Waveform equalizer, its control method, reception device, and its control method

Technical Field

0001The present invention relates to a waveform equalizer and a method for controlling the same, as well as a receiving apparatus and a method for controlling the same. More particularly, the invention relates to a waveform equalizer and a method for controlling the same, as well as a receiving apparatus and a method for controlling the same whereby better receiving characteristics are provided than before.

Background Art

0002In signal transmission systems, received waves can be distorted due to reflected waves occurring over transmission channels. For example, in the case of a terrestrial wave TV broadcast, the radio waves direct from a transmitting tower are interfered with by the waves coming therefrom but getting detoured through reflection on buildings and mountains before arrival, as shown in <figref idref="f0001">Fig. 1</figref>. Such interference between direct and reflected waves has been known as "ghost" since the era of analog TV broadcasting. It is still a big problem affecting the receiving characteristics of digital TV broadcasts.

0003Illustratively, as shown in <figref idref="f0001">Fig. 2</figref>, comparing the spectrum in effect when there are no reflected waves (shown on the left) with the spectrum in effect when there are reflected waves (shown on the right) reveals the following: that whereas power density is held constant with regard to frequency when there are no reflected waves (on the left), the level of power density drops at a certain frequency where signal distortion is caused by reflected waves (on the right).

0004The waveform equalizer is used as a device to remove such distortion. There are diverse structures of the waveform equalizer. Generally, the waveform equalizer may be structured as a filter having a coefficient that constitutes the reverse characteristic of the transmission channel in use. <figref idref="f0002">Fig. 3</figref> is explanatory of a waveform equalizer structured in such a manner.

0005As shown in <figref idref="f0002">Fig. 3</figref>, if the frequency characteristic of the channel coming from a broadcasting station is assumed to be H(f), then the waveform equalizer inside a receiving device receiving the broadcast signal is arranged to have a frequency characteristic of 1/H(f). This allows the waveform equalizer to output a interference-free signal to a demodulating/decoding section located downstream. That is, even if there exist reflected waves, it is possible to provide a spectrum without a dip as shown on the left of <figref idref="f0001">Fig. 2</figref>.

0006<figref idref="f0003">Fig. 4</figref> shows a structure of a waveform equalizer. This waveform equalizer is made up of registers 11₁ through 11_n each delaying the input signal by one clock pulse for output to the immediately subsequent stage, multipliers 12₀ through 12_n each multiplying the input signal by one of filter coefficients (tap coefficients) C₁₀ through C_1n, and adders 13₁ through 13_n adding up the products from the multipliers 12₀ through 12_n.

0007In the waveform equalizer of <figref idref="f0003">Fig. 4</figref>, the registers 11₁ through 11_n delay the input received signal by one clock pulse each, and the received signal thus delayed is multiplied by each of the filter coefficients (tap coefficients) C₁₀ through C_1n by each of the multipliers 12₀ through 12_n. Then, the adders 13₁ through 13_n add up the products from the multipliers 12₀ through 12_n. The resulting sum of the products is output as an equalized signal.

0008As a result, the equalized signal is an interference-free signal. Incidentally, the filter coefficients by which the delayed received signal is multiplied by each of the multipliers 12₀ through 12_n are acquired in keeping with impulse responses as shown in <figref idref="f0004">Fig. 5</figref>, to be discussed later.

0009Also, the waveform equalizers are roughly classified by the operating frequency into symbol rate equalizers and fractionally spaced equalizers.

0010Parenthetically, the symbol rate equalizer and fractionally spaced equalizer are discussed in detail in a book titled "<nplcit id="ncit0001" npl-type="b"><text>Digital Communication," written by John G. Proakis and translated by Koichi Sakaniwa et al into Japanese, published by Kagaku Gijutsu Shuppan, Inc. in November 1999 (ISBN: 978-4-87653-073-1 (4-87653-073-4</text></nplcit>)).

0011Comparing the two types of waveform equalizers reveals the following: that the symbol rate equalizer drives its filter using the symbol frequency of the transmitted signal, and that the fractionally spaced equalizer effects the driving using a frequency higher than the symbol frequency (usually by use of the frequency acquired by multiplying the symbol frequency by an integer multiple). For these reasons, there exist the following advantages and disadvantages regarding the two types of waveform equalizers.

0012First of all, where there are a sufficient number of taps, the fractionally spaced equalizer can perform equalization more accurately than the symbol rate equalizer. This is because the symbol rate equalizer does not satisfy sampling theorem and is thus theoretically incapable of reproducing the transmitted signal, whereas the fractionally spaced equalizer satisfies sampling theorem and is thus theoretically capable of reproducing the transmitted signal.

0013Second, given the same number of taps, the symbol rate equalizer can deal with longer delayed waves than the fractionally spaced equalizer. This is because the ability of a waveform equalizer to deal with longer delayed waves is determined by the length of the impulse response that can be expressed by the waveform equalizer in question. That is, as shown in <figref idref="f0004">Fig. 5</figref>, a fractionally spaced equalizer operating at an n-fold symbol rate needs n times as many taps as those for the symbol rate equalizer in order to express the impulse response of the same length.

0014<figref idref="f0004">Fig. 5</figref> is a graphic representation showing the relationship between impulse responses and filter coefficients.

0015In the graph of <figref idref="f0004">Fig. 5</figref>, the vertical axis stands for impulse responses. The higher the value along the vertical axis, the larger the impulse response value. The horizontal axis denotes the time of which the direction is from left to right as seen in the graph.

0016In <figref idref="f0004">Fig. 5</figref>, the arrows extending toward the waveform of the impulse response are spaced at intervals of T/2. Therefore, the space T for two arrows corresponds to the period of the symbol rate equalizer (i.e., symbol period), and the space T/2 for one arrow corresponds to the period of the fractionally spaced equalizer. That means the fractionally spaced equalizer requires twice as many taps as those for the symbol rate equalizer in order to express the impulse response of the same length.

0017If it were possible to have a sufficiently large number of taps, then a highly accurate fractionally spaced equalizer might well be utilized. However, always providing a sufficiently large number of taps is not realistic in terms of costs and other considerations. Thus the designers of waveform equalizers need to design optimal circuits by taking the above-mentioned advantages and disadvantages into account.

0018In view of such trade-offs, Patent Document 1 below proposes techniques whereby a symbol rate equalizer and a fractionally spaced equalizer are incorporated in a receiving apparatus in such a manner that one of them is selected for use depending on the channel. According to the proposal, a selection can be made between the symbol rate equalizer and the fractionally spaced equalizer as needed. <ul id="ul0001" list-style="none" compact="compact"><li>Patent Document 1: Japanese Patent Laid-open No. <patcit id="pcit0001" dnum="JPHEI3244220B"><text>Hei 3-244220</text></patcit></li></ul>

Disclosure of Invention

Technical Problem

0019However, although ordinary techniques including those disclosed by the above-cited Patent Document 1 make the most of the advantages and disadvantages in performance of both the symbol rate equalizer and the fractionally spaced equalizer, the demodulating apparatus is required to accommodate two bulky blocks of waveform equalizers inside. This drives up costs.

0020Specifically, instead of incorporating two waveform equalizers and selecting one of them for use, an ideal waveform equalizer design may well involve installing a single waveform equalizer capable of offering the performance advantages of both the symbol rate equalizer and the fractionally spaced equalizer.

0021The present invention has been made in view of the above circumstances and provides a single waveform equalizer controlled selectively for use either as a symbol rate equalizer or a fractionally spaced equalizer.

Technical Solution

0022According to one aspect of the present invention, there is provided a waveform equalizer for performing waveform equalization of an input signal, the waveform equalizer including: a filter including at least a group of delay devices connected serially to delay the input signal successively, a group of multipliers for multiplying an output from each of the delay devices by a filter coefficient, and a group of adders for adding up outputs from the multipliers in order to acquire a waveform-equalized output signal; and timing signal selecting means for selecting either a first or a second timing signal for driving the filter, the first timing signal driving the filter at intervals of a period of a symbol frequency of the input signal, the second timing signal driving the filter at intervals of a period shorter than that of the symbol frequency. The filter is driven either as a symbol rate equalizer in accordance with the first timing signal selected or as a fractionally spaced equalizer in keeping with the second timing signal selected.

0023The timing signal selecting means may first select the second timing signal to drive the filter as the fractionally spaced equalizer, then switch from the second timing signal to the first timing signal in keeping with the value of the filter coefficient after an error signal has settled, and drive the filter as the symbol rate equalizer using the first timing signal.

0024The timing signal selecting means may select either the first timing signal or the second timing signal in such a manner as to minimize channel error rate.

0025The timing signal selecting means may select either the first timing signal or the second timing signal in such a manner as to minimize an error signal inside the waveform equalizer.

0026The filter coefficient may be a predetermined fixed value, and the filter may be driven as a fixed coefficient equalizer based on the filter coefficient.

0027The filter coefficient may be a value determined adaptively based on adaptive equalization algorithm, and the filter may be driven as an adaptive equalizer based on the filter coefficient.

0028The period shorter than that of the symbol frequency may have a frequency that is an integer multiple of the symbol frequency.

0029A controlling method according to the first aspect of the present invention corresponds to the above-described waveform equalizer according to the first aspect of this invention.

0030The waveform equalizer and the method for controlling the same according to the first aspect of the present invention thus involve the use of a filter including at least a group of delay devices connected serially to delay the input signal successively, a group of multipliers for multiplying an output from each of the delay devices by a filter coefficient, and a group of adders for adding up outputs from the multipliers in order to acquire a waveform-equalized output signal, the filter being driven by either a first or a second timing signal, the first timing signal being selected to drive the filter at intervals of a period of a symbol frequency of the input signal, the second timing signal being selected to drive the filter at intervals of a period shorter than that of the symbol frequency. The filter is driven either as a symbol rate equalizer in accordance with the first timing signal selected or as a fractionally spaced equalizer in keeping with the second timing signal selected.

0031According to a second aspect of the present invention, there is provided a receiving apparatus for receiving a modulated signal derived from digital modulation of a carrier wave, the receiving apparatus including: a filter including at least a group of delay devices connected serially to delay the input signal successively, a group of multipliers for multiplying an output from each of the delay devices by a filter coefficient, and a group of adders for adding up outputs from the multipliers in order to acquire a waveform-equalized output signal; and timing signal selecting means for selecting either a first or a second timing signal for driving the filter, the first timing signal driving the filter at intervals of a period of a symbol frequency of the input signal, the second timing signal driving the filter at intervals of a period shorter than that of the symbol frequency. The filter is driven either as a symbol rate equalizer in accordance with the first timing signal selected or as a fractionally spaced equalizer in keeping with the second timing signal selected.

0032A controlling method according to the second aspect of the present invention corresponds to the above-described receiving apparatus according to the second aspect of this invention.

0033The receiving apparatus and the method for controlling the same according to the second aspect of the present invention thus involve the use of a filter including at least a group of delay devices connected serially to delay the input signal successively, a group of multipliers for multiplying an output from each of the delay devices by a filter coefficient, and a group of adders for adding up outputs from the multipliers in order to acquire a waveform-equalized output signal, the filter being driven by either a first or a second timing signal, the first timing signal being selected to drive the filter at intervals of a period of a symbol frequency of the input signal, the second timing signal being selected to drive the filter at intervals of a period shorter than that of the symbol frequency. The filter is driven either as a symbol rate equalizer in accordance with the first timing signal selected or as a fractionally spaced equalizer in keeping with the second timing signal selected.

Advantageous Effects

0034According to the first aspect of the present invention, as described above, a single waveform equalizer is controlled selectively for use as a symbol rate equalizer or as a fractionally spaced equalizer offering higher receiving characteristics than before.

0035Also, according to the second aspect of the present invention, a single waveform equalizer is controlled selectively for use as a symbol rate equalizer or as a fractionally spaced equalizer offering higher receiving characteristics than before.

Brief Description of Drawings

0036<ul id="ul0002" list-style="none" compact="compact"><li>[<figref idref="f0001">Fig. 1</figref>] <figref idref="f0001">Fig. 1</figref> is a view explanatory of radio interference.</li><li>[<figref idref="f0001">Fig. 2</figref>] <figref idref="f0001">Fig. 2</figref> is a view explanatory of changes in spectrum due to the presence or absence of reflected waves.</li><li>[<figref idref="f0002">Fig. 3</figref>] <figref idref="f0002">Fig. 3</figref> is a view explanatory of a waveform equalizer.</li><li>[<figref idref="f0003">Fig. 4</figref>] <figref idref="f0003">Fig. 4</figref> is a circuit diagram showing a structure of an ordinary waveform equalizer.</li><li>[<figref idref="f0004">Fig. 5</figref>] <figref idref="f0004">Fig. 5</figref> is a graph showing the relationship between impulse responses and filter coefficients.</li><li>[<figref idref="f0005">Fig. 6</figref>] <figref idref="f0005">Fig. 6</figref> is a block diagram showing a structure of a receiving apparatus embodying the present invention.</li><li>[<figref idref="f0006">Fig. 7</figref>] <figref idref="f0006">Fig. 7</figref> is a timing chart explanatory of what is output by a sampling synchronizing section.</li><li>[<figref idref="f0006">Fig. 8</figref>] <figref idref="f0006">Fig. 8</figref> is a view showing a typical sampling.</li><li>[<figref idref="f0007">Fig. 9</figref>] <figref idref="f0007">Fig. 9</figref> is a view explanatory of a received signal.</li><li>[<figref idref="f0008">Fig. 10</figref>] <figref idref="f0008">Fig. 10</figref> is a circuit diagram explanatory of a detailed structure of a fixed coefficient equalizer.</li><li>[<figref idref="f0009">Fig. 11</figref>] <figref idref="f0009">Fig. 11</figref> is a timing chart explanatory of a fixed coefficient equalizer operating in symbol rate mode.</li><li>[<figref idref="f0010">Fig. 12</figref>] <figref idref="f0010">Fig. 12</figref> is a timing chart explanatory of a fixed coefficient equalizer operating in fractionally spaced mode.</li><li>[<figref idref="f0011">Fig. 13</figref>] <figref idref="f0011">Fig. 13</figref> is a view explanatory of a method for estimating points of transmission as the source of an output signal of a waveform equalizer.</li><li>[<figref idref="f0012">Fig. 14</figref>] <figref idref="f0012">Fig. 14</figref> is a circuit diagram explanatory of a detailed structure of an adaptive equalizer.</li><li>[<figref idref="f0013">Fig. 15</figref>] <figref idref="f0013">Fig. 15</figref> is a timing chart explanatory of an adaptive equalizer operating in symbol rate mode.</li><li>[<figref idref="f0014">Fig. 16</figref>] <figref idref="f0014">Fig. 16</figref> is a timing chart explanatory of an adaptive equalizer operating in fractionally spaced mode.</li><li>[<figref idref="f0015">Fig. 17</figref>] <figref idref="f0015">Fig. 17</figref> is a block diagram showing a typical structure of a personal computer.</li></ul>

Explanation of Reference Numerals

003731 Receiving apparatus, 41 RFIC, 42 Demodulating LSI, 43 Host CPU, 51 A/D converting section, 52 Sampling synchronizing section, 53 Waveform equalizer, 54 Error correcting section, 71 Fixed coefficient equalizer, 81 Selector, 82 Filter, 83 Register, 84 Register, 85 Register, 91 Register, 92 Multiplier, 93 Adder, 101 Adaptive equalizer, 111 Selector, 112 Filter, 113 Register, 114 Register, 115 Register, 116 Arithmetic unit, 117 Delay circuit, 118 Register, 119 Multiplier, 120 Multiplier, 121 Adder, 122 Register, 123 Error signal computing section, 131 Register, 132 Multiplier, 133 Adder

Best Mode for Carrying Out the Invention

0038Some embodiments of the present invention will be described below in reference to the accompanying drawings.

0039<figref idref="f0005">Fig. 6</figref> is a block diagram showing a structure of a receiving apparatus embodying the present invention.

0040The receiving apparatus 31 is structured to include an RFIC (Radio Frequency Integrated Circuit) 41, a demodulating LSI (Large Scale Integration) 42, and a host CPU (Central Processing Unit) 43. In a system made up of the receiving apparatus 31 and an antenna 32 (called the digital TV receiving system hereunder), the broadcast wave of a BS (Broadcasting Satellite) digital broadcast from a broadcasting station is received illustratively by the antenna 32.

0041In the digital TV receiving system, the signal received by the antenna 32 is converted by the RFIC 41 into a baseband signal that is input to the demodulating LSI 42. In the ensuing description, the baseband signal input to the demodulating LSI 42 will be simply referred to as DT.

0042The demodulating LSI 42 performs predetermined processes on DT input from the RFIC 41 so as to acquire a transport stream (TS) that is then output.

0043The demodulating LSI 42 is structured to include an A/D (Analog/Digital) converting section 51, a sampling synchronizing section 52, a waveform equalizer 53, an error correcting section 54, and an I2CIF (Inter IC Interface) 55.

0044The A/D converting section 51 converts DT, which is an analog signal input from the RFIC 41, into a digital signal that is supplied to the sampling synchronizing section 52. In the ensuing description, the digital signal output from the A/D converting section 51 is called DT0.

0045The sampling synchronizing section 52 performs a symbol synchronizing process, which is a process for establishing the synchronization of a symbol point, on DT0 having undergone the conversion to digital form by the A/D converting section 51. More specifically, the symbol synchronizing process generates flags SEN1 and DEN1, the flag SEN1 indicating that DT1 constituting the received data before waveform equalization is a symbol point, the flag DEN1 indicating that DT1 is either a symbol point or an intermediate point between two symbol points.

0046The sampling synchronizing section 52 supplies the waveform equalizer 53 with the generated SEN1 and DEN1, as well as DT1 which constitutes the received data before waveform equalization and which corresponds to DT0.

0047Explained below in reference to the timing chart of <figref idref="f0006">Fig. 7</figref> is what is output by the sampling synchronizing section 52.

0048<figref idref="f0006">Fig. 7</figref> is a timing chart that shows, from the top down, DT1, SEN1, and DEN1 as the data output from the sampling synchronizing section 52. The direction of time is from left to right as seen in the chart. This direction of time is the same in the other timing charts to be discussed later.

0049As shown in <figref idref="f0006">Fig. 7</figref>, the sampling synchronizing section 52 successively outputs D0, D1, D2, D3, D4, ... as DT1 corresponding to DT0 input from the A/D converting section 51. The data D0, D1, D2, D3, D4, ... have been sampled from the analog data desired to be expressed as shown in the graph of <figref idref="f0006">Fig. 8</figref>.

0050Returning to <figref idref="f0006">Fig. 7</figref>, the flag SEN1 goes High when DT1 is at a symbol point (i.e., D0, D2, D4, ...); otherwise SEN1 stays Low. Thus the flag SEN1 goes High at intervals of a symbol period.

0051The flag DEN1 goes High when DT1 is at a symbol point or at an intermediate point between two symbol points (i.e., D0, D1, D2, D3, D4, ...) as shown in <figref idref="f0006">Fig. 7</figref>; otherwise DEN1 stays Low. Thus the flag DEN1 goes High at intervals of half the symbol period. Going High at intervals of half the symbol period, the flag DEN1 is always High when the flag SEN1 is High.

0052Incidentally, as shown in <figref idref="f0007">Fig. 9</figref>, the received signal DT may be expressed as an imbrication of signals each given by displacing the unit pulse by n × T (n stands for an integer and T for the symbol period). That is, the symbol points are the points sampled at t = n × T as indicated by solid points in <figref idref="f0007">Fig. 9</figref>, and the intermediate points are the points sampled at t = n × T + T/2 as shown by crosses in <figref idref="f0007">Fig. 9</figref>.

0053Returning to <figref idref="f0005">Fig. 6</figref>, the waveform equalizer 53 is supplied with not only DT1, SEN1 and DEN1 coming from the sampling synchronizing section 52 but also SEL from the host CPU 43 connected to the I2CIF 55 by way of an I2C bus 44. Based on these data, the waveform equalizer 53 performs a waveform equalizing process on DT1 which is the received data before waveform equalization, to obtain DT2 which is the received data having undergone waveform equalization. The data DT2 is then fed to the error correcting section 54.

0054Here, SEL is a signal (timing signal) that causes the waveform equalizer 53 to select between the symbol rate equalizer function and the fractionally spaced equalizer function. Thus in response to the signal SEL from the host CPU 43, the waveform equalizer 53 operates either as a symbol rate equalizer or as a fractionally spaced equalizer. The waveform equalizer 53 is timed in operation by SEN1 from the sampling synchronizing section 52 when acting a symbol rate equalizer, or by DEN1 when acting as a fractionally spaced equalizer, before feeding the resulting DT2 to the error correcting section 54.

0055Other equalizer types that can be used as the waveform equalizer 53 include one whose filter coefficient is fixed (called the fixed coefficient equalizer hereunder) or one which has its own filter coefficient controlled adaptively while observing the received signal (called the adaptive equalizer hereunder). These equalizers will be discussed later in detail.

0056Meanwhile, the timing for selecting between the symbol rate equalizer and the fractionally spaced equalizer is determined illustratively by one the following three methods.

0057The first method involves using channel conditions. More specifically, the host CPU 43 may estimate channel error rate and, in order to minimize the estimated channel error rate, determine whether the symbol rate equalizer or the fractionally spaced equalizer is to be operated. Upon determining that the symbol rate equalizer is to be operated, the host CPU 43 outputs a High-level SEL signal to the waveform equalizer 53. On the other hand, upon determining that the fractionally spaced equalizer is to be operated, the host CPU 43 outputs a Low-level SEL signal to the waveform equalizer 53.

0058The second method involves using internal status of the waveform equalizer 53. More specifically, the host CPU 43 may detect an error signal inside the waveform equalizer 53 and, in order to minimize the detected error signal, determine whether the symbol rate equalizer or the fractionally spaced equalizer is to be operated. Depending on the result of the determination, the host CPU 43 outputs the High-level or Low-level signal to the waveform equalizer 53.

0059Further, the third method involves selecting the functionality of the waveform equalizer 53 acting either as the symbol rate equalizer or as the fractionally spaced equalizer in a particularly timed manner. More specifically, the host CPU 43 may first output the Low-level signal to the waveform equalizer 53 to operate it as the fractionally spaced equalizer. Thereafter, depending on the value of the filter coefficient in effect after the error signal (e.g., error signal inside the waveform equalizer 53) has settled, the host CPU 43 determines whether or not to operate the waveform equalizer 53 as the symbol rate equalizer. Upon determining that the waveform equalizer 53 is to be operated as the symbol rate equalizer, the host CPU 43 outputs the High-level signal to the waveform equalizer 53. The signal causes the waveform equalizer 53 to switch from the fractionally spaced equalizer to the symbol rate equalizer in functionality.

0060The above-described three methods are only examples whereby a selection is made between the symbol rate equalizer and the fractionally spaced equalizer. Obviously, the selection can be made between the symbol rate equalizer and the fractionally spaced equalizer in an otherwise timed manner to take into account their advantages and disadvantages.

0061In this embodiment, the SEL signal is described as being set by the host CPU 43, but obviously the SEL may be set by other methods as well. The point is that the waveform equalizer 53 should be given an instruction to select between the symbol rate equalizer and the fractionally spaced equalizer, by any means or method.

0062The error correcting section 54 is supplied with not only DT2 from the waveform equalizer 53 but also SEN2 which is a flag indicating that DT2 is at a symbol point. The reason the error correcting section 54 is not supplied with DEN2 which is a flag indicating DT2 is either at a symbol point or at an intermediate point between symbol points is that the data constituting the intermediate points between symbol points is not needed following the waveform equalizing process. Thus DEN2 is left open and only SEN2 is input to the downstream block.

0063Based on DT2 and SEN2 coming from the waveform equalizer 53, the error correcting section 54 performs a channel error removing process to obtain TS that is output outside the demodulating LSI 42.

0064In the demodulating LSI 42, as described above, the waveform equalizer 53 operates as the symbol rate equalizer given the timing of SEN1 from the sampling synchronizing section 52 or as the fractionally spaced equalizer given the timing of DEN1 also from the sampling synchronizing section 52.

0065Meanwhile, as mentioned above, the waveform equalizer 53 may also be structured as a fixed coefficient equalizer or as an adaptive equalizer. What follows is a description of the waveform equalizer 53 structured as these equalizers. Described first in reference to <figref idref="f0008 f0009 f0010">Figs. 10 through 12</figref> is how the waveform equalizer 53 is operated illustratively as a fixed coefficient equalizer.

0066<figref idref="f0008">Fig. 10</figref> is a block diagram explanatory of a detailed structure of a fixed coefficient equalizer 71 as one variation of the waveform equalizer 53 in <figref idref="f0005">Fig. 6</figref>. The fixed coefficient equalizer 71 shown in <figref idref="f0008">Fig. 10</figref> is a six-tap waveform equalizer with its filter coefficient fixed. Although the six-tap fixed coefficient equalizer 71 is explained here in connection with this embodiment as a typical fixed coefficient equalizer, this is only an example; a fixed coefficient equalizer having other than six taps may also be used instead.

0067The fixed coefficient equalizer 71 in <figref idref="f0008">Fig. 10</figref>, which corresponds to the waveform equalizer 53 in <figref idref="f0005">Fig. 6</figref>, inputs DT1, SEN1 and DEN1 from the sampling synchronizing section 52 and SEL from the host CPU 43. The fixed coefficient equalizer 71 performs the waveform equalizing process based on these signals and outputs DT2, SEN2 and DEN2 derived from the process.

0068The fixed coefficient equalizer 71 is structured to include a selector 81, a filter 82, a register 83, a register 84, and a register 85. Incorporated in the six-tap waveform equalizer, the filter 82 is furnished with registers 91₁ through 91₅ which hold DT1 for a predetermined period each before output, multipliers 92₀ through 92₅ which multiply DT1 delayed by the registers 91₁ through 91₅ each by filter coefficients C₂₀ through C₂₅ respectively, and adders 93₁ through 93₅ which add up the products from the multipliers 92₀ through 92₅.

0069The elements above are explained hereunder, in no particular order. SEN1, DEN1 and SEL are input to the selector 81. If the input SEL represents 1, that means the equalizer operates as the symbol rate equalizer. Then the selector 81 selects SEN1 as en and outputs what is selected to the registers 91₁ through 91₅.

0070When en goes High, i.e., at intervals of a symbol period, the register 91₁ outputs the input DT1 to the register 91₂. Likewise, each of the registers 91₂ through 91₅ forwards the output from the preceding register to the immediately subsequent register per symbol period. That is, the input DT1 is shifted per symbol period from the register 91₁ to the register 91₂ to the register 91₃ to the register 91₄ to the register 91₅, in that order.

0071Also, the multiplier 92₀ multiplies DT1 by the filter coefficient C₂₀ and outputs the product to the adder 93₁. The adder 93₁ adds up the output from the multiplier 92₀ and the output from the multiplier 92₁ (i.e., product of the multiplication of DT1 delayed by the register 91₁, by the filter coefficient C₂₁), and outputs the sum to the adder 93₂. Likewise, each of the adders 93₂ through 93₅ proceeds to add up the sum from the preceding adder and the product input from the corresponding multiplier, and outputs the sum to the immediately subsequent adder.

0072As explained above, when the fixed coefficient equalizer 71 of <figref idref="f0008">Fig. 10</figref> operates as the symbol rate equalizer because the input SEL is 1, the product-sum operations carried out on DT1 and on the outputs from the registers 91₁ through 91₅ as well as on the filter coefficients C₂₀ through C₂₅ at the timing of SEN1 (i.e., per symbol period) results in a waveform-equalized output signal DT2 that is output.

0073On the other hand, if the input SEL represents 0, then the equalizer operates as the fractionally spaced equalizer. Then the selector 81 selects DEN1 as en and outputs what is selected to the registers 91₁ through 91₅.

0074When en goes High, i.e., at intervals of a half symbol period, each of the registers 91₁ through 91₅ forwards the output from the preceding register to the immediately subsequent register. That is, the input DT1 is shifted per half symbol period from the register 91₁ to the register 91₂ to the register 91₃ to the register 91₄ to the register 91₅, in that order.

0075As in the above-described case of the equalizer acting as the symbol rate equalizer, each of the adders 93₁ through 93₅ proceeds to add up the sum from the preceding adder and the product input from the corresponding multiplier, and outputs the sum to the immediately subsequent adder.

0076As explained above, when the fixed coefficient equalizer 71 of <figref idref="f0008">Fig. 10</figref> operates as the fractionally spaced equalizer because the input SEL is 0, the product-sum operations carried out on DT1 and on the outputs from the registers 91₁ through 91₅ as well as on the filter coefficients C₂₀ through C₂₅ at the timing of DEN1 (i.e., per half symbol period) result in a waveform-equalized output signal DT2 that is output.

0077The result of the product-sum operations is input from the adder 93₅ to the register 83. The register 83 outputs as DT2 the result of the product-sum operations held therein, at the timing of DEN1 (i.e., per half symbol period).

0078The register 84 holds SEN1 and delays it by one clock pulse before outputting it as SEN2. The register 85 holds DEN1 and delays it by one clock pulse before outputting it as DEN2. That is, the fixed coefficient equalizer 71 delays SEN1 and DEN1 by one clock pulse each before outputting them as SEN2 and DEN2, respectively.

0079The fixed coefficient equalizer 71 is selectively controlled in response to the SEL signal as described above, so as to operate either as the symbol rate equalizer or as the fractionally spaced equalizer. In the ensuing description, the mode in which the fixed coefficient equalizer 71 operates as the symbol rate equalizer will be called symbol rate mode, and the mode in which the fixed coefficient equalizer operates as the fractionally spaced equalizer will be referred to as fractionally spaced mode.

0080Described below in reference to the timing chart of <figref idref="f0009">Fig. 11</figref> is the fixed coefficient equalizer 71 operating in symbol rate mode.

0081In <figref idref="f0009">Fig. 11</figref>, the timings of SEN1, DEN1, DT1, SEL, R0, R1, SEN2, DEN2 and DT2 are shown from the top down in chart form.

0082Since in <figref idref="f0009">Fig. 11</figref> the timings before and after the registers acting as shift registers need only be known, <figref idref="f0009">Fig. 11</figref> shows the timing chart of registers R0 and R1 indicating the timings before and after each of the registers 91₁ through 91₅ in <figref idref="f0008">Fig. 10</figref>. Thus the timing chart of the registers R0 and R1 corresponds illustratively to the timing chart of the registers 91₁ and 91₂ in <figref idref="f0008">Fig. 10</figref>.

0083As explained above in connection with <figref idref="f0006">Fig. 7</figref>, the level of SEN1 goes High at intervals of the symbol period of DT1, and the level of DEN1 goes High at intervals of half the symbol period of DT1.

0084When the fixed coefficient equalizer 71 is made to operate in symbol rate mode, the level of SEL input to the selector 81 is brought High. The register R0 shifts and holds DT1 at the timing of SEN1, i.e., per symbol period, since the level of SEL is kept High in symbol rate mode. Likewise the register R1 shifts and holds DT1 per symbol period.

0085More specifically, if the register R0 holds D0 illustratively at a given time t0, then the register R1 holds D-2 which precedes D0 by one symbol period. Thereafter, at a time t2 subsequent to the time TO by one symbol period, the register R0 shifts D0 held therein to hold D2; the register R1 shifts D-2 held therein to hold D0.

0086The above operations are repeated so that when each of the times t0, t2, t4, t6, ..., is reached one after another, the register R0 holds D0, D2, D4, D6, ..., one after another. The register R1 holds one after another D-2, D0, D2, D4, ..., which precede by one symbol period each the data held in the register R1.

0087That is, although not shown, the relationship between the register R0 and the register R1 in <figref idref="f0009">Fig. 11</figref> may be applied to the registers 91₁ through 91₅ in <figref idref="f0008">Fig. 10</figref> as follows: when each of the times t0, t2, t4, t6, ..., is reached one after another, the register 91₁ holds D0, D2, D4, D6, ..., one after another; the register 91₂ holds one after another D-2, D0, D2, D4, ..., which precede by one symbol period each the data held in the register 91₁. Likewise, the register 91₃ holds D-4, D-2, D0, D2, ...; the register 91₄ holds D-6, D-4, D-2, D0, ...; and the register 91₅ holds D-8, D-6, D-4, D-2, ... one after another.

0088When the registers 91₁ through 91₅ shift their data at the timing of SEN1 as described above, the product-sum operations subsequent to the registers 91₁ through 91₅ are carried out on every other input data item and on the filter coefficient.

0089Then the result of the product-sum operations is held in the register 83 and output as DT2 at the timing of DEN1. That is, as shown in <figref idref="f0009">Fig. 11</figref>, ODO, OD1, OD2, OD3, ... resulting from the product-sum operations on DT1, on the outputs from the registers 91₁ through 91₅, and on the filter coefficients C₂₀ through C₂₅ are output at the timing of DEN1.

0090In addition, SEN1 that was input is delayed by one clock pulse by the register 84 before being output as SEN2. Likewise, DEN1 that was input is delayed by one clock pulse by the register 85 before being output as DEN2.

0091In the manner described above, the fixed coefficient equalizer 71 operates in symbol rate mode.

0092Described below in reference to the timing chart of <figref idref="f0010">Fig. 12</figref> is the fixed coefficient equalizer 71 operating in fractionally spaced mode.

0093In <figref idref="f0010">Fig. 12</figref>, as in <figref idref="f0009">Fig. 11</figref>, the timings of SEN1, DEN1, DT1, SEL, R0, R1, SEN2, DEN2 and DT2 are shown from the top down in chart form.

0094Since in <figref idref="f0010">Fig. 12</figref> the timings before and after the registers acting as shift registers need only be known, <figref idref="f0010">Fig. 12</figref> shows the timing chart of registers R1 and R2 indicating the timings before and after each of the registers 91₁ through 91₅ in <figref idref="f0008">Fig. 10</figref>. Thus the timing chart of the registers R1 and R2 corresponds illustratively to the timing chart of the registers 91₁ and 91₂ in <figref idref="f0008">Fig. 10</figref>.

0095Also, the portions in <figref idref="f0010">Fig. 12</figref> which correspond to those in the timing chart of <figref idref="f0009">Fig. 11</figref> and which are thus repetitive will not be discussed further hereunder where appropriate.

0096When the fixed coefficient equalizer 71 is made to operate in fractionally spaced mode, the level of SEL input to the selector 81 is brought Low. The register R1 shifts and holds DT1 at the timing of DEN1, i.e., per half symbol period, since the level of SEL is kept Low in fractionally spaced mode. Likewise the register R2 shifts and holds DT1 per half symbol period.

0097That is, whereas DT1 is shifted and held per symbol period when the fixed coefficient equalizer 71 is made to operate in symbol rate mode, DT1 is shifted and held per half symbol period when the fixed coefficient equalizer 71 is made to operate in fractionally spaced mode. Thus when each of the times t0, t1, t2, t3, t4, t5, t6, ... is reached one after another (i.e., at intervals of a half symbol period), the register R1 holds D0, D1, D2, D3, D4, D5, D6, D7, ..., one after another; the register R2 holds D-1 (not shown), D0, D1, D2, D3, D4, D5, D6, ..., one after another.

0098That is, although not shown, the relationship between the register R1 and the register R2 in <figref idref="f0010">Fig. 12</figref> may be applied to the registers 91₁ through 91₅ in <figref idref="f0008">Fig. 10</figref> as follows: when each of the times t0, t1, t2, t3, t4, t5, t6, ..., is reached one after another (i.e., at intervals of a half symbol period), the register 91₁ holds D0, D1, D2, D3, D4, D5, D6, D7, ..., one after another; the register 91₂ holds one after another D-1, D0, D1, D2, D3, D4, D5, D6, ..., which precede by half the symbol period each the data held in the register 91₁. Likewise, the register 91₃ holds D-2, D-1, D0, D1, D2, D3, D4, D5, ...; the register 91₄ holds D-3, D-2, D-1, D0, D1, D2, D3, D4, ...; and the register 91₅ holds D-4, D-3, D-2, D-1, D0, D1, D2, D3, ... one after another.

0099When the registers 91₁ through 91₅ shift and hold their data at the timing of DEN1 as described above, the product-sum operations subsequent to the registers 91₁ through 91₅ are carried out on the consecutively input data and on the filter coefficient.

0100Then the result of the product-sum operations is held in the register 83 and output as DT2 at the timing of DEN1. That is, as shown in <figref idref="f0010">Fig. 12</figref>, OD0, OD1, OD2, OD3, OD4, OD5, OD6, ... resulting from the product-sum operations on DT1, on the outputs from the registers 91₁ through 91₅, and on the filter coefficients C₂₀ through C₂₅ are output at the timing of DEN1.

0101In the manner described above, the fixed coefficient equalizer 71 operates in fractionally spaced mode.

0102As explained, the fixed coefficient equalizer 71 does not incorporate two waveform equalizers and select between the two equalizers. Instead, the fixed coefficient equalizer 71 incorporates a single waveform equalizer that is controlled to select between symbol rate mode and fractionally spaced mode. This makes it possible to switch the waveform equalizer between different operation rates while having the circuit resources such as registers and product-sum operation circuits shared between the switched rates.

0103Also, in keeping with the channel to be equalized, either symbol rate mode or fractionally spaced mode can be selected freely, so that higher receiving characteristics are made available than before.

0104Described below in reference to <figref idref="f0011 f0012 f0013 f0014">Figs. 13 through 16</figref> is an example of operating the waveform equalizer 53 as an adaptive equalizer, but before the description of the equalizer workings, the principle of adaptive equalization will be explained.

0105The adaptive equalizer is a waveform equalizer used for channels of which the filter coefficient with regard to a given waveform equalizer cannot be determined beforehand. The adaptive equalizer controls its own filter coefficient adaptively while monitoring the received signal.

0106Incidentally, the expression (1) shown below is the expression of principle (filter coefficient updating expression) of the adaptive equalizer. The expression (1) gives the principle of signed LMS (Least Mean Square) for the adaptive equalizer. That is, the LMS algorithm is one example of adaptive equalization algorithm.

0107<maths id="math0001" num="(1)"><math display="block"><mtable columnalign="left"><mtr><mtd><mfenced open="[" close="]"><mi>Expression</mi><mspace width="1em" /><mn mathvariant="normal">1</mn></mfenced></mtd></mtr><mtr><mtd><msubsup><mi mathvariant="normal">C</mi><mrow><mi mathvariant="normal">n</mi><mo mathvariant="normal">+</mo><mn mathvariant="normal">1</mn></mrow><mi mathvariant="normal">i</mi></msubsup><mo mathvariant="normal">=</mo><msubsup><mi mathvariant="normal">C</mi><mi mathvariant="normal">n</mi><mi mathvariant="normal">i</mi></msubsup><mo mathvariant="normal">+</mo><mi mathvariant="normal">λ</mi><mo mathvariant="normal">⋅</mo><mfenced open="[" close="]"><msubsup><mi mathvariant="normal">X</mi><mi mathvariant="normal">n</mi><mi mathvariant="normal">i</mi></msubsup></mfenced><mo>⁢</mo><msub><mi mathvariant="normal">e</mi><mi mathvariant="normal">n</mi></msub></mtd></mtr></mtable></math><img file="EP2178221A1_D0001.tif" /></maths>

0108In the expression (1) above, [a] = 1 when a > 0, [a] = 0 when a = 0, and [a] = -1 when a < 0.

0109In the expression (1), reference character i stands for a tap number, n for a time index, λ for a coefficient, and X_nⁱ for the value at the time n and tap number i. Furthermore, reference character e_n denotes the signal called the error signal in effect at the time n and represents equalization error. The adaptive equalizer exercises control in a manner minimizing the error signal e_n while updating the filter coefficient.

0110The error signal e_n represents the difference between the output signal of the waveform equalizer and the point of transmission as the source of that signal, and is given by the following expression:

0111<maths id="math0002" num="(2)"><math display="block"><msub><mi mathvariant="normal">e</mi><mi mathvariant="normal">n</mi></msub><mo mathvariant="normal">=</mo><msub><mi mathvariant="normal">Z</mi><mi mathvariant="normal">n</mi></msub><mo mathvariant="normal">-</mo><msub><mi mathvariant="normal">d</mi><mi mathvariant="normal">n</mi></msub></math><img file="EP2178221A1_D0002.tif" /></maths>

0112In the expression (2) above, Z_n denotes the output of the waveform equalizer (i.e., DT2 above), and d_n represents the estimated value of a transmitted symbol.

0113Also, Z_n is given by the following expression:

0114<maths id="math0003" num="(3)"><math display="block"><msub><mi mathvariant="normal">Z</mi><mi mathvariant="normal">n</mi></msub><mo mathvariant="normal">=</mo><mi mathvariant="normal">Σ</mi><mo>⁢</mo><msub><mi mathvariant="normal">C</mi><mi mathvariant="normal">n</mi></msub><mo mathvariant="normal">×</mo><msub><mi mathvariant="normal">X</mi><mi mathvariant="normal">n</mi></msub></math><img file="EP2178221A1_D0003.tif" /></maths>

0115There exist a number of techniques for estimating the point of transmission d_n as the source of the output signal of the waveform equalizer. Of these techniques, two will be explained below for illustration.

0116The first technique involves transmitting a fixed sequence for equalization purposes. According to this technique, the transmitting side transmits the fixed sequence at intervals of a transmission frame. When the receiving side determines the transmission frame and the position of the fixed sequence, the receiving side can determine the transmitted signal as the source in that fixed sequence interval alone.

0117The second technique involves assuming the signal point closest to the equalized signal point as the point of transmission. Illustratively in the case of QPSK (Quadrature Phase Shift Keying), according to this technique, one of four points A, B, C and D shown in <figref idref="f0011">Fig. 13</figref> constitutes the point of transmission. If a point X on the IQ plane is assumed to be received as shown in <figref idref="f0011">Fig. 13</figref>, then the point A closest to that point X is estimated to have been transmitted and the point X to have been received.

0118These techniques are used to acquire the estimated value of the transmitted symbol.

0119As for the updating of the filter coefficient, processing is carried out only on the signal of the symbol point regardless of the symbol rate equalizer or fractionally spaced equalizer being in operation.

0120The principle of adaptive equalization has been explained so far. It should be noted, however, that attempts to install the equalizer into actual circuits often fail to bring about an installation that satisfies the expression of the principle of adaptive equalization. The principal reason for this trouble is that it takes time to compute the error signal, which in turn prolongs feedback.

0121Thus the following expression is often used in practice as the filter coefficient updating expression:

0122<maths id="math0004" num="(4)"><math display="block"><mtable columnalign="left"><mtr><mtd><mfenced open="[" close="]"><mi>Expression</mi><mspace width="1em" /><mn mathvariant="normal">2</mn></mfenced></mtd></mtr><mtr><mtd><msubsup><mi mathvariant="normal">C</mi><mrow><mi mathvariant="normal">n</mi><mo mathvariant="normal">+</mo><mn mathvariant="normal">1</mn></mrow><mi mathvariant="normal">i</mi></msubsup><mo mathvariant="normal">=</mo><msubsup><mi mathvariant="normal">C</mi><mi mathvariant="normal">n</mi><mi mathvariant="normal">i</mi></msubsup><mo mathvariant="normal">+</mo><mi mathvariant="normal">λ</mi><mo mathvariant="normal">⋅</mo><mfenced open="[" close="]"><msubsup><mi mathvariant="normal">X</mi><mrow><mi mathvariant="normal">n</mi><mo mathvariant="normal">-</mo><mi mathvariant="normal">N</mi></mrow><mi mathvariant="normal">i</mi></msubsup></mfenced><mo>⁢</mo><msub><mi mathvariant="normal">e</mi><mrow><mi mathvariant="normal">n</mi><mo mathvariant="normal">-</mo><mi mathvariant="normal">N</mi></mrow></msub></mtd></mtr></mtable></math><img file="EP2178221A1_D0004.tif" /></maths>

0123In the expression (4) above, N stands for a circuit delay.

0124That is, the expression (4) utilizes the coefficient at a time n and the error signal at a time n-N in order to compute the coefficient at a time n+1.

0125Incidentally, the expression (4) as the filter coefficient updating expression allows for a delay of N clock pulses in error computations. The N-clock-pulse delay is applied likewise to the tap X_n to be multiplied by en. This is because the error signal en and the delay at the tap X_n need to be aligned precisely with one another so as to accomplish correct equalization.

0126The principle of adaptive equalization was described above. Explained next is the adaptive equalizer carrying out filter coefficient updates using the circuits applicable to the expression (4). In describing this embodiment, it is assumed that a d-1 clock pulse delay is involved in the computations performed by these circuits.

0127<figref idref="f0012">Fig. 14</figref> is a block diagram explanatory of a detailed structure of an adaptive equalizer 101 as a variation of the waveform equalizer 53 in <figref idref="f0005">Fig. 6</figref>. The adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> is a four-tap waveform equalizer that exemplifies the waveform equalizer controlling its filter coefficient adaptively. Although the four-tap adaptive equalizer 101 is explained below as a typical adaptive equalizer for this embodiment, obviously an adaptive equalizer having other than four taps is also acceptable.

0128Because the adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> corresponds to the waveform equalizer 53 in <figref idref="f0005">Fig. 6</figref>, DT1, SEN1 and DEN1 are input from the sampling synchronizing section 52 and SEL is input from the host CPU 43. Based on these signals, the adaptive equalizer 101 performs the waveform equalizing process and outputs DT2, SEN2 and DEN2 acquired through the process.

0129In the adaptive equalizer 101, a selector 111, a register 113, a register 114, and a register 115 correspond to the selector 81, register 83, register 84, and register 85 respectively in the fixed coefficient equalizer 71 of <figref idref="f0008">Fig. 10</figref>.

0130Although they have a different number of taps each, the filter 112 and the filter 82 in <figref idref="f0008">Fig. 10</figref> have basically the same structure: registers 131₁ through 131₃, multipliers 132₀ through 132₃, and adders 133₁ through 133₃ correspond to the registers 91₁ through 91₅, multipliers 92₀ through 92₅, and adders 93₁ through 93₅ respectively in <figref idref="f0008">Fig. 10</figref>. It should be noted that the adaptive equalizer 101 of <figref idref="f0012">Fig. 14</figref> has its filter coefficient updated adaptively. Thus whereas the multipliers 92₀ through 92₅ in <figref idref="f0008">Fig. 10</figref> use the fixed filter coefficients C₂₀ through C₂₅ in their multiplications, the multipliers 132₀ through 132₃, in <figref idref="f0012">Fig. 14</figref> use adaptively updated filter coefficients C₃₀ through C₃₃ in their multiplications.

0131That is, although the adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> has basically the same structure as the fixed coefficient equalizer 71 in <figref idref="f0008">Fig. 10</figref>, the adaptive equalizer 101 is different from its counterpart 71 in <figref idref="f0008">Fig. 10</figref> in that it has another downstream block for adaptively controlling the filter coefficients (C₃₀, C₃₁, C₃₂, C₃₃ in <figref idref="f0012">Fig. 14</figref>) in keeping with the received signal. Specifically, the adaptive equalizer 101 is structured further to include an arithmetic unit 116, delay circuits 117₁ through 117₃, registers 118₁ through 118₃, multipliers 119₁ through 119₄, multipliers 120₁ through 120₄, adders 121₁ through 121₄, registers 122₁ through 122₄, and an error signal computing section 123.

0132Thus the portions of the adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> which correspond to those of the fixed coefficient equalizer 71 in <figref idref="f0008">Fig. 10</figref> are repetitive and will not be discussed further hereunder where appropriate. The description below will center on the method for adaptively controlling the filter coefficients C₃₀ through C₃₃ to be multiplied by the multipliers 132₀ through 132₃ with the outputs from DT1 and the registers 131₁ through 131₃.

0133The elements above are explained hereunder, in no particular order. DT1 is input to the arithmetic unit 116. The arithmetic unit 116 performs operations involving the so-called sgn function and returns an integer (i.e., return value) indicating the sign of the number designated as an argument. That is, the arithmetic unit 116 carries out the operations corresponding to [X_n-Nⁱ] in the above-mentioned expression (4). The arithmetic unit 116 outputs the value x acquired through the operations to the delay circuit 117₂.

0134Symbol SEN1 is input to the delay circuit 117_1. The delay circuit 117 acquires sen_d by delaying SEN1 by d clock pulses and outputs sen_d to the registers 122₁ through 122₄.

0135The value x is input to the delay circuit 117₂ from the arithmetic unit 116. The delay circuit 117₂ acquires x0 by delaying x by d clock pulses and outputs x0 to the register 118₁ and multiplier 119₁.

0136The value en is input to the delay circuit 117₃ from the selector 111. The delay circuit 117₃ acquires en_d by delaying en by d clock pulses and outputs en_d to the registers 118₁ through 118₃.

0137The value x0 from the delay circuit 117₂ and en_d from the delay circuit 117₃ are input to the register 118₁. The register 118₁ turns x0 held therein into x1 when en_d goes High, and outputs x1 to the register 118₂ and multiplier 119₂.

0138When en_d goes High, the value x2 acquired by the register 118₂ delaying x1 is output likewise to the register 118₃ and multiplier 119₃; the value x3 acquired by the register 118₃ delaying x2 is output likewise to the multiplier 119₄.

0139The result of the product-sum operations is input to the error signal computing section 123 from the adder 133₃. The error signal computing section 123 computes the error signal using the result of the product-sum operations and outputs what is computed to the multipliers 119₁ through 119₄.

0140The value x0 from the delay circuit 117₂ and err from the error signal computing section 123 are input to the multiplier 119₁. The multiplier 119₁ multiplies x0 by err and outputs the product to the multiplier-120₁.

0141The output value from the multiplier 119₁ (i.e., product of x0 and err) and the coefficient λ are input to the multiplier 120₁. The multiplier 120₁ multiplies the output value by the coefficient λ and outputs the product to the adder 121₁.

0142The output value from the multiplier 120₁ (product of x0, error, and coefficient λ) and the filter coefficient C₃₀ from the downstream register 122₁ are input to the adder 121₁. The adder 121₁ adds the output value from the multiplier 120₁ to the filter coefficient C₃₀ from the register 122₁ and outputs the sum to the register 122₁.

0143The output value from the adder 121₁ (sum of the filter coefficient C₃₀ and the product of x0, err and coefficient λ) and sen_d from the delay circuit 117₁ are input to the register 122₁. When sen_d goes High, the register 122₁ turns the retained output value from the adder 121₁ into the filter coefficient C₃₀ and outputs the filter coefficient C₃₀ to the multiplier 132₀ and adder 121₁.

0144Here, the value held in the register 122₁ is acquired through computations that utilize the expression (4) described above in conjunction with the principle of adaptive equalization.

0145That is, the multiplier 119₁ computes [X_n-Nⁱ] × e_n-N, the multiplier 120₁ computes λ × [X_n-N¹] × e_n-N, and the adder 121₁ computes C_nⁱ+λ × [X_n-Nⁱ] × e_n-N, whereby the computations corresponding to the expression (4) are carried out. The register 122₁ outputs C_n+1ⁱ acquired through the computations corresponding to the expression (4), to the multiplier 132₀ and adder 121₁. Thereafter, the multiplier 132₀ multiplies C_n+1ⁱ by DT1, and the adder 121₁ computes C_n+2ⁱ using C_n+1ⁱ.

0146As with the multiplier 119₁ through the register 122₁, the multiplier 119₂ through the register 122₂, the multiplier 119₃ through the register 122₃, and the multiplier 119₄ through the register 122₄ carry out the computations corresponding to the expression (4). The filter coefficients C₃₁ through C₃₃ acquired through the computations are output to the multipliers 132₁ through 132₃ respectively, and are multiplied by the outputs from the registers 131₁ through 131₃ respectively.

0147As described above, the adaptively acquired filter coefficients C₃₀ through C₃₃ are input to the multipliers 132₀ through 132₃. The multipliers 132₀ through 132₃ perform computations multiplying these filter coefficients by DT1 and by the outputs from the registers 131₁ through 131₃.

0148In other words, in the adaptive equalizer 101 of <figref idref="f0012">Fig. 14</figref>, the error signal computing section 123 generates the error signal based on the result of the product-sum operations coming from the adder 133₃. The values x0, x1, x2 and x3 each delayed by the same amount as the error signal thus generated are used to perform the filter coefficient updating computations of the expression (4), whereby the filter coefficients C₃₁ through C₃₃ are updated.

0149As with the fixed coefficient equalizer 71 of <figref idref="f0008">Fig. 10</figref>, when the adaptive equalizer 101 of <figref idref="f0012">Fig. 14</figref> structured as described above operates as the symbol rate equalizer because SEL input to the selector 111 is 1, a waveform-equalized output signal DT2 is output at the timing of SEN1 (i.e., per symbol period) as representative of the result of the product-sum operations that utilize the adaptively controlled filter coefficients C₃₀ through C₃₃.

0150Also, when the adaptive equalizer 101 of <figref idref="f0012">Fig. 14</figref> operates as the fractionally spaced equalizer because SEL input to the selector 111 is 0, a waveform-equalized output signal DT2 is output at the timing of DEN1 (i.e., per half symbol period) as representative of the result of the product-sum operations that utilize the adaptively controlled filter coefficients C₃₀ through C₃₃.

0151When thus controlled selectively in keeping with SEL, the adaptive equalizer 101 operates either as the symbol rate equalizer or as the fractionally spaced equalizer.

0152Described below in reference to the timing chart of <figref idref="f0013">Fig. 15</figref> is the adaptive equalizer 101 operating in symbol rate mode.

0153In <figref idref="f0013">Fig. 15</figref>, the timings of SEN1, DEN1, DT1, SEL, en, R1, R2, SEN2, DEN2, DT2, err, x0, sen__d, en_d, x1, x2, C0, and C1 are shown from the top down in chart form.

0154Since in <figref idref="f0013">Fig. 15</figref> the timings before and after the registers acting as shift registers need only be known, <figref idref="f0013">Fig. 15</figref> shows the timing chart of the registers R1 and R2 corresponding to the registers 131₁ through 131₃ in <figref idref="f0012">Fig. 14</figref>. The timing chart of x0 corresponds to the output from the delay circuit 117₂ in <figref idref="f0012">Fig. 14</figref>; the timing chart of x1 and x2 corresponds to the outputs from the registers 118₁ through 118₃ in <figref idref="f0012">Fig. 14</figref>; the timing chart of C0 and C1 corresponds to the filter coefficients C₃₀ through C₃₃ in <figref idref="f0012">Fig. 14</figref>. These relationships also hold in the timing chart of <figref idref="f0014">Fig. 16</figref>, to be discussed later.

0155As described above, the level of SEN1 goes High at intervals of the symbol period of DT1; the level of DEN1 goes High at intervals of half the symbol period of DT1.

0156When the adaptive equalizer 101 is made to operate in symbol rate mode, the level of SEL input to the selector 111 is brought High. The level of en goes High per symbol period in keeping with SEN1.

0157Since the level of SEL is kept High in symbol rate mode, R1 and R2 are shifted at the timing of SEN1, i.e., per symbol rate.

0158In this manner, the registers 131₁ through 131₃ shift and hold their data at the timing of SEN1. Thus the product-sum operations subsequent to the registers 131₁ through 131₃ are performed on every other input data item and on the filter coefficient.

0159The result of the product-sum operations is held in the register 113 and output as DT2 at the timing of DEN1. That is, Z0, Z2, Z4, Z6, Z8, ... resulting from the product-sum operations performed on DT1 and on the outputs from the registers 131₁ through 131₃ as well as on the filter coefficients C₃₀ through C₃₃ are output at the timing of DEN1.

0160Incidentally, the adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> is assumed to have the delay of d-1 clock pulses as mentioned above. Thus as shown in <figref idref="f0013">Fig. 15</figref>, using Z0 resulting from the product-sum operation on DT2, the error signal computing section 123 illustratively computes E0 delayed by d-1 clock pulses relative to Z0 as err. The delay circuit 117₂ delays D0 by d clock pulses to attain synchronism with E0.

0161Likewise, the delay circuit 117₁ delays SEN1 by d clock pulses, and the delay circuit 117₃ delays en by d clock pulses. That is, the levels of sen_d and en_d having been delayed are brought High when SEN1 and en are delayed by d clock pulses each.

0162The registers 118₁ through 118₃ shift and hold their data at the timing of en_d. Thus the operations subsequent to the registers 118₁ through 118₃ are performed on every other input data item such as D0, D2, D4, D6, D8, ...

0163Also, the registers 122₁ through 122₄ shift their retained data at the timing of sen_d. This causes the filter coefficient C0 to be output at the timing of sen_d such as C0_0, C0_2, C0_4, C0_6, C0_8, ... Likewise the filter coefficient C1 is output at the timing of sen_d such as C1_0, C1_2, C1_4, C1_6, C1_8, ...

0164In the manner described above, the adaptive equalizer 101 operates in symbol rate mode.

0165Described below in reference to the timing chart of <figref idref="f0014">Fig. 16</figref> is the adaptive equalizer 101 operating in fractionally spaced mode.

0166In <figref idref="f0014">Fig. 16</figref>, as in <figref idref="f0013">Fig. 15</figref>, the timings of SEN1, DEN1, DT1, SEL, en, R1, R2, SEN2, DEN2, DT2, err, x0, sen_d, en_d, x1, x2, C0, and C1 are shown from the top down in chart form.

0167Also, the portions in <figref idref="f0014">Fig. 16</figref> which correspond to those in the timing chart of <figref idref="f0013">Fig. 15</figref> are repetitive and thus will not be discussed further hereunder where appropriate.

0168When the adaptive equalizer 101 is made to operate in fractionally spaced mode, the level of SEL input to the selector 111 is brought Low. The level of en goes High at intervals of the half symbol period in keeping with DEN1.

0169Since the level of SEL is kept Low in fractionally spaced mode, R1 and R2 are shifted at the timing of DEN1, i.e., at intervals of the half symbol period.

0170In this manner, the registers 131₁ through 131₃ shift and hold their data at the timing of DEN1. Thus the product-sum operations subsequent to the registers 131₁ through 131₃ are performed on the successively input data and on the filter coefficient.

0171Then, the result of the product-sum operations is held in the register 113 and output as DT2 at the timing of DEN1. That is, Z0, Z1, Z2, Z3, Z4, Z5, Z6, Z7, Z8, Z9, ... resulting from the product-sum operations performed on DT1 and on the outputs from the registers 131₁ through 131₃ as well as on the filter coefficients C₃₀ through C₃₃ are output at the timing of DEN1.

0172Incidentally, the adaptive equalizer 101 in <figref idref="f0012">Fig. 14</figref> is assumed to have the delay of d-1 clock pulses as mentioned above. Thus as shown in <figref idref="f0014">Fig. 16</figref>, the error signal computing section 123 illustratively computes E0 delayed by d-1 clock pulses relative to Z0 as err. The delay circuit 117₂ delays D0 by d clock pulses to attain synchronism with E0.

0173The registers 118₁ through 118₃ shift and hold their data at the timing of en_d. Thus the operations subsequent to the registers 118₁ through 118₃ are performed on the successively input data such as D0, D1, D2, D3, D4, D5, D6, D7, D8, D9, ...

0174Also, the registers 122₁ through 122₄ shift their retained data at the timing of sen_d. This causes the filter coefficient C0 to be output at the timing of sen_d such as C0_0, C0_2, C0_4, C0_6, C0_8, ... Likewise the filter coefficient C1 is output at the timing of sen_d such as C1_0, C1_2, C1_4, C1_6, C1_8, ...

0175In the manner described above, the adaptive equalizer 101 operates in fractionally spaced mode.

0176As described, the adaptive equalizer 101 does not incorporate two waveform equalizers and select between the two equalizers. Instead, the fixed coefficient equalizer 101 incorporates a single waveform equalizer that is controlled to select between symbol rate mode and fractionally spaced mode. This makes it possible to switch the waveform equalizer between different operation rates while having the circuit resources such as registers and product-sum operation circuits shared between the switched rates.

0177In that way, it is possible to build substantially the same scope of circuitry as one waveform equalizer and still be able to switch between the symbol rate equalizer and the fractionally spaced equalizer.

0178Also, in keeping with the channel to be equalized, either symbol rate mode or fractionally spaced mode can be selected freely, so that higher receiving characteristics are made available than before.

0179The series of the processes described above may be executed either by hardware or by software. Where the processes are to be carried out by software, the programs constituting the software may be either incorporated beforehand in the dedicated hardware of the computer to be used or installed from a suitable program recording medium into a general-purpose personal computer or like equipment capable of executing diverse functions.

0180<figref idref="f0015">Fig. 17</figref> is a block diagram showing a typical structure of a personal computer for executing the above-described series of processes using programs. A CPU (Central Processing Unit) 211 performs various processes in accordance with the programs recorded in a ROM (Read Only Memory) 212 or a recording section 218. A RAM (Random Access Memory) 213 may store necessary programs and data to be executed and operated on by the CPU 211. The CPU 211, ROM 212, and RAM 213 are interconnected by a bus 214.

0181An input/output interface 215 is also connected to the CPU 211 via the bus 214. The input/output interface 215 is connected with an input section 216 typically made of a microphone and with an output section 217 typically composed of a display and speakers. The CPU 211 carries out various processes in response to commands that are input from the input section 216. The CPU 211 outputs the result of the processing to the output section 217.

0182The recording section 218 connected to the input/output interface 215 is illustratively constituted by a hard disk, and records the programs and diverse data to be executed and operated on by the CPU 211. A communication section 219 communicates with external devices via networks such as the Internet and local area networks.

0183Also, programs may be acquired through the communication section 219 and recorded to the recording section 218.

0184When removable media 221 such as magnetic disks, optical disks, magneto-optical disks or semiconductor memories are attached to a drive 220 connected to the input/output interface 215, the drive 220 drives the attached medium to acquire programs and data therefrom. The programs and data thus acquired are transferred as needed to the recording section 218 for storage therein.

0185As shown in <figref idref="f0015">Fig. 17</figref>, the program recording media for accommodating computer-installable, computer-executable programs are constituted either by the removable media 221 provided as package media such as magnetic disks (including flexible disks), optical disks (including CD-ROM (Compact Disc Read-Only Memory) and DVD (Digital Versatile Disc)), magneto-optical disks, or a semiconductor memory; or by the ROM 212 and a hard disk drive making up the recording device 218 having the programs stored thereon temporarily or permanently. The programs may be recorded to the program recording media via the communication section 219 such as a router or a modem interfaced with wired or wireless communication media including local area networks, the Internet, and digital satellite broadcasts.

0186Incidentally, in this specification, the steps describing the programs stored on the storage medium represent not only the processes that are to be carried out in the depicted sequence (i.e., on a time series basis) but also processes that may be performed parallelly or individually and not chronologically.

0187Also in this specification, the term "system" refers to an entire configuration made up of a plurality of component devices.

0188Furthermore, it should be understood that the present invention when embodied is not limited to the above-described embodiments and that various modifications, variations and alternatives may be made of the invention so far as they are within the scope of the appended claims or the equivalents thereof.