KR100595500B1 - On-line adaptive equalizer - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an online adaptive equalizer. In an online adaptive equalizer that converges residual errors of a filter through a coefficient updater including a plurality of multipliers and an adder, the multiplier and the adder even after the residual errors converge to a predetermined level or less. As a normal operation, there was a problem that unnecessary power consumption is continued. Accordingly, the present invention has been made to solve the above-mentioned problems. If the residual error converges sufficiently below the predetermined value in the online adaptive equalizer, the error input of the coefficient updater is forcibly applied to 0 and the coefficient is increased. By significantly reducing the internal state transition probability of the multipliers and the adders in the updater, there is an effect of preventing power consumption and performance degradation of the coefficient updater.

Description

ON-LINE ADAPTIVE EQUALIZER}

1 is a block diagram showing the configuration of a conventional online adaptive equalizer.

Figure 2 shows the output of the error detector over time in Figure 1;

Figure 3 is a block diagram showing the configuration of the present invention online adaptive equalizer.

4 is a diagram illustrating power consumption of a coefficient updater over time in FIG. 3;

*** Description of the symbols for the main parts of the drawings ***

100: filter 110: coefficient updater

120: error detector 130: convergence determiner

140: multiplexer

The present invention relates to an online adaptive equalizer. In particular, in an adaptive equalizer that removes multipath interference of a signal in an environment in which channel characteristics are changed, the residual error input of the coefficient updater is set to 0 when the residual error component is less than a predetermined value. By reducing the internal state transition probability of the multiplier in the coefficient updater to minimize the power consumption.

1 is a block diagram showing a configuration of a conventional online adaptive equalizer, as shown in FIG. 10); An error detector (30) for calculating an error of the online adaptive equalizer through the difference between the input data and the output signal (y (n)) of the filter (10); A coefficient updater for updating the coefficients c ₀ (n) to c _N-1 (n) of the filter 10 by using the error e (n-1) calculated by the error detector 30 ( 20), wherein the filter 10 includes a plurality of delay storages R10 to R1N for sequentially delaying and storing input data; A plurality of coefficient reservoirs R ₂₀ to R _2N-1 which receive and store respective updated coefficients c ₀ (n + 1) to c _N-1 (n + 1); Respectively, the output signals c ₀ (n) to c _N-1 (n) of the coefficient reservoirs R ₂₀ to R _2N-1 and the output signals of the delay reservoirs R ₁₀ to R _{1N are} respectively received. A plurality of multipliers M _{10 to} M _{1N-1 to} multiply; It consists of an adder (S ₁₀ ) for receiving the output signal of the multipliers (M ₁₀ ~ M _1N-1 ) and add it, the coefficient updater 20 is an error calculated by the error detector (30) (e (n-1)) and a plurality of multiplicative operations for receiving the output signals d (n) to d (n + 1-N) of the plurality of delay storages R ₁₀ to R _2N-1 . A multiplier of M _{20 to} M _2N-1 ; Each of the output signals _{(c 0 (n) ~c N} -1 (n)) of the multiplier (M ₂₀ ~M _2N-1) output signal and said stored plurality of coefficient groups (R ₂₀ ~R _2N-1) of It is composed of a plurality of adders (S _{20 ~} S _2N-1 ) to receive the input and add it, it will be described in detail with reference to FIG.

The filter 10 using the LSM algorithm filters and outputs input data through the coefficients c ₀ (n + 1) to c _N-1 (n + 1) updated by the coefficient updater 20. The error detector 30 receiving the output signal y (n) of the filter 10 and the input data calculates an error e (n-1) of the online adaptive equalizer based on the difference between the two output signals. Done.

At this time, the filter 10 sequentially stores delayed input data in a plurality of delay storages R ₁₀ to R _1N , and updates the coefficient c ₀ (n + 1) input from the coefficient update unit 20. C _-1 to n _-1 (n + 1) are stored in the plurality of coefficient reservoirs R ₂₀ to R _2N , and then the respective delay reservoirs R ₁₀ to R _1N-1 are represented as in Equation 1 below. After multiplying through a plurality of multipliers (M _{10 to} M _1N-1 ), an add operation is performed through an adder (S ₁₀ ) to output the multiplication operation.

Here, i is a tap number, and a delay value of delay data used for updating coefficients in each tap is a case in which an associated delay time during an error operation is assumed to be one cycle.

The error detection method of the error detector 30 varies slightly from system to system, but the error of the adaptive equalizer is calculated using the output y (n) of the filter 10 and the input data. The coefficient updater 20 receiving the error e (n-1) calculated by the detector 30 uses coefficients c ₀ (n) to c _N− of the filter 10 by using Equation 2 below. ₁ (n)) is updated.

Here, μ is a constant related to the convergence speed. As the value of μ increases, the convergence speed increases, and when the value of μ decreases, the convergence speed decreases.

In this case, multipath interference and white noise are introduced together through the input data, and the multipath interference as a linear component is completely removed from the online adaptive equalizer, but the white noise as a nonlinear component is not sufficiently removed as shown in FIG. 2. Even after convergence, the error component remains due to the white noise component.

 As described above, in the online adaptive equalizer that converges the residual error of the filter through a coefficient updater including a plurality of multipliers and adders, the power consumption is unnecessary as the multiplier and the adder operate normally even after the residual error converges to a predetermined level or less. There was a problem that takes place.

Accordingly, the present invention has been made to solve the above-mentioned conventional problems. When the residual error component is less than or equal to a predetermined value, the residual state input of the coefficient updater is set to 0 to shift the internal state of the multiplier in the coefficient updater. The objective is to provide an online adaptive equalizer that reduces the probability to minimize power consumption.

A configuration of the present invention for achieving the above object is a filter for filtering the input data using a minimum mean square algorithm; An error detector for calculating an error based on a difference between the input data and the output signal of the filter; A convergence determiner configured to output a convergence determination signal by comparing the error calculated by the error detector with a preset value; A multiplexer which selects and outputs an error of the error detector according to a convergence determination signal of the convergence determiner; And a coefficient updater for updating the coefficients of the filter using the output signal of the multiplexer.

Hereinafter, with reference to the accompanying drawings, the operation and effect of an embodiment of the present invention will be described in detail.

3 is a block diagram showing the configuration of the online adaptive equalizer of the present invention, and as shown therein, a filter 100 for filtering input data using an LSM algorithm; An error detector 120 that calculates an error e (n-1) of the online adaptive equalizer based on the difference between the input data and the output signal y (n) of the filter 100; A convergence determiner (130) for outputting a convergence determination signal (ADS) by comparing the error (e (n-1)) calculated by the error detector (120) with a preset value; A multiplexer (140) for selecting and outputting an error (e (n-1)) or 0 of the error detector 120 by a convergence determination signal (ADS) of the convergence determiner (130); And a coefficient updater 110 for updating the coefficients c ₀ (n) to c _N-1 (n) of the filter 100 by using the output signal of the multiplexer 140. The filter 100 And the coefficient updater 110 are configured in the same manner as in FIG. 1 and will be described in detail with reference to FIG. 4 attached to the operation process according to the present invention.

The operation of receiving the first input data and equalizing the same operates in the same manner as in FIG. That is, when the error detector 120 calculates the error e (n-1) using the difference between the output signal y (n) of the filter 100 and the input data, the error of the error detector 120 is calculated. The convergence determiner 130 that receives (e (n-1)) outputs a convergence determination signal ADS to the multiplexer 140 when the error e (n-1) is greater than a preset value. The error e (n-1) calculated by the error detector 120 is output to the coefficient updater 110.

In addition, the coefficient updater 110 that receives the error e (n-1) of the error detector 120 may include delayed input data d (n-1) to d (n + 1-N). After multiplying them, the process of adding the output signals c ₀ (n) to c _N-1 (n) of the previous filter 100 is repeated to converge in the direction of reducing the residual error e (n-1). Let's do it.

The convergence determiner 130 determines that the multipath interference, which is applied as input data, is completely removed when the residual error e (n-1) decreases continuously below a predetermined value for a predetermined time. Accordingly, the multiplexer 140 outputs 0 instead of the error e (n-1) of the error detector 120 from the multiplexer 140 to the coefficient updater 110 through a convergence determination signal ADS.

Accordingly, as the coefficient updater 110 receives 0 instead of the error e (n-1) of the error detector 120, the multipliers M _{20 to} M _2N-1 and the adder S _20. The probability of the internal state transition of ˜S _2N-1 ) decreases drastically.

That is, the plurality of multipliers M _{20 to} M _2N-1 multiply the delayed data inputs d (n-1) to d (n + 1-N) and 0, which are randomly applied, and internal state transitions. Significantly reduced, and the plurality of adders S _{20 to} S _2N-1 each have a previous coefficient value c ₀ (n) to c _N-1 (n) and the multiplier M _{20 to} M _2N-1 . By adding the output of, the internal state transition does not occur at all, so no power is consumed.

Therefore, as shown in FIG. 4, the power consumption of the coefficient updater 110 is significantly reduced when the error e (n-1) of the error detector 120 is lower than or equal to a predetermined value and when zero is applied. Decreases.

Thereafter, when the channel environment is changed and the residual error e (n-1) of the equalizer becomes large, the convergence determiner 130 needs to re-converge to reduce the error. The error e (n-1) of the error detector 120 is output to the coefficient updater 110 through the multiplexer 140 to converge the residual error.

As described in detail above, the present invention forcibly applies an error input of the coefficient updater to 0 when the residual error sufficiently converges below a predetermined value in the online adaptive equalizer, thereby shifting the internal state of the multiplier and the adder in the coefficient updater. By significantly reducing the probability, there is an effect of preventing power consumption and performance degradation of the coefficient updater.

Claims (1)

A filter for filtering input data using a least mean square algorithm; An error detector for calculating an error based on a difference between the input data and the output signal of the filter; A convergence determiner configured to output a convergence determination signal by comparing the error calculated by the error detector with a preset value; A multiplexer which selects and outputs an error of the error detector according to a convergence determination signal of the convergence determiner; And a coefficient updater for updating the coefficients of the filter using the output signal of the multiplexer.

Images