JP2006197349A - Receiver for ultrawide bandwidth radio system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a receiver for an ultrawide bandwidth radio system wherein a transmission error rate is enhanced. <P>SOLUTION: Adopting the DDFSE sequence estimation for a Viterbi decoder reduces the number of states of an equalizer for the MBOK DS-UWB and trying the reduction in the number of states taking into account data mapping of the MBOK DS-UWB configures the equalizer with less computation complexity. The receiver employs the M-ary Bi-orthogonal Keying system provided with: a pulse correlation unit taking a correlation between a received signal and a transmission pulse waveform; a rake synthesizer for recovering the energy distributed in multi-paths; a spread sequence correlation unit for taking correlation between an output of the rake synthesizer and the spread sequence used for the communication; the Viterbi decoder for suppressing errors included in an output of the spread sequence correlation unit; an M-ary Bi-orthogonal Keying decoder for decoding an output of the Viterbi equalizer and providing an output; and a communication path estimation unit for estimating a state of a communication path including a multi-path environment in the communication path. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a receiver for an ultra wide bandwidth (UWB) wireless system with improved transmission error rate.

  Ultra Wide Bandwidth (UWB) wireless communication is attracting attention as wireless communication capable of low power consumption and high-speed data transmission in a short distance. Wireless communication using such a UWB signal includes a PHY (physical layer) for high-speed data transmission in a wireless personal area network (WPAN). For example, a direct sequence UWB (DS-UWB) system using a root-raised cosine pulse signal or a multi-band orthogonal frequency division multiplex (MB) -OFDM) and the like are known.

  The DS-UWB system is attracting attention as a system capable of performing high-speed data transmission up to over 1 Gbps with a simple configuration, but has a drawback of being easily affected by multipath. For example, in the DS-UWB system that is normally used, UWB pulse signals that are continuous at intervals of about 1 nsec are transmitted. In general, the multipath delay time in the UWB transmission line reaches about 60 nsec, so that interpulse interference occurs in which the delayed pulse signal interferes with other pulse signals. Therefore, it is indispensable to adopt an inter-pulse interference cancellation technique on the receiving side, and for this purpose, a decision feedback equalizer (DFE) or the like is used.

  In the DS-UWB system, multi-coding or shortening of the spread code is performed in order to increase the speed according to the data rate. Among them, as a method for performing the multi-coding, multi-dimensional orthogonal keying (M-ary bi-orthogonal keying: MBOK) is used. The DS-UWB (MBOK DS-UWB) method using MBOK can maintain a spreading gain even when performing high-speed data transmission, but it is complicated in the equalizer used on the receiving side by multi-coding. The degree will increase.

FIG. 1 shows an MBOK DS-UWB information mapping method. In MBOK, it is possible to transmit 1 og ₂ M-bit information in one symbol by selecting one spreading sequence from among M spreading sequences. In this way, multi-level processing is performed by increasing the number of spreading sequences M, thereby realizing high-speed data transmission. In MBOK, if a set of M spreading sequences is C, half of the M / 2 spreading sequences are other polarity-inverted spreading sequences as follows.

Here, each spreading sequence has a length L _c and is expressed as follows.

ここで、ｉ番目のシンボルとして選択された拡散系列を

Here, the spreading sequence selected as the i-th symbol is

Then, the transmission signal is obtained by replacing each chip of the spreading sequence with a UWB pulse signal p (t) having a time width T _p as follows.

Here, N _s is the number of symbols, T _c is the T _c = L _c × T _p represents a symbol interval.

  In general, a multipath channel is modeled with discrete samples as follows.

Here, L represents the total number of paths, and h _l and τ _l represent the gain and delay time of each path.

  At this time, the received signal is represented by the next convolution of the transmission signal x (t) and the response h (t) of the communication path.

Here, n (t) is AWGN (additive white Gaussian noise) having an average of 0 and a variance σ ² .

D. Cassioli, M. Z. Win, F. Vatalaro, and A. F. Molish, "performance of low-complexity Rake reception in a realistic UWB channel," Proc. IEEE ICC'03, vol.2, pp.763-767, 2003.

  As described above, the DS-UWB method using MBOK can maintain the spreading gain even when performing high-speed data transmission, but the complexity of the equalizer used on the receiving side can be increased by multi-coding. It will increase.

  In the present invention, the number of states of an equalizer for MBOK DS-UWB is reduced by adopting DDFSE (Delayed Decision Feedback Sequence Estimation) as a sequence estimation algorithm in Viterbi equalization, and in addition, data mapping of MBOK DS-UWB By trying to reduce the number of states in consideration of the above, an equalizer with a small amount of calculation is configured.

  By applying DDFSE to MBOK DS-UWB receivers in ultra-wideband wireless communications, and further reducing the number of states using the features of MBOK, the amount of Viterbi equalization in MBOK DS-UWB can be reduced. it can. By reducing the number of states, the metric calculation amount in the Viterbi equalizer can be reduced with a loss of 0.5 dB or less.

  A receiver for an ultra-wideband radio system according to the present invention is a receiver for a multi-element orthogonal keying communication device that uses an ultra-wideband radio signal having a bandwidth by spread modulation larger than a bandwidth by a data-modulated signal. Correlation between the received signal and the pulse correlator that correlates the transmitted pulse waveform, the Rake synthesizer for recovering energy distributed by multipath, and the correlation between the Rake synthesizer output and the spreading sequence used for communication A spreading sequence correlator, a Viterbi equalizer for suppressing errors included in the output of the spreading sequence correlator, a multi-factorial orthogonalization keying decoder that decodes and outputs the output of the Viterbi equalizer, And a communication path estimator for estimating a communication path state including a multipath environment in the communication path.

  In the receiver for the ultra wideband wireless system, the Viterbi equalizer regards the communication path having multipath characteristics as an encoder having a convolutional code constraint length equal to a predetermined number of delay symbols. The equalization is performed by performing an operation equivalent to the operation of equalization.

  In the receiver for the ultra-wideband wireless system described above, in the branch metric evaluation in Viterbi equalization, based on DDFSE, interference from delay symbols exceeding the specified value is detected in advance by decision feedback estimation from the output of the spread sequence correlator. The branch metric is calculated by the maximum likelihood sequence estimation for the resulting value.

  In addition, in order to reduce the amount of computation in Viterbi equalization, when performing survival path selection in the maximum likelihood sequence estimation unit in DDFSE, it is necessary to consider in subsequent state transitions by performing positive / negative polarity determination of each spread sequence together The number of states can be reduced, and the amount of calculation can be reduced.

Hereinafter, embodiments of the present invention will be described in detail along the MBOK DS-UWB system. The MBOK DS-UWB information mapping method is the same as described above, and is shown in FIG. That is, by selecting one spreading sequence from M spreading sequences, 1 og ₂ M-bit information is transmitted in one symbol. If the set of M spreading sequences is C, as shown in FIG. 1, half of the M / 2 codes are codes obtained by reversing the polarity of the other codes.

  FIG. 2 shows a receiver configuration example of the present invention. After correlating at the pulse signal level with the pulse correlator 1, the rake synthesizer 2 performs rake synthesis, and then the spread sequence correlator 3 with M / 2 spread sequences from c1 to cM / 2. Get correlation. Viterbi equalization by the Viterbi equalizer 4 is performed at the symbol level, decoded by the MBOK decoder 5 and output. By adopting Viterbi equalization at the symbol level, it is possible to ease the demand for processing speed. Below, it demonstrates along the flow of a signal.

  The received signal r (t) is subjected to Rake synthesis at the pulse signal level after correlation with the pulse signal p (t). Regarding the correlation with the pulse signal p (t), in the signal corresponding to the j-th chip of the i-th symbol, the correlation output is

となる。また、Ｒａｋｅ合成出力は次のように与えられる。

It becomes. The Rake composite output is given as follows.

ここでα_fとτ_fは、ｆ番目のフィンガにおける合成重みと遅延量を表す。

Here, α _f and τ _f represent the composite weight and delay amount in the f-th finger.

The Rake combined output is correlated with M / 2 codes. The correlation output with the spread sequence c _m (mε1,..., M / 2) is given by the following equation.

When ignoring the AWGN term, Equation 7 is divided into a desired component and an inter-pulse interference component (IPI).

  In general, a communication path with inter-pulse interference due to multipath can be considered in the same way as a convolutional coding apparatus using multipath, so that the influence can be removed by the same operation as decoding of a normal convolutional code. Is possible. If this inter-pulse interference component can be removed by equalization processing, the error rate characteristics are greatly improved. In the present invention, interference removal by viterbi equalization using a trellis diagram is performed. A correlation value for each code is obtained for a signal from which interference effects caused by multipath or the like are removed by Viterbi equalization, and converted to a bit string by an MBOK demodulator. In equalization, it is not necessary to limit to Viterbi equalization, and other known equalization may be used. However, in decision feedback equalization, error propagation occurs at a low SNR, and error rate characteristics are improved. I can't get it.

[Reduction of calculation amount in Viterbi equalization for MBOK DS-UWB]
Next, a calculation amount reduction method proposed by the present invention for equalization for MBOK DS-UWB will be described. In order to reduce the calculation amount of the equalization processing, the following two methods are taken.
・ Adoption of DDFSE ・ After reducing the number of trellis states considering MBOK DS-UWB, each will be explained.

[MBOK DS-UWB equalization using DDFSE]
DDFSE is a hybrid type of maximum likelihood sequence estimation (MLSE) and decision feedback estimation (DFE). The amount of calculation can be reduced by reducing the number of delay symbols η considered in MLSE. Hybrid type. When performing interference cancellation by virtue equalization by DDFSE, if the number of delay symbols to be considered is η and the number of delay symbols to be equalized by MLSE is (μ <η), the delay symbol of the difference between η and μ is Equalized by DFE. If η is set to a low value, the number of delay symbols to be equalized is reduced and the amount of calculation is reduced. However, depending on the multipath environment, equalization cannot be performed sufficiently, and interference components remain after equalization. Further, since the calculation amount of DFE is lower than that of MLSE, the calculation amount can be reduced by setting μ to a low value.

  FIG. 3A shows an example of a trellis diagram in MLSE in MBOK DS-UWB. In this figure, the number of delay symbols to be considered η = 2. In the figure, the spread sequence set written in each state (indicated by a circle) is a delay symbol set that assumes that inter-pulse interference has occurred, and is indicated by a branch (a line connecting the circle and the circle). The spreading sequence that corresponds to the current symbol.

Here, the branch metrics in MLSE and DDFSE are calculated as follows.
・ MLSE

  FIG. 3B shows an example of a trellis diagram of DDFSE in MBOK DS-UWB. This figure shows a case where the number of delay symbols to be considered η = 2 and the number of symbols μ used for decision feedback estimation = 1. The interference removal in the determination feedback estimation is performed from the determined symbol (spread sequence) and the channel estimation value.

ここで、ハット（＾）は、判定済みのシンボルを表す。
また、ｆ_nは通信路推定値より計算された重み係数である。 Here, the branch metrics in MLSE and DDFSE are calculated as follows.
・ DDFSE

Here, a hat (^) represents a determined symbol.
F _n is a weighting coefficient calculated from the channel estimation value.

In the communication path where the maximum delay time of the multipath signal is T _dm , MLSE needs to consider the number of delay symbols included in the following equation.

ここで、次のセイル関数は、Ａ以上の最小の整数を示す。

Here, the following sail function indicates the smallest integer equal to or greater than A.

  In FIG. 3, η = 2 is set for simplification as described above. In DDFSE, the number of symbols to be equalized is η = 2, and the number of delay symbols handled as MLSE is μ = 1. The number of states is M to the power of η in MLSE and M is the power of μ in DDFSE. Therefore, when η = 2 and μ = 1, the branch metric calculation amount can be reduced to 1 / M compared to MLSE by applying DDFSE.

[Reduction of number of states considering MBOK DS-UWB]
In MBOK DS-UWB, half of the M codes to be transmitted are codes whose polarity is inverted. Since the distance between codes with different polarities is large, there is little performance degradation even if a method is adopted in which the polarity is first determined for each symbol and only the inter-pulse interference from the spreading sequence having the polarity remaining as a result of the determination is eliminated. it is conceivable that. An example of a trellis diagram in consideration of this point is shown in FIG. As in FIG. 3, the delay symbols considered in DDFSE are η = 2 and μ = 1. FIG. 4 shows a positive polarity symbol. Diffusion series with different positive and negative polarities were parallel branches. FIG. 4 shows an example in which a positive polarity symbol is selected. By expressing spreading sequences with different positive and negative polarities as parallel (parallel) branches and selecting only one polarity to remain when surviving paths are selected, the number of states considered in MLSE can be halved. M / 2) × (M to the power of μ-1). Therefore, in the case of η = 2 and μ = 1 as shown in FIG. 3, the calculation amount can be halved by adding the polarity determination to the survival path selection. Therefore, the metric calculation amount can also be reduced to half compared with DDFSE. Using the DDFSE (Reduced State DDFSE: RS-DDFS) algorithm in which the number of states is reduced, performance evaluation was performed for equalization processing in MBOK DS-UWB. This is shown below.

[Performance evaluation]
In the following, performance evaluation is performed for the proposed computational complexity reduction equalization. First, comparison between sequence estimations taking into account the UWB channel model is performed, and then the simulation results are shown.

[Distance characteristics in UWB channel model]
In order to qualitatively compare the performance of the proposed RS-DDFSE algorithm with MLSE or DDFSE, here we compare the mean square distance for a certain error event. FIG. 5 shows trellis diagrams in which error events in 4BOK DS-UWB are represented by bold lines in (a) MLSE, (b) DDFSE, (c) RS-DDFSE, and each estimation algorithm. The error event is {c2, -c1, c1, c1}. In DDFSE, the number of delay symbols is η = 2 and μ = 1, respectively. As can be seen from each trellis diagram, it can be seen that the path length of the error event is shortened by reducing the amount of calculation.

  By considering the distance to the correct path (all C1), it is possible to evaluate how much the error event affects the performance. Table 1 shows the mean square distance between the error event and the correct path in each UWB channel model.

ここで、Ｒａｋｅフィンガ数は１６本とし、また雑音は加えないものとした。

Here, the number of Rake fingers was 16 and no noise was added.

Table 2 shows the parameters of each channel model.

  The mean square distance shown in Table 2 is a value obtained by obtaining 100 realization values for each channel model and averaging the square distance values between correct paths and error events in each realization value. Expressed as an expression

ここで、

here,

は、正しいパスの拡散系列相関値ベクトル、

Is the spreading sequence correlation vector for the correct path,

はエラーイベントの拡散系列相関値ベクトルを表している。

Represents a spread series correlation value vector of an error event.

If there is no inter-pulse interference, square distance is 6 (= (1 ² tens 1 ²⁾ Ten 2 ²⁾ from becoming, in a LOS (Line Of Sight) environment between transmission and reception is foreseeable CM1, pulse interference It can be seen that the same distance can be obtained as when there is no. On the other hand, as the energy dispersion increases, the energy obtained by Rake synthesis decreases, so the mean square distance decreases particularly in CM4. It can be seen that the proposed RS-DDFSE shows approximately the same distance as MLSE and DDFSE in each channel model. From this, it can be said that there is almost no difference in error rate characteristics between MLSE and DDFSE in this error event.

[Computer simulation results]
Next, performance evaluation by computer simulation is performed for the Viterbi equalization processing using the proposed sequence estimation algorithm RS-FDDFSE. Table 3 shows the parameters in the simulation. For Rake combining, S-rake (Non-patent Document 1) that coherently combines the upper N _{f of the} path gain is used.

  FIG. 6 shows a simulation result of Viterbi equalization using RS-DDFSE in the channel models CM3 and CM4. For comparison, these figures also show the error rate characteristics when equalization is not performed and when MLSE and DDFSE are used. It can be seen that the effect of equalization appears more in CM4 than in CM4. This is because, as shown in Table 2, CM4 is the channel model with the largest delay spread, and thus the influence of interpulse interference is also large. From the simulation results, it can be seen that although RS-DDFSE has a calculation amount halved compared with DDFSE, the error rate characteristic is less degraded to 0.5 dB or less. This result is consistent with the average distance comparison results for error events in the previous section.

  The present invention assumes a receiver of a multi-element orthogonal keying communication device using an ultra-wideband radio signal having a bandwidth by spread modulation larger than the bandwidth by a data modulated signal. As is clear from the above, the bandwidth expansion by multiplying the spread sequence is not essential, and can be applied to the receiver of any multi-way orthogonal keying communication device using spread modulation. Can do.

It is a figure which shows the information mapping method of MBOK DS-UWB. It is a block diagram which shows the example of a receiver structure of this invention. The example of the trellis diagram of MLSE in MBOK DS-UWB is shown. An example of a trellis diagram is shown. (A) MLSE, (b) DDFSE, (c) RS-DDFSE, and trellis diagrams in which error events in 4BOK DS-UWB are represented by bold lines. The simulation result of Viterbi equalization using RS-DDFSE is shown.

Explanation of symbols

1 Pulse Correlator 2 Rake Synthesizer 3 Code Correlator 4 Viterbi Equalizer 5 MBOK Decoder 6 Channel Estimator

Claims (4)

A receiver of a communication device of a multi-element orthogonal keying method using an ultra-wideband radio signal having a bandwidth by spread modulation larger than a bandwidth by data modulation signal
A pulse correlator for correlating the received signal with the transmitted pulse waveform;
A rake synthesizer that collects multipath distributed energy;
A spreading sequence correlator that correlates the output of the rake combiner with the spreading sequence used for communication;
A Viterbi equalizer for suppressing inter-pulse interference included in the output of the spread sequence correlator;
A multi-element orthogonal orthogonal keying decoder for decoding and outputting the output of the Viterbi equalizer;
A channel estimator that estimates the multipath environment in the channel;
A receiver for an ultra-wideband wireless system, comprising:   In the Viterbi equalizer, an operation equivalent to an operation for equalizing a channel having multipath characteristics as an encoder having a convolutional code constraint length equal to a predetermined number of delay symbols is performed. The receiver for an ultra-wideband wireless system according to claim 1, wherein the receiver is equalized.   The branch metric in the Viterbi equalization described above is characterized in that a branch metric is obtained for a value obtained by subtracting in advance a term having an obvious effect due to multipath characteristics from an output of a spreading sequence correlator. Receiver for ultra wideband wireless system.   In the multi-element orthogonal keying decoder, the output of the Viterbi equalizer is subjected to polarity determination and assigned to each polarity group, and each trellis diagram is associated with a predetermined correspondence relationship. 4. The receiver for an ultra wideband radio system according to claim 1, wherein the decoding result of the group signal substitutes the decoding of the signal of the other group in accordance with the correspondence relationship.

Images