embodiment, the case where PEP is calculated from the frequency response of subcarriers will be described.

Non-Patent Document 4 assumes that the frequency response of the propagation path does not fluctuate within the OFDM symbol. In other words, it is possible to fluctuate at frame-by-frame or shorter intervals. However, in the wireless LAN environment, it is assumed that the movement of the wireless device is small to some extent, so there may be cases where propagation paths can be regarded as identical in a plurality of frames (the fluctuation of the propagation paths can be ignored within observation time). In such a case, the frequency response H _kw of the subcarrier (or the impulse response of the propagation path) can be regarded as a constant. Therefore, conditional PEP can be calculated by assuming that the frequency response (or impulse response) of the subcarrier does not change with time in Equation (6) and treating it as a constant. In this case, the expected value may be obtained only for Δ _w (i _w ) as follows with respect to equation (6).

  However, since the frequency response of the subcarrier and the impulse response of the propagation path can be equivalently converted as in equation (2), equation (17) can also be calculated from the impulse response.

Similar to the first embodiment, w = 1, ..., Δ w (i w) = d min 2 become claim respect d _f is expected to be dominant in the J terms. The P _j of this dominant term is denoted as P _min . Therefore, equation (17) is approximated as follows.

  Thus, conditional PEP may be calculated. Such processing enables prediction of FER by estimating subcarrier frequency response (or impulse response) instead of the power delay profile.

  The other processes are the same as those described in the first embodiment, and therefore the description will not be repeated.

  Therefore, in the present specification, the terms are used as follows.

  The “propagation path characteristic information” refers to information representing the attenuation of propagation path power, and typically means the power delay profile described above. Furthermore, as described above, when satisfying the condition that propagation path fluctuation can be ignored within the observation time, instantaneous information of the propagation path estimated in the communication path, that is, each subcarrier in the convolutional coding OFDM system It also includes the frequency response of (equivalently impulse response).

  As described above, according to the configuration of the present embodiment, it is possible to predict the frame error rate by accurately predicting the post-decoding BER even when errors in the communication path can not be regarded as random errors. Frame error rate prediction apparatus can be realized. Also, by using such a frame error rate prediction apparatus, a wireless communication apparatus capable of adaptively selecting an appropriate MCS and performing communication even when errors in the communication path can not be regarded as random errors. And a wireless communication system can be realized.

  The embodiment disclosed this time is an illustration of a configuration for specifically implementing the present invention, and does not limit the technical scope of the present invention. The technical scope of the present invention is indicated not by the description of the embodiment but by the scope of claims, and includes modifications within the scope of wording and meaning of the scope of claims. Is intended.

  1000 transmission apparatus, 1002 error correction coding section, 1004 interleaving section, 1010 modulation section, 1012 OFDM modulation section, 1014 RF section, 1020 antenna, 1100 reception processing section, 1110, 1110 'adaptive rate control section, 2000 reception apparatus, 2002 Antenna, 2010 RF unit, 2012 OFDM demodulator, 2014 demodulator, 2016 deinterleaver, 2018 error correction unit, 2100 power delay profile / noise power estimation unit, 2110 transmission processing unit, 3010 BER prediction unit, 3020, 3050 MCS storage Part, 3030 FER prediction part, 3040 MCS selection part.

Claims (11)

A frame error rate prediction apparatus of a communication system which communicates by a convolutional coding scheme that can be decoded by a trellis diagram, and in which the same interleaving process is repeatedly performed on transmission bits in a frame,
Storage means for storing information of a combination of a coding rate of convolutional coding and a modulation method which are preset as adopted in the convolutional coding method;
The convolutional coding for each of the pair of the coding rate and the modulation scheme stored in the storage means, based on the propagation path characteristic information representing the attenuation of the propagation path power and the noise power estimated in the communication path Calculate the pairwise error rate of the error path in the decoding process for each of the positions of the section of the trellis diagram corresponding to the size of the frame, and predict the frame error rate as the sum of the pairwise error rates calculated at each of the positions. A frame error rate prediction device comprising: a frame error rate prediction means.   The frame error rate predicting means is configured to divide the transmission bit into block lengths set such that propagation path fluctuation with respect to a coded sequence after de-interleave processing can be regarded as periodic in the frame. Calculating a block error rate as a sum of the pairwise error rates in a section of the trellis diagram corresponding to the one block, and assuming that the block error rates are equal to each other, the block error rate and the number of blocks included in the frame The frame error rate predicting apparatus according to claim 1, wherein the frame error rate is predicted from.   3. The frame error rate predicting apparatus according to claim 2, wherein the communication system performs communication according to orthogonal frequency division multiplexing, and the block is one symbol in orthogonal frequency division multiplexing.   The frame error rate predicting apparatus according to claim 1, wherein the channel characteristic information is a power delay profile estimated in the communication path.   The frame error rate predicting apparatus according to claim 1, wherein the channel characteristic information is an impulse response of the channel. A wireless communication apparatus that communicates by a convolutional coding method, comprising:
Receiving means for receiving, as feedback information, propagation path characteristic information representing attenuation of propagation path power and noise power estimate transmitted from the receiving apparatus;
Storage means for storing information of a combination of a coding rate of convolutional coding and a modulation method which are preset as adopted in the convolutional coding method;
And selecting means for selecting a set that maximizes throughput among the sets of the coding rate and the modulation scheme stored in the storage means based on the feedback information.
The selection means is
The pairwise error rate of the error path in the decoding process for the convolutional coding is calculated for each of the pair of the coding rate and the modulation scheme stored in the storage means based on the propagation path characteristic information and the noise power. Frame error rate predicting means for calculating a frame error rate as a sum of the pairwise error rates calculated at each position of the section of the trellis diagram corresponding to the size of the frame;
And d) modulation scheme selection means for selecting the set that achieves the maximum throughput within a predetermined frame error rate based on the predicted frame error rate.
A wireless communication apparatus, further comprising transmission means for performing convolutional coding and modulation processing on transmission data according to the set selected by the modulation scheme selection unit and transmitting the transmission data.   The frame error rate predicting means is configured to divide the transmission bit into block lengths set such that propagation path fluctuation with respect to a coded sequence after de-interleave processing can be regarded as periodic in the frame. Calculating a block error rate as a sum of the pairwise error rates in a section of the trellis diagram corresponding to the one block, and assuming that the block error rates are equal to each other, the block error rate and the number of blocks included in the frame The wireless communication apparatus according to claim 6, wherein the frame error rate is predicted from.   The wireless communication apparatus according to claim 7, wherein the communication system performs communication according to orthogonal frequency division multiplexing, and the block is one symbol in orthogonal frequency division multiplexing.   The wireless communication apparatus according to claim 6, wherein the channel characteristic information is a power delay profile estimated in the communication path.   The wireless communication apparatus according to claim 6, wherein the channel characteristic information is an impulse response of the channel. A wireless communication system for communicating according to a convolutional coding OFDM system, comprising:
Equipped with a receiver,
The receiving device is
In the communication path, propagation path characteristic information representing attenuation of power of the propagation path and communication path state estimation means for estimating noise power;
First transmission means for transmitting the estimated channel characteristic information and noise power;
Further comprising a transmitter;
The transmitting device is
Receiving means for receiving, as feedback information, the channel characteristic information and the estimated value of the noise power transmitted from the receiving apparatus;
Storage means for storing information of a combination of a coding rate of convolutional coding and a modulation scheme set in advance as employed in the convolutional coding OFDM system;
Selecting means for selecting a set that maximizes throughput among the sets of the coding rate and the modulation scheme stored in the storage means based on the feedback information;
The selection means is
The pairwise error rate of the error path in the decoding process for the convolutional coding is calculated for each of the pair of the coding rate and the modulation scheme stored in the storage means based on the propagation path characteristic information and the noise power. Frame error rate predicting means for calculating a frame error rate as a sum of the pairwise error rates calculated at each position of the section of the trellis diagram corresponding to the size of the frame;
Modulation scheme selecting means for selecting the set that achieves the maximum throughput within a predetermined frame error rate based on the predicted frame error rate;
A wireless communication system, further comprising: second transmission means for performing convolutional coding and modulation processing on transmission data and transmitting the transmission data according to the set selected by the modulation scheme selection unit.