JP2006115221A - Line estimation method and reception device using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To ensure a good reception error rate and to reduce the arithmetic processing load of line estimation. <P>SOLUTION: A candidate section detection part 121 of a line estimation means 120 narrows down candidate sections by detecting peaks of channel impulse responses of transmission lines on the basis of a reception sequence R(t). An error operation part 122 targets only narrowed-down sections to calculate square errors of section lengths (Tap_Length) of candidate sections detected by the candidate section detection parts 121. A selection part 123 selects a section length giving the minimum square error (J_min) out of section lengths calculated by the error operation part 122. A maximum likelihood sequence estimation part 130 outputs an estimated transmission sequence T'(t) on the basis of the reception sequence R(t), a channel impulse response (Estimated_CIR) outputted from the selection part 123, and the section length. Thus the processing load of line estimation can be reduced by calculating square errors of only narrowed-down candidate sections. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a receiving device such as a mobile portable terminal device for digital communication provided with a channel estimation response and channel estimation means for estimating an equalization interval used in an adaptive equalizer.

  In recent years, technological development for performing high-speed digital transmission in mobile communication and the like has been actively performed.

  As digital mobile communication systems, GSM (Global System for Mobile Communications) and IS-54 (TIA TDMA Air Interface Standard), PCD, PHS, and other time division multiplex (TDMA) communication systems are known. Yes.

  In such a communication system that performs high-speed digital transmission, a relatively large transmission line delay occurs with respect to the symbol period. Therefore, the bit error rate does not improve even when the received power is increased. "Characteristic" occurs.

  Since this floor characteristic causes a deterioration of the demodulation characteristic, adaptive equalization for removing waveform distortion due to frequency selective fading is indispensable in a TDMA communication system.

  Known adaptive equalizers used in the TDMA system include a decision feedback equalizer (DFE) and a maximum likelihood sequence estimator (MLSE). The maximum likelihood sequence estimator is generally called a Viterbi equalizer (see, for example, Non-Patent Document 1). These adaptive equalizers are composed of transversal filters, and their Tap coefficients are determined by the channel impulse response (CIR) from the channel estimator described below so as to maximize the received S / N. Operate.

  Here, the channel estimation means used in the above adaptive equalizer will be described taking the European GSM system as an example. The channel estimation means estimates the symbol synchronization and the equalization interval length by estimating the transfer function expressed by the channel impulse response of the propagation path using the midamble autocorrelation characteristics of the received sequence. Then, the influence of level fluctuation due to fading is mitigated by performing a convolution operation on the reciprocal of the estimated transfer function and the received sequence.

  FIG. 5 is a block diagram showing the basic configuration of the channel estimation means used in the maximum likelihood sequence estimator. FIG. 6 is a diagram illustrating a configuration of a normal burst (NB) in the European GSM system.

  As shown in FIG. 5, the channel estimation means 50 includes convolution calculators 51 and 52, a detection / squeezing unit 53, and a square error calculation unit 54.

  In FIG. 5, a convolution calculator 51 performs a convolution operation between a received sequence and an expected training sequence (TSC: training sequence), and calculates a channel impulse response he (t) from the autocorrelation characteristics of the training sequence.

  The detection / narrowing unit 53 detects the peak value (CIR peak) of he (t), and generates candidates {he_1 (t),... He_n (t)} up to an interval including an effective path. Next, the convolution calculator 52 performs a convolution operation between the channel impulse response of the candidate {he_1 (t),... He_n (t)} and the training sequence of the expected value for narrowing down.

  The square error calculator 54 calculates an error J = J_1 in each channel impulse response candidate he_1 (t) based on the received sequence and the output sequence of the convolution calculator 52, and sends the error J = J_1 to the detection / narrowing unit 53. This process is repeated for he (t) = {he_1 (t) ·· he_n (n)}, and in the detection / restriction unit 53 in which J = {J1 ·· Jn} is obtained, {J1 ·· Jn} The minimum value J_min and he (t) that gives it are determined. The impulse response he (t) that takes the minimum value (J_min) of the errors J in each channel impulse response candidate calculated by the square error calculator 54 is a result that is very close to the true impulse response of the transmission path.

  In the GSM system, as shown in FIG. 6, the training sequence (TSC) is located at the center of the normal burst because of the midamble format. At the time of receiving and demodulating the normal burst, the receiving apparatus uses a known sequence informed by the broadcast channel among the eight types of training sequences.

In the channel estimation means 50 shown in FIG. 5, when performing channel estimation, the convolution calculators 51 and 52 perform convolution operations on all timing candidates of a known training sequence and a training sequence that is a candidate for a received training sequence. Estimate delay profile consisting of main wave and effective delay wave from inside, and output estimated channel impulse response (Estimated_CIR) and interval length (Tap_Length) required for equalization to maximum likelihood sequence estimator (MLSE) .
Gottfried Ungerboeck, `` Adaptive Max-imum-Likelihood Receiver for Carrier-Modulated Data-Transmission Systems '', IEEE Transactions on Communications, Vol.COM-22, No.5, pp624-636, May 1974

  However, in the channel estimation means in the conventional receiver, an impulse is obtained by performing a least square error calculation in a combination of all equalization interval length candidates and a timing candidate consisting of the sum of the reception window width and the training sequence code. Seeking a response.

  For this reason, the conventional receiving apparatus has a problem that the processing load due to the processing of the error calculation and the minimum value search increases.

  Also, in conventional receivers, the number of received bursts to be demodulated in one frame is doubled or quadrupled in order to support a high rate, as in GPRS (General Packet Radio Service) multislot. ... and will increase.

  For this reason, in a receiving apparatus such as a conventional mobile portable terminal apparatus, it is a problem to reduce the processing amount without causing performance degradation when the channel estimation means is installed.

  The present invention has been made in view of the above points, and is a channel estimation method capable of ensuring a good reception error rate and reducing the calculation processing load of channel estimation, and reception of a mobile portable terminal using the channel estimation method. An object is to provide an apparatus.

  In order to solve this problem, the receiving apparatus of the present invention includes channel estimation means for outputting a channel impulse response and section length of a transmission path based on a received sequence, and channel impulses output by the received sequence and the channel estimation means. Maximum likelihood sequence estimation means for outputting an estimated transmission sequence based on the response and the section length, and the channel estimation means detects a peak of the channel impulse response of the transmission path based on the received sequence, and A candidate section detecting means for detecting a candidate section by setting a threshold value of a valid candidate path according to a peak value and applying the threshold value to a channel impulse response to narrow down a valid candidate path, and detection by the candidate section detecting means An error calculating means for calculating a square error only for the section length of the candidate section, and the section length calculated by the error calculating means. A configuration comprising a selection means for selecting a segment length that gives the smallest square error.

  Here, the following three methods (1) to (3) can be cited as methods for reducing the processing load of the channel estimation means.

  (1) The effective path candidate determination threshold (Tap_th) is introduced, and the effective path candidate is determined by applying Tap_th to the result of the convolution operation between the training sequence and the expected value training sequence in the received sequence. As a result, as shown in FIG. 2 or FIG. 3, the effective candidate paths are narrowed down, and the least square error calculation and the minimum value search are performed on the section.

  (2) When the effective path is determined to be a long path channel as shown in FIG. 2 in the method (1), the convolution operation and the least square error operation are performed from the maximum equalization interval to a predetermined minimum value. To do.

  (3) In addition to (1) and (2), the frame likelihood value (Frame Likelihood F) output after impulse response estimation is used, and the likelihood satisfies a predetermined threshold (Frame Likelihood F_Th). In the case where there is no noise, that is, in a state where the SNR is measured, the increase in the probability of outputting an estimation result in which noise is regarded as an effective path is considered, and the effective path candidate determination threshold (Tap_th) is increased to increase the noise. The probability of erroneous determination that a path due to power is regarded as an effective path is reduced.

  According to the present invention, it is possible to secure a good reception error rate and reduce the processing load of channel estimation.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the drawings, components having the same configuration or function and corresponding parts are denoted by the same reference numerals and description thereof will not be repeated.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of a communication system using the receiving apparatus according to Embodiment 1 of the present invention.

  The receiving apparatus 110 according to the first embodiment includes a receiving antenna 111, an RF receiving unit 112, an LPF (low-pass filter) 113, an A / D (A / D converter) 114, a storage memory 115, a channel estimation unit 120, A likelihood sequence estimation unit 130 is provided.

  The line estimation unit 120 includes a candidate section detection unit 121, an error calculation unit 122, and a selection unit 123.

  The reception antenna 111 receives radio waves radiated from a transmission antenna 103 described later. The reception RF unit 112 converts the frequency of the received signal in the carrier wave band received by the reception antenna 111 into the baseband. An LPF (low-pass filter) 113 limits the band of the received signal frequency-converted to the baseband by the reception RF unit 112. An A / D (A / D converter) 114 converts a received signal in a band limited by an LPF (low-pass filter) 113 into a digital value. The accumulation memory 115 accumulates the reception signal output from the A / D (A / D converter) 114 by the reception window width, and outputs a reception sequence R (t) in equalization processing units.

  The channel estimation means 120 estimates the Tap length from the channel impulse response and the effective path length based on the received sequence R (t) and the known training sequence (TSC_expect). The maximum likelihood sequence estimation unit 130 uses the Viterbi algorithm, the received sequence R (t) output from the storage memory 115, the channel impulse response (Estimated_CIR) obtained by the channel estimation means 120, and the Tap length. A transmission prediction sequence T ′ (t) is estimated from (Tap_Length).

  In FIG. 1, a baseband transmission sequence T (t) is modulated by a modulation unit 101. The output of modulation section 101 is up-converted to a carrier band in RF transmission section 102 for carrier transmission. The output of the RF transmission unit 102 is radiated into the air by the transmission antenna 103.

  The baseband signal obtained via the reception antenna 111 and the RF reception unit 112 is digitized into a complex IQ reception sequence via an LPF (low-pass filter) 113 and an A / D (A / D converter) 114, Once stored in the storage memory 115.

  Thereafter, a received sequence R (t) having a length necessary for processing by channel estimation means 120 and maximum likelihood sequence estimation section 130 is input from accumulation memory 115 to channel estimation means 120 and maximum likelihood sequence estimation section 130.

  The channel estimation means 120 uses the autocorrelation characteristic of the normal burst midamble (training sequence) shown in FIG. 6 and the training sequence in the received sequence R (t) and the known training sequence (TSC_expect). The peak value is obtained by convolution calculation between and the channel impulse response. Then, the estimated channel impulse response (Estimated_CIR) and the section length (Tap_Length) are output.

  Maximum likelihood sequence estimation section 130 estimates a likely data sequence from reception sequence R (t), channel impulse response (Estimated_CIR), and section length (Tap_Length), and outputs estimated transmission sequence T ′ (t).

  In receiving apparatus 110 according to Embodiment 1, channel estimation that can reduce the processing load is realized by the above method. Here, the GSM system that is most widely used among digital cellular telephones will be described as an object, but the present invention is not limited to this. The standard specification of the GSM system is described in the ETSI / GSM Series 03 Air Interface Specification, GSM PN Paris, and will not be described in detail.

  Hereinafter, the processing method in the channel estimation means 120 of this example will be described in detail. Here, the GSM system will be described based on the normal burst shown in FIG. 6, but the processing method of this example can also be applied to synchronization (SB).

  The present invention can also be applied to a preamble type TDMA burst as shown in FIG.

  The processing method of this example can also be applied to all receiving apparatuses that use an adaptive equalizer that performs channel estimation using the correlation characteristics of a training sequence.

  In the channel estimation means 120 of this example, first, the peak detection of the channel impulse response is performed in the candidate section detection unit 121 as a first step.

The reception sequence {Ri} of the channel estimation means 50 shown in FIG. 5 can be expressed as the following equation (1) by the transmission sequence {Si} and the transmission path {hc (t)}. In Expression (1), a symbol in which “*” is surrounded by “◯” indicates a convolution operation.

  The received sequence {Ri} shown in the equation (1) is first subjected to a broadcast pattern midamble and convolution operation in the channel estimation means 120. As a result, a finite-length channel impulse response as shown in FIG. 2 or FIG. 3 is obtained.

Next, as shown in FIGS. 2 and 3, the effective path candidate determination threshold (Tap_th) is applied to the channel impulse response, and the first path exceeding the effective path candidate determination threshold (Tap_th) is started. And the maximum power | P | of the absolute value of the path is searched. Next, the number of peaks exceeding the threshold value obtained by multiplying the value | P | by a predetermined effective path candidate determination threshold value (Tap_th) is counted. Here, the effective path candidate determination threshold (Tap_th) is given as a value obtained by multiplying the absolute value | Pmax | of the power P of the path with the largest power P by the coefficient Tap_th, as shown in the equation (2). It is done.

  Next, as a second step, a section length (Tap_Length) candidate is calculated. If a section including a path having a value equal to or greater than the effective path candidate determination threshold (Tap_th) is represented as a section length (Tap_Length) by an integral multiple of the symbol length T, for example, in the case of a GSM system, the longest section length is “6T” or “5T” (T: symbol period) is a realistic number. However, the present invention is not limited to these section lengths. In the following description, the section length is described as being the maximum “5T” and the minimum “3T”, but the operation in the channel estimation unit 120 of this example is the same with “6T”. The candidate for the section length estimated by the candidate section detector 121 is sent to the next error calculator 122.

  Next, in the error calculation unit 122, if the section length candidate is “5T”, {5, 4, 3} is a section length (Tap_Length) candidate. When the section length candidate is “4T”, {4, 3} is a section length (Tap_Length) candidate. Further, when the section length candidate is “3T”, {3} is a section length (Tap_Length) candidate. Then, the square error for each number of candidates for each section length is obtained and sent to the selection unit 123.

  Next, as a third step, the selection unit 123 narrows down the section length (Tap_Length). In the selection unit 123, narrowing-down is performed by searching for the minimum value of each square error based on the square error sent from the error calculation unit 122. When the number of peaks in which the absolute value | P | of the path power P in the candidate section detection unit 121 exceeds the valid path candidate determination threshold (Tap_th) is “5” or more, the section length (Tap_Length) is { Narrowing is performed with 5,4,3} as candidates. If the number of peaks is “4”, the section length (Tap_Length) is {4, 3} as a candidate, and if the number of peaks is “3” or less, the section length (Tap_Length) is Narrowing is performed as {3}.

  Here, if the number of candidate peaks is “5” or more, first, the section length (Tap_Length) = 5 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. Is called. The square error between the calculation result and the reception sequence is “J5”.

  Next, the section length (Tap_Length) = 4 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. The square error between this result and the received sequence is “J5”.

  Next, the section length (Tap_Length) = 4 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. The square error between this result and the received sequence is “J4”.

  Next, a section length (Tap_Length) = 3 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. The square error between this result and the received sequence is “J3”.

  Next, the minimum value search of each square error “J5”, “J4”, “J3” is performed, the result of the minimum value is “Jmin”, the section length at that time is identified as (Tap_Length), and the section length (Tap_Length) is The channel impulse response he (t) is obtained, and the inverse characteristic thereof is input to the maximum likelihood sequence estimation unit 130 as (Estimated_CIR).

  Next, when the number of candidate peaks is “4”, first, the section length (Tap_Length) = 4 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. . The square error between the calculation result and the reception sequence is “J4”.

  Next, a section length (Tap_Length) = 3 is provisionally determined, and a convolution operation is performed between the corresponding channel impulse response and the broadcast training sequence. The square error between this result and the received sequence is “J3”.

  Next, the minimum value search of each square error “J4” and “J3” is performed, the result of the minimum value is “Jmin”, the section length at that time is identified as (Tap_Length), the section length (Tap_Length) and its channel impulse response he (t) is obtained, and its inverse characteristic (Estimated_CIR) is input to the maximum likelihood sequence estimation unit 130.

  When the number of peaks is “3”, “J_min” is equal to the preset minimum value of the set section length (Tap_Length) = 3. The inverse characteristics of the section length (Tap_Length) and the channel impulse response he (t) obtained here are input to the maximum likelihood sequence estimation unit 130 as (Estimated_CIR).

  Here, the estimated range of the section length (Tap_Length) has been described as “3” to “5”, but is not limited to this range.

(Embodiment 2)
Next, the receiving apparatus according to Embodiment 2 of the present invention will be described. FIG. 4 is a block diagram showing a configuration of a communication system using the receiving apparatus according to Embodiment 2 of the present invention.

  As shown in FIG. 4, receiving apparatus 410 according to the second embodiment has the same configuration as receiving apparatus 110 according to the first embodiment, except that the configuration of channel estimation means 420 is different.

  That is, channel estimation means 420 of receiving apparatus 410 according to Embodiment 2 is different from candidate section detecting section 121 and selecting section 123 of receiving section 110 of receiving apparatus 110 according to Embodiment 1 as candidate section detecting section 421 and selecting section. A portion 422 is provided.

  Selection section 422 of channel estimation means 420 in receiving apparatus 410 according to Embodiment 2 newly introduces moving average value M based on past samples of frame likelihood. Then, the frame likelihood and the frame likelihood threshold (as instantaneous fluctuations) are considered in consideration of the reception state by comparing the moving average value M, which is the estimation result of the reception state, with the state estimation threshold (frame_likelihood_S_th). frame_likelihood_th). The frame likelihood F is obtained from the square sum of the estimated channel impulse response (Estimated_CIR) with a square error J_min (“J_min” with the minimum square error input to the selection unit 422). Here, the larger the frame likelihood F and its moving average value M, the higher the likelihood.

Based on these condition comparison results, the effective path candidate determination threshold (Tap_th) in the next frame is determined as shown in the following Table 1 (indicated by “th” in Table 1).

  In Table 1, “th_L”, “th_M”, and “th_H” of the effective path candidate determination threshold (Tap_th) are coefficients for the absolute value of the maximum path shown in Expression (2). Here, each determined valid path candidate determination threshold (Tap_th) satisfies 0 <Pth_L ≦ Pth_M ≦ Pth_H <1.

  On the other hand, the candidate section detection unit 421 uses the “th” of the valid path candidate determination threshold (Tap_th) determined as shown in Table 1, as in the receiving apparatus 110 according to Embodiment 1, and Performs processing in bursts.

  As a result, in receiving apparatus 410 according to Embodiment 2, the effective path candidate determination threshold (Tap_th) can be set higher in a situation where the reception level is reduced due to multipath fading. The probability of erroneously selecting the short path channel as the long path channel can be reduced by the path according to.

  Note that “=” in Table 1 may be included in any of the inequality signs.

  The channel estimator of the present invention can also be applied to a transmission system having a preamble-type training sequence as shown in FIG.

  Since the receiving apparatus according to the present invention can secure a good reception error rate and reduce the processing load, it is useful as a receiving apparatus such as a mobile portable terminal device for digital communication.

The block diagram which shows the structure of the communication system using the receiver which concerns on Embodiment 1 of this invention. The figure for demonstrating the channel impulse response in a receiver The figure for demonstrating the other channel impulse response in a receiver FIG. 3 is a block diagram showing a configuration of a communication system using a receiving apparatus according to Embodiment 2 of the present invention. Block diagram showing the basic configuration of channel estimation means used in the maximum likelihood sequence estimator The figure which shows the structure of the TDMA burst in a midamble format The figure which shows the structure of the TDMA burst in a preamble format.

Explanation of symbols

110,410 Receiver 111 Receiver antenna 112 RF receiver 113 LPF (low pass filter)
114 A / D (A / D converter)
115 Storage Memory 120, 420 Line Estimation Unit 121, 421 Candidate Section Detection Unit 122 Error Calculation Unit 123, 422 Selection Unit 130 Maximum Likelihood Sequence Estimator

Claims (8)

A channel estimation means for outputting the channel impulse response of the transmission path and the section length based on the received sequence;
Maximum likelihood sequence estimation means for outputting an estimated transmission sequence based on the received sequence and the channel impulse response and section length output by the channel estimation means,
The line estimation means includes
Based on the received sequence, the channel impulse response peak of the transmission path is detected, the threshold value of the effective candidate path is set according to the peak value, and the threshold value is applied to the channel impulse response to narrow down the effective candidate path. Candidate section detecting means for detecting candidate sections,
Error calculation means for performing a square error calculation only for the section length of the candidate section detected by the candidate section detection means;
A receiving apparatus comprising: selection means for selecting a section length that gives the smallest square error among the section lengths calculated by the error calculation means. Maximum likelihood based on a correlation value obtained by convolution calculation of an expected value between a predetermined first training sequence included in a reception sequence received through a transmission line and digitally modulated and a known second training sequence broadcast in advance A channel estimation means for estimating the channel impulse response of
The line estimation means includes
Correlation calculating means for performing a correlation calculation between the first training sequence and the second training sequence;
The maximum correlation peak value is obtained from the peak value of the correlation calculation result sequence by the correlation calculation means, and the number of peak values exceeding the predetermined threshold value is counted to determine equalization interval candidates. A segmentation section candidate determination means;
The channel impulse in the section length is determined by determining the length of the equalization according to the correlation calculation sequence by the correlation calculation means at the timing of the maximum correlation peak value and the number of the correlation peak values exceeding the threshold value. Search for the square error of the response, reduce the interval length by 1 and repeat the search for the square error of the channel impulse response to obtain an interval length that gives the smallest square error, and obtain an interval length that minimizes the square error An estimation processing means for estimating a channel impulse response.   When the frame likelihood is smaller than a predetermined threshold based on the frame likelihood calculated based on the minimum square error identified by the estimation processing means and the square sum of the channel impulse response. The receiving apparatus according to claim 2, wherein the threshold value of the next frame is controlled to a predetermined value.   A minimum square error identified by the estimation processing means, a frame likelihood calculated based on the sum of squares of the channel impulse response, and a likelihood of a reception state obtained from a past average value of the frame likelihood; 4. The receiving apparatus according to claim 2, wherein the threshold is controlled when the frame likelihood is smaller than a predetermined threshold and the likelihood of the reception state is low.   The receiving apparatus according to claim 4, wherein a past average value of the frame likelihood is calculated by a moving average.   The receiving apparatus according to claim 4, wherein the threshold value is controlled for each reception burst. A channel estimation step for outputting the channel impulse response and section length of the transmission path based on the received training sequence,
A maximum likelihood sequence estimation step of outputting an estimated transmission sequence based on the received training sequence and the channel impulse response and interval length output in the channel estimation step;
The channel estimation step includes a candidate section detection step for detecting a candidate section by performing peak detection of a channel impulse response of a transmission path based on the received sequence, and a section length of the candidate section detected in the candidate section detection step 7. An error calculation step for performing a square error calculation, and a selection step for selecting a section length that gives the smallest square error among the section lengths calculated in the error calculation step. A channel estimation method in the receiving apparatus according to claim 1. In the channel estimation step, the received training sequence and the channel impulse response and interval length output in the channel estimation step are set as the longest interval in the error calculation step for performing the square error calculation in a direction to narrow down the interval length. 8. The channel estimation method according to claim 7, further comprising an error calculation step for performing the square error calculation.

Images