JP4365125B2 - Adaptive equalization method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an adaptive equalization method used for land communication and the like, and more particularly to an adaptive equalization method for compensating for multipath fading that occurs when performing high-speed data transmission.
[0002]
[Prior art]
Conventionally, it is generally known that a decision feedback equalizer (DFE: Decision Feedback Equalizer) or the like is effective as an adaptive equalization method for countermeasures against multipath fading. DFE can be realized on a relatively small scale, and can be equalized even when the delay time is large.
[0003]
DFE will be described with reference to FIG. FIG. 3 is a configuration block diagram of a conventional DFE. As shown in FIG. 3, the conventional DFE includes an FF filter (Feed Forward Filter) 150, an FB filter (Feed Back Filter) 160, a complex adder 112, a data determiner 113, a switch 115, a training signal, and the like. It comprises a generator 114 and a tap coefficient update unit 121.
[0004]
When an input signal (received baseband signal) is input to the FF filter 150, the FF filter 150 executes a matching process for obtaining a diversity effect incorporating a direct wave and a delayed wave. Delay registers 101 to 105 store transmission data signals at ½ symbol intervals, and complex multipliers 107 to 111 are used for equalization corresponding to outputs of delay registers 101 to 105, respectively. Complex multiplication is performed with the complex tap coefficient ww (n).
[0005]
The FB filter 160 receives the determination output of the previous symbol or the reference training signal from the training signal generator 120, which is switched by the SW 115 depending on whether the input signal to be equalized is a training signal or a data signal.
[0006]
The FB filter 160 feeds back a determination output or a reference training signal, and performs arithmetic processing in the complex adder 112 to remove residual delayed waves. The delay registers 116 and 117 delay the number of symbols corresponding to the maximum delay time of the multipath delay wave generated in the propagation path, and the complex multipliers 118 and 119 respectively correspond to the outputs of the delay registers 116 and 117 to which they correspond, respectively. Complex multiplication is performed with the complex tap coefficient ww (n).
[0007]
The output signal of the FF filter 150 and the output signal of the FB filter 160 are input to the complex adder 112, and both signals are added by the complex adder 112, and the resulting output signal is input to the data determination unit 113. The The data determination unit 113 selects one mapping value corresponding to the signal point (mapping value) from which the signal from the complex adder 112 is drawn, and this output signal is the result of the equalization processing of the symbol. Is output as demodulated data and used as a decision feedback value to the FB filter 160.
[0008]
The complex adder 120 calculates a difference between the output of the SW 115 and the equalized output that is the output of the complex adder 112, and outputs the difference as an equalization error. The tap coefficient updating unit 121 updates the equalization complex tap coefficient ww (n) by an MMSE (Minimum Mean Square Error) algorithm such as RLS or LMS, and minimizes the equalization error.
[0009]
In this way, by performing complex multiplication with the complex tap coefficient for equalization ww (n) that is updated in a timely manner, multi-wave component diversity combining and canceling operations are performed to realize an equalization function.
[0010]
[Problems to be solved by the invention]
In the conventional equalization method as described above, the training signal at the head of the transmission data is performed, but since the training of the training signal is not sufficient, the convergence time of the equalization error becomes long and the convergence is not completed completely. There is a problem in that equalization of data signals is performed and an equalized result that is erroneously determined is output.
[0011]
An object of the present invention is to provide an adaptive equalization method capable of improving transmission quality while realizing adaptive equalization.
[0012]
[Means for Solving the Problems]
The training equalization process in the conventional adaptive equalization method is performed only once from the beginning of the transmission data.In the present invention, the equalization process for training is performed a plurality of times, so that The transmission quality is improved by reducing the equalization error.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Next, the present invention will be described with reference to the drawings.
[0014]
FIG. 1 shows a frame structure of transmission data used for a multi-training adaptive equalization method according to the present invention. In the frame configuration according to the present invention, data is arranged after the training signal.
[0015]
FIG. 2 shows a configuration of a multi-training adaptive equalization method according to an embodiment of the present invention.
[0016]
When an input signal (reception baseband signal) is input, the frame detection unit 10 captures reception timing by detecting a training signal (synchronization word) in the frame. Based on the captured reception timing, the frame detection unit 10 outputs a training signal to the training signal memory 11 and a data signal to the data memory 13, and stores them.
[0017]
The training signal stored in the training signal memory 11 is input to the FF filter 14 during training. The FF filter 14 performs a filter operation using the tap coefficient wwFF. As the tap coefficient wwFF, a value stored in the tap coefficient update unit 18 is used.
[0018]
The reference training signal output from the training signal generator 19 is input to the FB filter 15. The FF filter 15 performs a filter operation using the tap coefficient wwFB. As the tap coefficient wwFB, a value stored in the tap coefficient update unit 18 is used.
[0019]
The complex adder 16 adds the filter outputs of the FF filter 14 and the FB filter 15 to obtain an equalized output. The complex adder 17 adds the equalized output of the complex adder 16 and the reference training signal, and calculates the equalization error power sum.
[0020]
The tap coefficient updating unit 18 calculates the optimum tap coefficients (tap coefficient wwFF and tap coefficient wwFB) so that the calculated value is minimized in the equalization error power sum calculated by the complex adder 17. When the optimum tap coefficient is calculated, the tap coefficient updating unit 18 updates and stores a new tap coefficient instead of the old tap coefficient. With the new tap coefficient, the FF filter 14 and the FB filter 15 perform respective filter operations.
[0021]
The equalization error threshold value determination unit 22 receives the equalization error power sum calculated by the complex adder 17. The equalization error threshold determination unit 22 calculates a training equalization error by cumulatively adding the training symbol to the equalization error power sum.
[0022]
The equalization error threshold determination unit 22 has a predetermined value as a threshold for sufficiently converging the training equalization error, and the magnitude of the calculated training equalization error power sum and the threshold is large or small. Comparison is performed, and training equalization / data equalization switching control is performed based on the comparison result.
[0023]
The training / data equalization switching control unit 23 performs switching control of the FF filter input signal selector 12 and the FB filter input signal selector 20 based on the comparison result. The training equalization process is repeated a plurality of times until the calculated training equalization error is equal to or less than a predetermined threshold value, and the tap coefficient optimum for data equalization is calculated.
[0024]
The FF filter input signal selector 12 selects and outputs the data stored in the data memory 13 when it is determined that the training equalization error is not more than a predetermined threshold value. When the determination is made, the FB filter input signal selector 20 selects the equalization result determination value output from the data determination unit 21 and outputs it as demodulated data.
[0025]
【The invention's effect】
As described above, the adaptive equalization method of the present invention is a method of repeatedly performing equalization for training, so that the equalization error can be reduced more reliably and the equalization result can be determined more reliably. Can be improved. In addition, by repeatedly performing training, it is possible to reduce the training signal length in one frame and increase the information signal length, and thus it is possible to increase the amount of information signal transmission per frame. Furthermore, by reducing the length of the training signal, it is possible to sufficiently follow the change in the transmission environment in the wireless communication under the environment where the change is fast, so that the transmission quality can be improved.
[Brief description of the drawings]
FIG. 1 shows a frame configuration used for a multi-training adaptive equalization method according to the present invention.
FIG. 2 shows a configuration of a multi-training adaptive equalization method according to an embodiment of the present invention.
FIG. 3 shows a configuration of a DFE method (Decision Feedback Equalizer) according to a conventional embodiment.
[Explanation of symbols]
10: Frame detection unit, 11: Training signal memory, 12: FF filter input signal selector, 13: Data memory, 14: FF filter, 15: FB filter, 16, 17: Complex adder, 18: Tap coefficient update unit , 19: training signal generator, 20: FB filter input signal selector, 21: data judgment unit, 22: equalization error threshold judgment unit, 23: training / data equalization switching control unit
101, 102, 103, 104: 1/2 symbol delay element, 107, 108, 109, 110, 111: complex multiplier, 112: complex adder, 113: data decision unit, 114: training signal generator, 115: FB filter input signal selector, 116, 117: 1 symbol delay element, 118,119: Complex multiplier, 120: Complex adder, 121: Tap coefficient update unit, 150: FF filter, 160: FB filter

Claims (1)

A frame detection unit that detects a received frame and separates a training signal unit and a data unit included in the frame;
A training signal memory unit for storing the training signal unit;
A data memory unit for storing the data unit;
Equalization processing means for performing equalization processing by inputting a signal from the training signal memory unit or the data memory unit;
The training signal memory unit is connected to one of the input terminals, the data memory unit is connected to the other input terminal, and the equalization processing means is connected to the output terminal,
The equalization processing means comprises:
When receiving the frame, the training signal unit stored in the training signal memory unit is input via the switch to perform equalizer tap coefficient training, and at the end of training, the equalization error is compared with a threshold value, and the equalization is performed. When the error is larger than the threshold value, the training signal unit stored in the training signal memory unit is input again and the equalizer tap coefficient training is repeated. On the other hand, when the equalization error is less than the threshold value, the switch The adaptive equalizer is characterized in that the data part stored in the data memory part is switched and the data part is subjected to equalization processing.

Images