WO2012089140A1 - 时变系统中预存储滤波器系数组的选择方法及装置 - Google Patents

时变系统中预存储滤波器系数组的选择方法及装置 Download PDF

Info

Publication number
WO2012089140A1
WO2012089140A1 PCT/CN2011/084904 CN2011084904W WO2012089140A1 WO 2012089140 A1 WO2012089140 A1 WO 2012089140A1 CN 2011084904 W CN2011084904 W CN 2011084904W WO 2012089140 A1 WO2012089140 A1 WO 2012089140A1
Authority
WO
WIPO (PCT)
Prior art keywords
scene
filter
quantization
filter coefficient
quantized
Prior art date
Application number
PCT/CN2011/084904
Other languages
English (en)
French (fr)
Inventor
许百成
Original Assignee
意法·爱立信半导体(北京)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 意法·爱立信半导体(北京)有限公司 filed Critical 意法·爱立信半导体(北京)有限公司
Publication of WO2012089140A1 publication Critical patent/WO2012089140A1/zh

Links

Classifications

    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03HIMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00Networks using digital techniques
    • H03H17/02Frequency selective networks
    • H03H17/0294Variable filters; Programmable filters
    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03HIMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00Networks using digital techniques
    • H03H17/02Frequency selective networks
    • H03H17/04Recursive filters
    • H03H2017/0477Direct form I

Definitions

  • the present invention relates to the field of wireless communication and digital signal processing technologies, and in particular, to a method and apparatus for selecting a pre-stored filter coefficient set applied to a B-inch variable system.
  • Filters are widely used in the field of wireless communication and digital signal processing technology, and their characteristics are determined by the filter structure and filter coefficients.
  • the scenes of filter applications are often time-varying, that is, a set of filter coefficient groups often cannot meet the needs of all scenarios. Therefore, an adaptive filter has been developed, that is, the coefficient set of the filter is calculated and adjusted in real time. In this way, the adaptive filter can flexibly cope with various situations.
  • adaptive filters also have their shortcomings in practical applications.
  • the computational load of adaptive filters tends to be large, which is very unfavorable for the power-saving performance of devices. It also takes up extra computing resources and Storage resources increase hardware costs.
  • a more common compromise method is to pre-store different sets of filter coefficients, and select a corresponding set of coefficients according to different scenarios, which reduces the amount of calculation and meets the needs of different scenarios.
  • the two main problems that need to be solved when applying this method are: ⁇ ) Pre-storing which group coefficients; Pre-storage coefficients and mapping between various scenarios.
  • the optimal filter coefficient depends on the current channel's extension and signal-to-noise ratio, that is, the application scenario is delayed.
  • the extension and the signal-to-noise ratio together determine if all possible delay spreads and signal-to-noise ratios are quantized separately: i) RMS Delay (root mean square delay) corresponding to the delay spread is quantized to 50:50: 50ns is the starting point and 50ns is the step size.
  • the 1000 ns delay spread range is quantized to get a total of 21 values.
  • the signal-to-noise ratio SNR is quantized as -5:2 relieve5: (indicating that -5dB is the starting point and 2.5 is the step size for quantization), and the range of the signal-to-noise ratio of Bécé ⁇ is quantized to obtain a total of 15 values.
  • the technical problem to be solved by the present invention is to provide a method and a device for selecting a pre-stored filter coefficient group that should be used in a time-varying system, which can cover all application scenarios with low filter performance loss, and can greatly reduce Select the amount of calculation when pre-storing the filter coefficient group to improve the selection efficiency.
  • the present invention provides the following solutions:
  • a method for selecting a pre-stored set of filter coefficients in a time varying system comprising:
  • Step A Quantify each parameter of the defined variable scene in the system, and obtain a quantization scenario defined by the quantization parameter group, where each quantization parameter group includes a quantization parameter obtained by quantizing each parameter ;
  • Step B determining, according to the obtained quantization scene and a predetermined filter structure, a filter coefficient group corresponding to each quantization scene, and filtering caused by filtering each of the quantized scenes by using a filter having the filter coefficient group Performance loss;
  • Step C Select, according to the filter performance loss, a number of the selected quantized scenes that are greater than the first group of pre-stored filter coefficient groups, and select the filter coefficient group corresponding to the selected quantized scene as the to-be-selected Filter coefficient group;
  • Step D selecting, by using a traversal manner, a filter coefficient group having the best filtering performance for all quantized scenes and having a group number equal to the first group number, as the pre-stored Filter coefficient group.
  • the method further includes:
  • Step E obtaining, for each quantization scenario, a minimum filter that can be obtained by filtering the quantized scene by using a filter corresponding to each filter coefficient group in the pre-stored filter coefficient group by using a traversal manner The performance loss, and thus the mapping relationship between the filter coefficient group corresponding to the minimum performance loss and the quantization scenario.
  • the step A specifically includes:
  • each parameter is separately quantized according to a range of values of each parameter of the time-varying scenario defined in the time-varying system, to obtain a quantized quantization parameter;
  • a quantized scene defined by the set of quantization parameters is determined based on the quantized quantization parameters, wherein each set of quantization parameters uniquely corresponds to a quantized scene.
  • the step B specifically includes:
  • Determining by using a filter corresponding to any quantization scenario, filtering each quantized scene, and performing filtering performance loss caused by filtering each quantized scene with a filter corresponding to each quantized scene, specifically It is theoretically calculated by filtering simulation or according to the filter coefficient generation process.
  • the step C specifically includes:
  • each selection process specifically includes: removing all the items involved in the column number set from the loss matrix The elements on the column, and the elements on all the rows that are equal to the value of the column number set, are selected as the basis for selection of the selection process; based on the selection of the selection process, the selection has the most The performance loss tolerance line of the required element of the threshold, the row number of the row is added to the row number set, and the column number of all elements on the row that meet the performance loss tolerance threshold requirement is added to the column number set; When the number of the row numbers in the row number set is greater than the number of groups of the pre-stored filter coefficient groups, determining a set of quantized scenes corresponding to all the row numbers in the row number set, and further according to the set of the quantized scenes All the quantization parameter groups of the quantization scene determine the corresponding filter coefficient group to obtain a set of filter coefficient groups to be selected.
  • the number of row numbers in the row number set is less than or equal to the number of groups of pre-stored filter coefficient groups
  • the step D specifically includes:
  • the set of filter coefficients to be selected selects all possible combinations of pre-stored filter coefficient sets
  • the present invention also provides a device for selecting a filter coefficient group pre-stored in a time-varying system, and a packet scene quantizer for quantizing each parameter of a time-varying scene in the time-varying system to obtain quantization a quantization scenario defined by the parameter group, each quantization parameter group includes a quantization parameter obtained by quantizing each parameter;
  • a performance loss generator configured to determine, according to the obtained quantization scene and a predetermined filter structure, a filter coefficient group corresponding to each quantization scene, and filter each of the quantized scenes by using a filter having the filter coefficient group The resulting loss of filtering performance
  • a filter coefficient group initial selector selecting a first quantized scene to be selected that is greater than a pre-stored filter coefficient group, and filtering the corresponding quantized scene to be selected a set of coefficients as a set of filter coefficients to be selected;
  • a filter coefficient group secondary selector configured to select, by means of traversing, a group of the filter coefficients to be selected that has optimal filtering performance for all quantization scenarios, the number of groups being equal to the first group number
  • the set of filter coefficients is used as the pre-stored set of filter coefficients.
  • the above selection device further includes:
  • mapping table generator configured to obtain, by using a traversal manner, a filter corresponding to each filter coefficient group in the pre-stored filter coefficient group for each quantization scenario, and separately filtering the quantized scene The minimum filter performance loss obtained is obtained, and then the mapping relationship between the filter coefficient group corresponding to the minimum performance loss and the quantization scene is established.
  • the scene quantizer is specifically configured to:
  • each parameter is separately quantized according to a range of values of each parameter of the time-varying scenario defined in the time-varying system, to obtain a quantized quantization parameter;
  • a quantized scene defined by the set of quantization parameters is determined based on the quantized quantization parameters, wherein each set of quantization parameters uniquely corresponds to a quantized scene.
  • the performance loss generator is specifically used to:
  • the performance loss generator is further configured to obtain the filter performance loss by filtering simulation or by theoretical calculation according to a filter coefficient generation process.
  • the filter coefficient group initial selector is specifically used to:
  • each selection process specifically includes: all the involved in the plex loss matrix The elements on the column, and the elements on all the rows that are equal to the value of the column number set, are selected as the basis for selection of the selection process; based on the selection of the selection process, the selection has the most The performance loss tolerance line of the required element of the threshold, the row number of the row is added to the row number set, and the column number of all elements on the row that meet the performance loss tolerance threshold requirement is added to the column number set;
  • the filter coefficient group initial selector is further configured to: when the number of row numbers in the row number set is less than or equal to the number of groups of pre-stored filter coefficient groups, reduce the predetermined performance loss tolerance threshold, and then return The step of performing one or more selection processes on the selection basis until the number of elements in the selection basis that meet the predetermined performance loss tolerance threshold requirement is zero.
  • the filter coefficient group secondary selector is specifically configured to:
  • the method and apparatus for selecting a pre-stored filter coefficient set provided by the present invention first selects a set of filter coefficient groups to be selected by means of two selections, and then uses the traversal mode to The pre-stored filter coefficient group is selected in the filter coefficient group to be selected, and since the number of filter coefficient groups to be selected is smaller than the number of quantization scenes, the combination of possible pre-stored filter coefficient groups is greatly Reduced, so embodiments of the invention can be greatly reduced Traversing the amount of computation increases the selection efficiency of the pre-stored filter coefficient set.
  • the filter coefficient group to be selected is selected according to whether the filtering performance loss satisfies the tolerance threshold for the first time, and the pre-stored filter coefficient group is selected based on the filter coefficient group to be selected.
  • JA can cover all quantization scenarios with lower performance loss, and can give a mapping relationship between pre-stored filter coefficient groups and all quantization scenes.
  • FIG. 1 is a flowchart of a method for selecting a pre-storage filter coefficient group according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of a general filter structure referenced in the embodiment of the present invention
  • FIG. 3 is a schematic structural diagram of a device for selecting a pre-stored filter coefficient group according to an embodiment of the present invention. Detailed ways
  • the invention provides a method for selecting a pre-storage filter coefficient group, and performs pre-storage filter coefficient group selection by means of two screening methods. When the same task is completed, the calculation amount is greatly reduced, and can be given
  • the mapping relationship between the M sets of filter coefficients and all scenes covers all time-varying scenarios with low performance loss.
  • the time-varying scene is defined by a total of P parameters of ⁇ ,..., and the optimal filter coefficients will be generated based on these P parameters. That is, the P parameters uniquely define a specific time-varying scene, and uniquely define a set of optimal filter coefficients corresponding to the time-varying scene.
  • a method for selecting a pre-storage filter coefficient group applied to a time-varying system specifically includes the following steps:
  • Step 11 Quantify each parameter of the time-varying system in the time-varying system to obtain a quantization scenario defined by the quantization parameter group, and each quantization parameter group includes a quantization parameter obtained by quantizing each parameter.
  • Step 12 Determine, according to the obtained quantization scene and a predetermined filter structure, a filter coefficient group corresponding to each quantization scene, and filter that is caused by filtering each of the quantized scenes by using a filter having the filter coefficient group Loss of performance.
  • Step 3 According to the filter performance loss, select a first set of quantized scenes whose number is greater than a pre-stored filter coefficient group, and select a filter coefficient group corresponding to the selected quantized scene as a to-be-selected Filter coefficient set.
  • Step 14 Select, by using a traversal manner, a filter coefficient group having the best filtering performance for all quantization scenarios and having the number of groups equal to the first group number, as the pre-stored Filter coefficient group.
  • the embodiment adopts the method of filtering twice to finally determine the pre-stored filter coefficient group, wherein the first screening is from the filter coefficient group corresponding to all the quantization scenarios, and the number of screening is greater than the first a set of filter coefficients (to be assumed to be M) to be selected; and then, by a second screening, from the set of filter coefficients to be selected, the pre-stored filter of the first set of numbers is filtered out Group of coefficients. It can be seen that the second screening is performed based on the set of filter coefficients to be selected obtained by the first screening, and the possible combinations thereof are common.
  • the implementation may further include the following steps:
  • Step 15 For each quantization scenario, obtain, by using a traversal manner, a minimum filter that can be obtained by separately filtering the quantized scene by using a filter corresponding to each filter coefficient group in the pre-stored filter coefficient group.
  • the performance loss and thus the mapping relationship between the filter coefficient group corresponding to the minimum performance loss and the quantization scenario.
  • the set of filter coefficients mapped by the quantized scene is a set of filter coefficients having a minimum filter performance loss for the quantized scene in a pre-stored set of filter coefficients.
  • the present embodiment provides a mapping relationship between the group of filter coefficients of the group and all the quantized scenes, and thus these mapping relationships can be saved, so as to be actually filtered. According to this, select the appropriate filter coefficient group.
  • the above step 11 specifically includes:
  • Step ⁇ 1 According to the value range of each parameter of the time-varying scenario in the time-varying system, each parameter is separately quantized to obtain a quantized quantization parameter;
  • Step 112 Determine, according to the quantized quantization parameter, a quantization scene defined by a quantization parameter group, where each quantization parameter group uniquely corresponds to a quantization scene.
  • the filter coefficient group corresponding to the quantization scene may be generated according to the quantization parameter group of each quantization scenario, and the filter corresponding to the quantization scene is determined.
  • the possible time-varying scene is quantized to obtain a quantized scene, and specifically, each parameter used to define the time-varying scene is quantized. That is, for the parameter ft, e [l, P], the possible values from small to large are quantized as , ⁇ ⁇ , its minimum possible value, v3 ⁇ 4 is its maximum possible value, ⁇ represents the quantization that can be obtained after the parameter ft is quantized. The number of values.
  • Each quantization scene is uniquely determined by e [l, Ay, l, U.
  • the optimal filter coefficients depend on the delay spread and signal to noise ratio of the current channel, that is, its time varying scene. It is determined by the two parameters of delay spread and signal to noise ratio.
  • Each parameter has a value range.
  • the cell radius determines the value range of the delay extension. Based on the cell radius in the prior art, the delay extension ranges from 0 to 5000 ns (this delay spread). The corresponding RMS delay is approximately 0 ⁇ 1000ns).
  • the minimum signal-to-noise ratio required by the system during normal operation may be used as the lower limit of the signal-to-noise ratio (the lowest signal-to-noise ratio may be found according to relevant technical standards or device parameters); Ideally without noise (taking into account the thermal noise of the board) the signal-to-noise ratio of the channel to determine the highest signal-to-noise ratio as the upper limit of the signal-to-noise ratio.
  • those skilled in the art can determine according to factors affecting the parameter or related technical standards. A possible range of values for this parameter is not described here.
  • the above step 12 specifically includes:
  • Step 121 Determine, for a predetermined filter structure, a filter coefficient group corresponding to each quantization scene according to a quantization parameter group corresponding to each quantization scene.
  • the corresponding filter coefficient group is determined according to the quantization parameter group, and can be determined according to the manner in which the corresponding filter coefficient is determined according to the parameter in the prior art.
  • Step i22 Determine, according to the filter coefficient group corresponding to each quantization scenario, a filter corresponding to each quantization scene.
  • Step 123 Determine, for each quantization scenario, filtering each quantized scene by using a filter corresponding to any quantization scenario, and filtering each quantized scene by using a filter corresponding to each quantized scene.
  • Loss of filtering performance can be theoretically calculated by filtering the truth or according to the filter coefficient generation process. The following are examples:
  • e [l, N] represents the filter performance loss caused by filtering the first scene with the filter corresponding to the first scene.
  • the performance loss matrix obtained by simulation can be obtained as follows:
  • the filter coefficient generation process is obtained by theoretical calculation. Please refer to the filter structure used in FIG. 2, which includes a plurality of delay units 22 and a plurality of adders 21 for filtering the input signal X(n) to obtain The filtered output signal Y(n).
  • ai , a 2 ... a p are the coefficients of the feedback loop
  • bj, b 2 ... b Q are the coefficients of the forward path.
  • / ⁇ denotes the filter sample ⁇ ( ⁇ ) i ideal value, which represents the value of the actual filtered output of the filtered sample, ie Y(n) in Fig. 2;
  • X represents the input signal X(n);
  • W « Represents the filter coefficient set corresponding to the first quantization scene;
  • R represents fj in the jth quantization scene, output Y ( n ) the autocorrelation matrix of the input samples, ie
  • Ri represents the cross-correlation vector of the vectors I and h in the first quantization scene, which is a column direction:
  • the above step 13 specifically includes:
  • Step 131 generating a loss matrix L of the filter performance loss, the elements of the missing matrix / ⁇ f,./ e [l,N] represents the element in the . / column of the row of the loss matrix L, which means that the filter corresponding to the first quantization scene filters the ./th quantization scene, as opposed to / Performance loss caused by filters corresponding to the quantized scene filtering the . / quantization scene.
  • Step 132 Perform more than one selection process on the selection basis until the number of elements in the selection basis that meet the predetermined performance loss tolerance threshold requirement is zero, wherein each selection process specifically includes:
  • Step 133 The number of row numbers in the row number set is greater than the number of groups of pre-stored filter coefficient groups (M), determine a set of quantized scenes corresponding to all row numbers in the row number set, and further perform the quantization according to the A set of quantization parameters of all quantized scenes in the set of scenes, determining a corresponding set of filter coefficients, and obtaining a set of filter coefficients to be selected.
  • M pre-stored filter coefficient groups
  • Step 134 When the number of row numbers in the row number set is less than or equal to the number of groups (M) of pre-stored filter system arrays, after the predetermined performance loss tolerance threshold is decreased, return to step 132.
  • Step 13 determines the initial set of filter coefficients based on the performance loss matrix L and the tolerance threshold to determine the initial ⁇ ' ( (' is usually greater than ⁇ ).
  • the tolerance threshold ⁇ is the maximum filter performance loss that can be tolerated when filtering the _/th quantization scene by the filter corresponding to the second quantization scene.
  • the threshold is ⁇ to ensure that the M' group filter coefficients obtained by the initial selection cover all quantized scenes ⁇ , and the maximum performance loss introduced will not exceed.
  • the above steps 131 to 134 can be specifically implemented by the following matrix operations:
  • Step I defines a flag matrix F whose size is ⁇ ⁇ ⁇ , the element 0, ./ e [l, N] in the flag matrix represents the element located in the ith row and column of the flag matrix F.
  • the elements in the flag matrix are initially set according to the loss matrix L and the tolerance threshold p.
  • the tolerance threshold P is less than the tolerance threshold P, the performance loss caused by filtering the first quantization scene by the filter corresponding to the first quantization scene is less than tolerance.
  • the threshold P is set, the corresponding position in the flag matrix is set to 1, otherwise it is set to 0.
  • it can be represented by the following pseudo code: For i::: 1: N
  • Step 2 Select the initial filter group of the M' group to be selected, and step 2 may include the following steps:
  • Step A determining whether there is an element other than 0 in the flag matrix F, and if yes, proceeding to step B; otherwise, proceeding to step D;
  • Step B selecting a row having the largest element and value from the flag matrix F, recording the row number of the row into the row number set, and recording the column number of all the elements not on the row on the row to the column number
  • the collection ⁇ Afo ⁇ then go to step C.
  • Step C The element in the flag matrix F is set according to the column number of the column number record, and the setting process includes: setting the column number in the flag matrix F equal to the column number of the column number record All elements are set to 0, and all elements on the line in the flag matrix F whose row number is equal to the column number of the column number set record are set to 0, and then returns to step A.
  • Step D determining whether the number of line numbers recorded in the line number set is greater than the number of groups of pre-stored filter coefficient groups (M), and if yes, proceeding to step E, otherwise, decreasing the tolerance threshold ⁇ and returning to the step.
  • the tolerance threshold p by reducing the tolerance threshold p, the number of line numbers recorded in the line number set can be increased. Since the number of row numbers recorded in the row number set is the number M' of the filter coefficient group to be selected, M' is larger, and the subsequent operation amount is larger, that is, the larger the tolerance threshold, the more the filtering performance requirement is. Low, and the subsequent calculation is smaller; the smaller the tolerance threshold p, the higher the filtering performance requirement, and the larger the subsequent calculation. Therefore, it can be set by considering the filtering performance requirement of the system and the subsequent calculation amount.
  • Step E determining a set of quantized scenes corresponding to the row numbers recorded in the row number set, and further determining corresponding filter coefficient groups according to the quantization parameter groups of all the quantized scenes in the set of the quantized scenes, to obtain a set of to be selected Filter coefficient group. For example, when the line number of the line number record recorded in the line number set includes line numbers of 3, 5, 27, 122, etc., the quantized scene corresponding to the line numbers includes the quantized scenes of the third, fifth, 27, 122, etc., These can be determined from the set of quantization parameters for these quantization scenarios. The filter coefficient group corresponding to the scene is quantized to obtain a filter coefficient group to be selected.
  • step 2 can be specifically expressed by the following pseudo code:
  • the scene covered by the recording coefficient ⁇ is Afapp ie Fd ⁇ , :) is not 0 scene number inildxArrayim ) ⁇ f ifia /7save the index fj -------- 0, i G idxMapped , j ---- ---- 1,... ⁇ ' ?
  • the M' number of filter coefficient groups to be selected are recorded in an array. If ⁇ 'less than M can change the tolerance threshold to a smaller value, repeat steps 1 and 2 until M' is greater than or equal to M
  • the above steps 14 1 ⁇ 2 include:
  • Step 141 Select, according to the number of groups of pre-stored filter coefficient groups, the set of filter coefficients to be selected to select all possible combinations of pre-stored filter coefficient sets.
  • Step 142 For each combination, determine, by using a traversal manner, a minimum performance loss that can be obtained by filtering each quantized scene by a filter corresponding to each filter coefficient group in each combination, and calculating each combination. The sum of the minimum performance losses that can be achieved for all quantized scenarios.
  • Step 143 selecting a combination corresponding to the smallest sum value as a pre-stored filter system array.
  • the above step 14 is to select M groups from the M' filter coefficient groups for pre-storage, so there are a total of ⁇ / (possible combinations. Here, by traversing all possible combinations, the best ⁇ is found. a set of filter coefficients for pre-storage.
  • the specific process can be represented by the following pseudo code. C Possible combinations of M
  • i best arg min[E(i)], / ⁇ 1,2,, . . U
  • the optimal M pre-stored filter coefficient sets are selected.
  • the foregoing step 15 is used to generate a mapping relationship between the quantization scene and the filter coefficient group in the pre-stored filter coefficient group, and the mapping relationship may be determined according to the performance loss matrix L, that is, the pre-stored filter coefficient.
  • L the performance loss matrix
  • a set of filter coefficient groups having the smallest filter performance loss for each quantization scene is found as a filter coefficient group mapped by the quantization scene. This is represented by the form of the map T.
  • the following pseudo code can be used to indicate the following:
  • the embodiment of the present invention further provides a pre-storage filter coefficient group selection device in a time-varying system.
  • the selection device specifically includes:
  • each parameter of the time-varying scene defined in the time-varying system is quantized, and a quantization scene defined by the quantization parameter group is obtained, and each quantization parameter group includes the quantized each parameter Quantitative parameter;
  • a performance loss generator configured to determine, according to the obtained quantization scene and a predetermined filter structure, a filter coefficient group corresponding to each quantization scene, and filter each of the quantized scenes by using a filter having the filter coefficient group The resulting loss of filtering performance;
  • the filter coefficient group initial selector, ffi selects, according to the filter performance loss, a number of the quantized scenes to be selected that is greater than the first group of pre-stored filter coefficient groups, and filters the corresponding quantized scene to be selected. a set of coefficients as a set of filter coefficients to be selected;
  • a filter coefficient group secondary selector configured to select, by traversing mode, a filter coefficient having a group number equal to the first group number for optimal filtering performance from all the filter coefficient groups to be selected Group, as the pre-stored filter coefficient set.
  • the selecting apparatus may further include: a mapping table generator, configured to obtain, by using a traversal manner, filtering corresponding to each filter coefficient group in the pre-stored filter coefficient group for each quantization scenario The minimum filter performance loss that can be obtained by filtering the quantized scene separately, and then establishing a mapping relationship between the filter coefficient group corresponding to the minimum performance loss and the quantized scene.
  • a mapping table generator configured to obtain, by using a traversal manner, filtering corresponding to each filter coefficient group in the pre-stored filter coefficient group for each quantization scenario The minimum filter performance loss that can be obtained by filtering the quantized scene separately, and then establishing a mapping relationship between the filter coefficient group corresponding to the minimum performance loss and the quantized scene.
  • the scene quantizer is specifically configured to:
  • quantized scenes defined by the set of quantization parameters are determined, wherein each set of quantization parameters uniquely corresponds to a quantized scene. For a predetermined filter structure, determining a filter coefficient group corresponding to each quantization scene according to a quantization parameter group corresponding to each quantization scene;
  • the performance loss generator is further used for theoretically calculating by filter simulation or according to a filter coefficient generation process to obtain the filter performance loss.
  • the filter coefficient group initial selector is specifically configured to:
  • each selection process specifically includes: removing the column number set from the loss matrix The elements on all the columns, and the elements on all the rows that are equal to the value of the set of column numbers, are selected as the basis for selection of the selection process; based on the selection of the selection process, the selection has the most a row of elements of a predetermined performance loss tolerance threshold requirement, adding the row number of the row to the row number set, and adding the column number of all elements of the row that meet the performance loss tolerance threshold requirement to the column number set ;
  • the filter coefficient set initial selector is further configured to reduce the predetermined performance loss tolerance threshold when the number of row numbers in the row number set is less than or equal to the number of groups of pre-stored filter coefficient groups Thereafter, the step of performing one or more selection processing on the selection basis until the number of elements in the selection base that meet the predetermined performance loss tolerance threshold requirement is zero is returned.
  • the filter coefficient group secondary selector is specifically configured to:
  • the minimum performance loss that can be obtained by filtering each quantized scene by the filter corresponding to each filter coefficient group in each combination is determined by traversing mode, if each combination is for all The sum of the minimum performance losses that can be obtained by quantifying the scene;

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Mathematical Physics (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Description

本发明涉及无线通信和数字信号处理技术领域, 特别涉及一种应用于 B寸 变系统的预存储滤波器系数组的选择方法和装置。
滤波器广泛地应用于无线通信和数字信号处理技术领域, 其特性由滤波 器结构和滤波系数决定。 在无线通信系统和数字信号处理领域中, 滤波器应 用的场景在很多时候是时变的, 也就是说一组滤波器系数组往往不能满足所 有场景的需求。 于是, 人们研究出了自适应的滤波器, 即滤波器的系数组是 实时计算并调整的。 这样, 自适应滤波器可以灵活地应对各种情况。 然而, 自适应滤波器在实际应用中也有其不可忽视的缺点, 自适应滤波器的计算量 往往比较大, 这对设备的省电性能是非常不利的, 同^还要占用额外的计算 资源和存储资源, 增加了硬件成本。
实际上, 较为常用的一种折中的方法是, 预存储凡组不同的滤波器系数 组, 根据不同的场景选择相应的一组系数组, 这样既降低了计算量又能满足 不同场景的需求。 应用这种方法所需要解决两个主要问题是: ί) 预存储哪凡 组系数; 预存储系数与各种场景之间的映射关系。
在实际选择过程中, 如果采用比较直观的遍历式检测的方法进行选择, 其运算量是非常巨大的。 以 LTE系统中的频域信道估计为例, 在给定滤波器 结构的情况下, 最优的滤波器系数取决于当前信道的^延扩展和信噪比, 也 即, 其应用场景由时延扩展和信噪比共同决定, 如果将所有可能的时延扩展 和信噪比分别进行量化: i) 按照时延扩展对应的 RMS Delay (均方根时延) 量化为 50:50: (表示以 50ns为起始点, 50ns为步长进行量化), 则 1000 ns 的时延扩展范围量化后得到一共 21个值。 ii) 信噪比 SNR量化为 - 5:2„5: (表 示以- 5dB为起始点, 2.5为步长进行量化), 贝 ύί Β的信噪比的范围量化后 得到一共 15个值。由任一时延扩展的量化参数和任一信噪比的量化参数组成 的组合, 可以定义一个量化场景, 这样, 一共可以得到 21 X 15= 315个量化 场景。 这 315个量化场景对应着最优的 315组滤波系数组。 如果在实际应用 中预存储 8组滤波器系数组, 就需要从 3】5组中选择出 8组滤波器系数组来 覆盖所有的场景, 那么可能的组合共有 5 =2, 19811236238094e+ 15种, 将超过 200 万亿种情况。 对于如此巨大的数字, 如果采用直观的遍历式检验的方式 进行选择, 巨量运算将耗费大量时间, 在实际应^中几乎是不可行的。 发明内容
本发明所要解决的技术问题是提供一种应^于时变系统的预存储滤波器 系数组的选择方法和装置, 能够以较低的滤波性能损失来覆盖所有的应用场 景, 并能够极大地降低选择预存储滤波器系数组时的运算量, 提高选择效率。
为解决上述技术问题, 本发明提供方案如下:
一种时变系统中预存储的滤波器系数组的选择方法, 包括:
步骤 A, 对所述 ^变系统中定义^变场景的每个参数进行量化, 获得由 量化参数组所定义的量化场景, 每个量化参数组包括有所述每个参数量化后 得到的量化参数;
步骤 B, 根据所获得的量化场景和预定的滤波器结构, 确定每一量化场 景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波器对所有量化 场景分别进行滤波所引起的滤波性能损失;
步骤 C, 根据所述滤波性能损失, 选择出数量大于预存储的滤波器系数 组的第一组数的待选择的量化场景, 将该待选择的量化场景对应的滤波器系 数组作为待选择的滤波器系数组;
步骤 D, 通过遍历方式, 从待选择的滤波器系数组中选择出针对所有量 化场景具有最佳滤波性能的、 组数等于所述第一组数的滤波器系数组, 作为 所述预存储的滤波器系数组。
优选地, 上述选择方法中, 所述步骤 D之后还包括:
步骤 E, 针对每一量化场景, 通过遍历方式, 获得采用所述预存储的滤 波器系数组中的每一滤波器系数组对应的滤波器, 对该量化场景分别进行滤 波所能够获得的最小滤波性能损失, 进而建立该最小性能损失对应的滤波器 系数组与该量化场景之间的映射关系。 优选地, 上述选择方法中, 所述步骤 A具体包括:
根据所述时变系统中定义时变场景的每个参数的取值范围, 对每个参数 分别进行量化, 得到量化后的量化参数;
根据量化后的量化参数, 确定由量化参数组所定义的量化场景, 其中, 每个量化参数组唯一对应于一个量化场景。
优选地, 上述选择方法中, 所述步骤 B具体包括:
对于预定的滤波器结构, 根据每个量化场景对应的量化参数组, 确定每 个量化场景对应的滤波器系数组;
根据每个量化场景对应的滤波器系数组, 确定每个量化场景对应的滤波 器;
针对每一量化场景, 确定采用任一量化场景对应的滤波器对该每一量化 场景进行滤波, 相对于采^该每一量化场景对应的滤波器对该每一量化场景 进行滤波所引起的滤波性能损失。
优选地, 上述选择方法中,
所述确定采用任一量化场景对应的滤波器对该每一量化场景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每一量化场景进行滤波所引起 的滤波性能损失, 具体是通过滤波仿真或根据滤波器系数生成过程通过理论 计算得到的。
优选地, 上述选择方法中, 所述步骤 C具体包括:
生成一滤波性能损失的损失矩阵, 所述损失矩阵的元素 ^表示第 i个量 化场景对应的滤波器对第 个量化场景进行滤波,相对于第 /个量化场景对应 的滤波器对第 个量化场景进行滤波所引起的性能损失;
对选择基础进行一次以上的选择处理, 直至选择基础中符合预定的性能 损失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体包括: 从损 失矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述列号集合 的值相等的所有行上的元素, 得到本次选择处理的选择基础; 在本次选择处 理的选择基础上, 选择出具有最多的、 符合预定的性能损失容忍门限要求的 元素的行, 将该行的行号增加到行号集合中, 同时将该行上符合所述性能损 失容忍门限要求的所有元素的列号增加到列号集合中; 在所述行号集合中行号的数量大于预存储的滤波器系数组的组数时, 确 定所述行号集合中所有行号所对应的量化场景的集合, 进而根据所述量化场 景的集合中所有量化场景的量化参数组, 确定对应的滤波器系数组, 得到一 组待选择的滤波器系数组。
优选地, 上述选择方法中,
在所述行号集合中行号的数量小于或等于预存储的滤波器系数组的组数
B寸, 减小所述预定的性能损失容忍门限后, 返回所述对选择基础进行一次以 上的选择处理, 直至选择基础中符合预定的性能损失容忍门限要求的元素的 数目为零的步骤。
优选地, 上述选择方法中, 所述步骤 D具体包括:
根据预存储的滤波器系数组的组数, 丛所述待选择滤波器系数组选择出 预存储的滤波器系数组的所有可能的组合;
针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波器系数组 对应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能损失, 计 算该每一组合针对所有量化场景所能获得的最小性能损失的和值;
选择出最小的所述和值所对应的组合, 作为预存储的滤波器系数组。 本发明还提供了一种时变系统中预存储的滤波器系数组的选择装置, 包 场景量化器,用于对所述时变系统中定义时变场景的每个参数进行量化, 获得由量化参数组所定义的量化场景, 每个量化参数组包括有所述每个参数 量化后得到的量化参数;
性能损失生成器, 用于根据所获得的量化场景和预定的滤波器结构, 确 定每一量化场景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波 器对所有量化场景分别进行滤波所引起的滤波性能损失;
滤波器系数组初始选择器, ^于根据所述滤波性能损失, 选择出数量大 于预存储的滤波器系数组的第一组数的待选择的量化场景, 将该待选择的量 化场景对应的滤波器系数组作为待选择的滤波器系数组;
滤波器系数组二次选择器, 用于通过遍历方式, 从待选择的滤波器系数 组中选择出针对所有量化场景具有最佳滤波性能的、 组数等于所述第一组数 的滤波器系数组, 作为所述预存储的滤波器系数组。
优选地, 上述的选择装置中, 还包括:
映射表生成器, 用于针对每一量化场景, 通过遍历方式, 获得采用所述 预存储的滤波器系数组中的每一滤波器系数组对应的滤波器, 对该量化场景 分别进行滤波所能够获得的最小滤波性能损失, 进而建立该最小性能损失对 应的滤波器系数组与该量化场景之间的映射关系。
优选地, 上述的选择装置中,
所述场景量化器具体用于:
根据所述时变系统中定义时变场景的每个参数的取值范围, 对每个参数 分别进行量化, 得到量化后的量化参数;
根据量化后的量化参数, 确定由量化参数组所定义的量化场景, 其中, 每个量化参数组唯一对应于一个量化场景。
优选地, 上述的选择装置中,
所述性能损失生成器具体用于:
对于预定的滤波器结构, 根据每个量化场景对应的量化参数组, 确定每 个量化场景对应的滤波器系数组;
根据每个量化场景对应的滤波器系数组, 确定每个量化场景对应的滤波 器;
针对每一量化场景, 确定采 ^任一量化场景对应的滤波器对该每一量化 场景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每一量化场景 进行滤波所引起的滤波性能损失。
优选地, 上述的选择装置中,
优选地, 所述性能损失生成器进一步用于通过滤波仿真或根据滤波器系 数生成过程通过理论什算, 得到所述滤波性能损失。
优选地, 上述的选择装置中,
所述滤波器系数组初始选择器具体用于:
生成一滤波性能损失的损失矩阵, 所述损失矩阵的元素 ^表示第 i个量 化场景对应的滤波器对第 个量化场景进行滤波,相对于第_/个量化场景对应 的滤波器对第 _/个量化场景进行滤波所引起的性能损失; 对选择基础进行一次以上的选择处理, 直至选择基础中符合预定的性能 损失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体包括: 丛损 失矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述列号集合 的值相等的所有行上的元素, 得到本次选择处理的选择基础; 在本次选择处 理的选择基础上, 选择出具有最多的、 符合预定的性能损失容忍门限要求的 元素的行, 将该行的行号增加到行号集合中, 同时将该行上符合所述性能损 失容忍门限要求的所有元素的列号增加到列号集合中;
在所述行号集合中行号的数量大于预存储的滤波器系数组的组数时, 确 定所述行号集合中所有行号所对应的量化场景的集合, 进而根据所述量化场 景的集合中所有量化场景的量化参数组, 确定对应的滤波器系数组, 得到一 组待选择的滤波器系数组。
优选地, 上述的选择装置中,
所述滤波器系数组初始选择器进一步用于在所述行号集合中行号的数量 小于或等于预存储的滤波器系数组的组数时, 减小所述预定的性能损失容忍 门限后, 返回所述对选择基础进行一次以上的选择处理, 直至选择基础中符 合预定的性能损失容忍门限要求的元素的数目为零的步骤。
优选地, 上述的选择装置中,
所述滤波器系数组二次选择器具体用于:
根据预存储的滤波器系数组的组数, 从所述待选择滤波器系数组选择出 预存储的滤波器系数组的所有可能的组合;
针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波器系数组 对应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能损失, 计 算该每一组合针对所有量化场景所能获得的最小性能损失的和值;
选择出最小的所述和值所对应的组合, 作为预存储的滤波器系数组。 从以上所述可以看出, 本发明提供的预存储滤波器系数组的选择方法和 装置, 通过两次选择的方式, 首先选择出一组待选择的滤波器系数组, 进而 利用遍历方式, 从待选择的滤波器系数组中选择出预存储的滤波器系数组, 由于待选择的滤波器系数组的数量要小于量化场景的数量, 从而可能的预存 储的滤波器系数组的组合也就大大减少, 因此本发明实施例可以极大地降低 遍历运算量, 提高预存储的滤波器系数组的选择效率。 并且, 本发明实施例 在首次选择 B寸根据滤波性能损失是否满足容忍门限来选择出待选择的滤波器 系数组, 再基于该待选择的滤波器系数组选择出预存储的滤波器系数组, JA 而能够较低的性能损失来覆盖所有的量化场景, 并且能够给出预存储滤波器 系数组和所有量化场景之间的映射关系。 附图说明
图 1为本发明实施例所述的预存储滤波器系数组的选择方法的流程图; 图 2为本发明实施例中引用的通用的滤波器结构的示意图;
图 3 为本发明实施例所述预存储滤波器系数组的选择装置的结构示意 图。 具体实施方式
本发明提供了一种预存储滤波器系数组的选择方法, 通过两次筛选的方 式进行预存储滤波器系数组选择, 在完成相同任务的情况下, 极大地降低了 运算量, 并能够给出 M组滤波器系数和所有场景之间的映射关系, 以较低的 性能损失来覆盖所有的时变场景。
为不失一般性和便于后文描述, 首先进行以下定义或说明:
1、 假设最终保留的预存储的滤波器系数组的数目为 M。
2、 假设时变场景由由 ^,… 共 P个参数定义, 最优的滤波器系数将 根据这 P个参数生成。 即这 P个参数既唯一地定义了一个特定的时变场景, 也唯一地定义了对应于这一个时变场景的一组最优滤波器系数。
3、如何根据尸个参数生成滤波器系数组可以利用现有技术的方法, 为节 约篇幅, 本文中不再赘述。
请参照图 1, 本发明实施例所述的应用于时变系统的预存储滤波器系数 组的选择方法, 具体包括以下步骤:
步骤 11, 对时变系统中定义时变场景的每个参数进行量化, 获得由量化 参数组所定义的量化场景, 每个量化参数组包括有所述每个参数量化后得到 的量化参数。 步骤 12, 根据所获得的量化场景和预定的滤波器结构, 确定每一量化场 景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波器对所有量化 场景分别进行滤波所引起的滤波性能损失。
步骤】 3, 根据所述滤波性能损失, 选择出数量大于预存储的滤波器系数 组的第一组数的待选择的量化场景, 将该待选择的量化场景对应的滤波器系 数组作为待选择的滤波器系数组。
步骤 14, 通过遍历方式, 从待选择的滤波器系数组中选择出针对所有量 化场景具有最佳滤波性能的、 组数等于所述第一组数的滤波器系数组, 作为 所述预存储的滤波器系数组。
通过以上步骤, 本实施例采^两次筛选的方式, 最终确定预存储的滤波 器系数组, 其中第一次筛选是从所有量化场景对应的滤波器系数组中, 筛选 出数量大于所述第一组数 (假设为 M) 的待选择的滤波器系数组; 然后, 通 过第二次筛选, 从待选择的滤波器系数组中, 筛选出数量为所述第一组数的 预存储的滤波器系数组。 可以看出, 第二次筛选, 是基于第一次筛选所得到 的待选择的滤波器系数组进行, 其可能的组合共有 个。 而现有技术从 所有量化场景对应的滤波器系数组中选择, 其可能的组合共有 = 个, Ν 表示所有量化场景的数量。 由于待选择的滤波器系数组的数量要小于量化场 景的数量, 从而可能的预存储的滤波器系数组的组合也就大大减少, 因此本 实施倒可以极大地降低步骤 14中的遍历运算量,提高预存储的滤波器系数组 的选择效率。
在上述步骤 14之后, 本实施倒还可以进一步包括以下步骤:
步骤 15, 针对每一量化场景, 通过遍历方式, 获得采用所述预存储的滤 波器系数组中的每一滤波器系数组对应的滤波器, 对该量化场景分别进行滤 波所能够获得的最小滤波性能损失, 进而建立该最小性能损失对应的滤波器 系数组与该量化场景之间的映射关系。
也就是说, 对于一个量化场景, 该量化场景所映射的滤波器系数组, 是 在预存储的滤波器系数组中对于该量化场景具有最小滤波性能损失的滤波器 系数组。这样, 通过以上步骤 15, 本实施例给出了 Μ组滤波器系数组和所有 量化场景之间的映射关系, 进而可以保存这些映射关系, 以便于在实际滤波 中据此选择合适的滤波器系数组。
为了帮助理解上述步骤, 以下对上述步骤做进一步的说明。
上述步骤 11具体包括:
步骤 Π 1 ,根据所述时变系统中定义时变场景的每个参数的取值范围,对 每个参数分别进行量化, 得到量化后的量化参数;
步骤 112, 根据量化后的量化参数, 确定由量化参数组所定义的量化场 景, 其中, 每个量化参数组唯一对应于一个量化场景。
本实施例中, 可以根据每个量化场景的量化参数组, 生成该量化场景对 应的滤波器系数组, 确定该量化场景对应的滤波器。
对于时变场景来说, 通过 P个参数可以定义无限多个时变场景, 然而在 实际的预存储滤波器系数组选择过程中, 只能对有限个场景进行处理。 本实 施例通过以上步骤 11 , 对可能的时变场景进行量化, 得到量化场景, 具体来 说, 是对用于定义时变场景的各个参数进行量化。 即对于参数 ft, e [l,P], 将 其可能值从小到大依次量化为 , . ·ν , 为其最小可能值, v¾为 其最大可能值, ^表示参数 ft量化后能够得到的量化值的个数。 每一个量化 场景由 e [l,Ay, l,U唯一确定。 籍此, 可将所有可能场景 量化为 A^I¾ 个场景。 例如, 以 UTE系统中的频域信道估计为例,在给定滤波器结构的情况下, 最优的滤波器系数取决于当前信道的时延扩展和信噪比, 也即, 其时变场景 由时延扩展和信噪比这两个参数共同决定。 每个参数都有一个取值范围, 例 如, 小区半径决定了时延扩展的取值范围, 基于现有技术中的小区半径, 时 延扩展的取值范围大概在 0〜5000ns (该时延扩展对应的 RMS delay大约在 0〜1000ns)。 再例如, 对于当前信道的信噪比, 则可以根据系统正常工作时 要求的最低信噪比, 作为信噪比的下限 (最低信噪比可以根据相关技术标准 或设备参数查找的); 再根据没有噪声的理想情况下(可以考虑电路板的热噪 声) 信道的信噪比, 来确定最高信噪比, 作为信噪比的上限。 对于其它各种 参数, 本领域技术人员都能够根据影响该参数的因素或相关技术标准, 确定 该参数的一个可能的取值范围, 此处不再一一赘述。
对于 ΕΤΈ系统中的频域信道估 % 如果: 0 按照时延扩展 RMS Delay 量化为 50:50: (表示以 50ns为起始点, 50ns为步长进行量化), 则 0〜1000 ns 的时延扩展范围量化后得到一共 21个值。 ii) 信噪比 SNR量化为 5:2.5 : (表 示以- 5dB为起始点, 2.5为步长进行量化), 贝^ 5dB〜25dB的信噪比的范 量 化后得到一共 15个值。每个量化场景由任一时延扩展的量化参数和任一信噪 比的量化参数组成的组合所定义, 这样可以定义得到 2】 X 15= 315个量化场 景, 每个量化场景唯一对应于一组量化参数组。
上述步骤 12具体包括:
步骤 121, 对于预定的滤波器结构, 根据每个量化场景对应的量化参数 组, 确定每个量化场景对应的滤波器系数组。 这里, 根据量化参数组, 确定 对应的滤波器系数组, 可以按照现有技术的根据参数确定对应的滤波器系数 的方式进行确定。
步骤 i22, 根据每个量化场景对应的滤波器系数组, 确定每个量化场景 对应的滤波器。
步骤 123, 针对每一量化场景, 确定采用任一量化场景对应的滤波器对 该每一量化场景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每 一量化场景进行滤波所引起的滤波性能损失。 这里, 滤波性能损失可以通过 滤波饬真或根据滤波器系数生成过程通过理论计算得到的, 以下分别举例说 明:
假设 , , e [l,N]表示用第 个场景对应的滤波器对第 个场景进行滤波所 引起的滤波性能损失。 所谓通过仿真得到性能损失矩阵, 可按照如下方式得 到:
1 ) 使 ffi第 个量化场景对应于滤波器系数对第 个量化场景进行滤波, 得到滤波输出的信噪比为 SV¾;,,./ e [l,N]; 使用第 个量化场景对应于滤波器 系数对第 J个量化场景进行滤波, 得到滤波输出的信噪比为 SW^, [1,N]。
2 ) 如果信噪比用 dB表示, 则 5Λ¾ Λ [1 ] ; 如果信噪比用 线性值表示, 贝 υ /; [ , ΛΠ
Figure imgf000012_0001
所谓根据滤波器系数生成过程通过理论计算得到, 请参考图 2 用的滤波器结构, 其中包括多个延迟单元 22和多个加法器 21, 用于对输入 信号 X(n)进行滤波处理, 得到滤波后的输出信号 Y(n)。 图 2 中, ai、 a2...ap 为反馈环路的系数, bj、 b2...bQ为前向通路的系数, 这里, 中的下标 P表
2
:滤波器中横向结构网络的阶数, 即横向结构网络由 P个前向通路 m 要 P个延迟单元 22 , bQ中的下标 Q表示滤波器中反馈网络的阶数, 即反馈网 络由 Q个反馈环路构成, 需要 Q个延迟单元 22。 图 2表 ^ '以n下^的T滤^波 i^ffi枉, 可以^以下公式表示:
p
Yin) - Z hp . X{n― p) - aj(n― q )
Figure imgf000013_0001
引起的误差
E h— h
E \ |/i -- W( · Χ
E 2 R
::: 1 - 2 Re(wi0
Figure imgf000013_0002
0[1, Λ'·]
匕式中: /ί表示滤波样点 Υ(η) i理想值, 表示滤波样点的实际滤波 输出的值, 即图 2中的 Y(n); X表示输入信号 X(n); W«表示用于第 个量 化场景对应的滤波器系数组; (" 2,--' ,^,'- ], 这里表示为一个行 向量; R 表示第 j个量化场景下 ffi于 ii 输出 Y ( n ) 的输入样点的自相关 矩 阵 , 即
I [X(n), X(n - 1), - - - X(n -- P), Y(n - 1), Υ(η - 2), - - - Υ(η -- 0)
Figure imgf000013_0003
Ri 表示第 /个量化场景下向量 I与 h的互相关向量, 是一个列向:
2 ) i
Figure imgf000013_0004
上述步骤 13具体包括:
步骤 131, 生成一滤波性能损失的损失矩阵 L, 失矩阵的元素 /^ f,./ e [l,N]表示位于损失矩阵 L第 行第. /列的元素,其含义为第 个量化场 景对应的滤波器对第 ./个量化场景进行滤波,相对于第. /个量化场景对应的滤 波器对第. /个量化场景进行滤波所引起的性能损失。
步骤 132, 对选择基础进行一次以上的选择处理, 直至选择基础中符合 预定的性能损失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体 包括:
从损失矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述 列号集合的值相等的所有行上的元素, 得到本次选择处理的选择基础;
在本次选择处理的选择基础上, 选择出具有最多的、 符合预定的性能损 失容忍门限要求的元素的行, 将该行的行号增加到行号集合中, 同 B寸将该行 上符合所述性能损失容忍门限要求的所有元素的列号增加到列号集合中。
步骤 133, 在所述行号集合中行号的数量大于预存储的滤波器系数组的 组数 (M ) 确定所述行号集合中所有行号所对应的量化场景的集合, 进 而根据所述量化场景的集合中所有量化场景的量化参数组, 确定对应的滤波 器系数组, 得到一组待选择的滤波器系数组。
步骤 134, 在所述行号集合中行号的数量小于或等于预存储的滤波器系 数组的组数 (M ) 时, 减小所述预定的性能损失容忍门限后, 返回步骤 132。
步骤 13根据性能损失矩阵 L和容忍门限 来确定出初始的 Μ' ( Μ'通常 会大于 Μ ) 组滤波器系数组。 这里的容忍门限 ρ是指, 用第 /个量化场景对 应的滤波器对第_/个量化场景进行滤波时, 所能容忍的的最大的滤波性能损 失。该门限是 ^来保证通过初始选择得到的 M'组滤波器系数在覆盖所有量化 场景^, 所引入的最大性能损失不会超过 。 这里, 上述步骤 131〜134具体 可以通过以下的矩阵操作来实现:
步骤 I , 定义一个标志矩阵 F , 其大小为 Ν Χ Ν, 标志矩阵中的元素 0,./ e [l, N]表示位于标志矩阵 F第 i行第 列的元素。 根据损失矩阵 L和容 忍门限 p对标志矩阵中的元素进行初始置位, 当 小于容忍门限 P时 , 即用 第 个量化场景对应的滤波器对第 个量化场景进行滤波所引起的性能损失小 于容忍门限 P时, 将标志矩阵中的相应位置的 . 置 1 , 反之则置 0。 具体可用 如下伪代码表示: for i::: 1: N
forj = 1 : N
./;, 0
if < p
Λ i
end
end
步骤 2 ,选择初始的 M'组的待选择的滤波器系数组,步骤 2具悻又可以包 括以下步骤:
步骤 A,判断标志矩阵 F中是否存在不为 0的元素,若是,则进入步骤 B; 否则进入步骤 D;
步骤 B, 从标志矩阵 F中选择出具有最大的元素和值 的一行, 将该行的 行号记录到行号集合 中, 并将该行上所有不为 0 的元素的列号 记录到列号集合^ Afo^ 中, 然后进入步骤 C。
步骤 C,根据列号集合记录的列号,对标志矩阵 F中的元素进行置位处理, 所述置位处理包括: 将标志矩阵 F中列号等于列号集合记录的列号的列上的 所有元素置 0, 以及将标志矩阵 F中行号等于列号集合记录的列号的行上的 所有元素置 0, 然后返回步骤 A。
步骤 D,判断行号集合中记录的行号的数量是否大于预存储的滤波器系数 组的组数 (M), 若是, 则进入步骤 E, 否则, 减小所述容忍门限 ^后返回步 骤 。
这里, 减小容忍门限 p, 可以使得行号集合中记录的行号的数量增加。 由 于行号集合中记录的行号的数量即是待选择的滤波器系数组的数量 M' , M' 越大, 后续的运算量也就越大, 即容忍门限 越大, 滤波性能要求就越低, 而后续计算量越小; 容忍门限 p越小, 滤波性能要求就越高, 而后续 算量 越大。 因此, 可以综合考虑系统的滤波性能需求和后续的运算量来设定。
步骤 E, 确定行号集合中记录的行号对应的量化场景的集合, 进而根据所 述量化场景的集合中所有量化场景的量化参数组,确定对应的滤波器系数组, 得到一组待选择的滤波器系数组。 例如, 在行号集合中记录的行号记录的行 号包括 3、 5、 27、 122等行号时, 则这些行号对应的量化场景包括第 3、 5、 27、 122 等量化场景, 于是可以根据这些量化场景的量化参数组, 确定这些 量化场景对应的滤波器系数组, 得到待选择的滤波器系数组。
上述步骤 2具体可用如下伪代码表示:
m ― 0;
N
while : ,. > 0
Figure imgf000016_0001
= arg χ(β:) Ι ?' "' Λ
记录系数^ 覆盖的场景为 Afapp 即 Fd^ , :)中不为 0的场景序号 inildxArrayim ) ^ fifia /7save the index f j -------- 0, i G idxMapped , j -------- 1,… Λ'?— ;
f] ; = 0, / = L ' * - N, j G idxA4apped;
m ― m + 1
end
Mf - m
通过滤波器系数组初始选择器, M'个待选择的滤波器系数组的序号记录 在数组 中。 如果 Μ'小于 M可以将容忍门限换成一个更小的值, 重 复步骤 1和 2, 直至 M'大于或等于 M
上述步骤 14具体包括:
步骤 141, 根据预存储的滤波器系数组的组数, 丛所述待选择滤波器系数 组选择出预存储的滤波器系数组的所有可能的组合。
步骤 142, 针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波 器系数组对应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能 损失, 计算该每一组合针对所有量化场景所能获得的最小性能损失的和值。
步骤 143, 选择出最小的所述和值所对应的组合, 作为预存储的滤波器系 数组。
上述步骤 14是从 M'个滤波器系数组中选出 M组用于预存储, 因此共有 ί/ ( 种可能的组合方式。 这里通过对所有可能的组合进行遍历式检验, 找 到最佳的 Μ个用于预存储的滤波器系数组。具体过程可用如下伪代码表示如 C
Figure imgf000017_0001
possible combinations of M
// candidates from initial candidates set
// C is with size U x M
for / =^ 1 : U
Cf = C(/,:);/7 get ith possible combination in C
E( = 0
for j:::: 1: N
:E(0 E(0 -f miR
end
end
ibest = arg min[E(i)], / ^ 1,2,, . . U
idxArrayBest ^ C(U;
这样, 通过以上步骤 14进行滤波器系数组的二次筛选, 最佳的 M个预存 储的滤波器系数组就被选择出来了。
上述歩骤 15用以生成量化场景与预存储的滤波器系数组中的滤波器系数 组之间的映射关系, 该映射关系可以根据性能损失矩阵 L确定下来, 也即在 预存储的滤波器系数组中, 找到对于每一个量化场景的滤波性能损失最小的 一组的滤波器系数组, 作为该量化场景所映射的滤波器系数组。 这里通过映 射表 T的形式表示出来。 具体可用如下伪代码表示如下:
for j - 1: N
/cm;.,™ argmin(' ! ) ,Α' e i bcArrayBest
τ(./Χ
end
以上介绍了本实施例所述的时变系统中预存储滤波器系数组的选择方法, 基于以上选择方法, 本发明实施例还提供了一种时变系统中预存储滤波器系 数组的选择装置, 请参照图 3 , 所述选择装置具体包括:
场景量化器, ^于对所述时变系统中定义时变场景的每个参数进行量化, 获得由量化参数组所定义的量化场景, 每个量化参数组包括有所述每个参数 量化后得到的量化参数;
性能损失生成器, 用于根据所获得的量化场景和预定的滤波器结构, 确定 每一量化场景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波器 对所有量化场景分别进行滤波所引起的滤波性能损失; 滤波器系数组初始选择器, ffi于根据所述滤波性能损失, 选择出数量大于 预存储的滤波器系数组的第一组数的待选择的量化场景, 将该待选择的量化 场景对应的滤波器系数组作为待选择的滤波器系数组;
滤波器系数组二次选择器, 用于通过遍历方式, 从待选择的滤波器系数组 中选择出针对所有量化场景具有最佳滤波性能的、 组数等于所述第一组数的 滤波器系数组, 作为所述预存储的滤波器系数组。
优选地, 所述选择装置还可以包括: 映射表生成器, 用于针对每一量化场 景, 通过遍历方式, 获得采用所述预存储的滤波器系数组中的每一滤波器系 数组对应的滤波器, 对该量化场景分别进行滤波所能够获得的最小滤波性能 损失, 进而建立该最小性能损失对应的滤波器系数组与该量化场景之间的映 射关系。
优选地, 所述场景量化器具体用于:
根据所述^变系统中定义 ^变场景的每个参数的取值范围,对每个参数分 别进行量化, 得到量化后的量化参数;
根据量化后的量化参数, 确定由量化参数组所定义的量化场景, 其中, 每 个量化参数组唯一对应于一个量化场景。 对于预定的滤波器结构, 根据每个量化场景对应的量化参数组, 确定每个 量化场景对应的滤波器系数组;
根据每个量化场景对应的滤波器系数组, 确定每个量化场景对应的滤波 ;
针对每一量化场景,确定采用任一量化场景对应的滤波器对该每一量化场 景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每一量化场景进 行滤波所引起的滤波性能损失。
优选地,所述性能损失生成器进一歩用于通过滤波仿真或根据滤波器系数 生成过程通过理论计算, 得到所述滤波性能损失。
优选地, 所述滤波器系数组初始选择器具体用于:
生成一滤波性能损失的损失矩阵, 所述损失矩阵的元素^表示第 i个量 化场景对应的滤波器对第 个量化场景进行滤波,相对于第„/个量化场景对应 的滤波器对第 个量化场景进行滤波所引起的性能损失;
对选择基础迸行一次以上的选择处理,直至选择基础中符合预定的性能损 失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体包括: 从损失 矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述列号集合的 值相等的所有行上的元素, 得到本次选择处理的选择基础; 在本次选择处理 的选择基础上, 选择出具有最多的、 符合预定的性能损失容忍门限要求的元 素的行, 将该行的行号增加到行号集合中, 同时将该行上符合所述性能损失 容忍门限要求的所有元素的列号增加到列号集合中;
在所述行号集合中行号的数量大于预存储的滤波器系数组的组数时,确定 所述行号集合中所有行号所对应的量化场景的集合, 进而根据所述量化场景 的集合中所有量化场景的量化参数组, 确定对应的滤波器系数组, 得到一组 待选择的滤波器系数组。
优选地,所述滤波器系数组初始选择器进一步用于在所述行号集合中行号 的数量小于或等于预存储的滤波器系数组的组数时, 减小所述预定的性能损 失容忍门限后, 返回所述对选择基础进行一次以上的选择处理, 直至选择基 础中符合预定的性能损失容忍门限要求的元素的数目为零的步骤。
优选地, 所述滤波器系数组二次选择器具体用于:
根据预存储的滤波器系数组的组数,从所述待选择滤波器系数组选择出预 存储的滤波器系数组的所有可能的组合;
针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波器系数组对 应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能损失, if算 该每一组合针对所有量化场景所能获得的最小性能损失的和值;
选择出最小的所述和值所对应的组合, 作为预存储的滤波器系数组。 以上所述仅是本发明的实施方式, 应当指出, 对于本技术领域的普通技术 人员来说, 在不脱离本发明原理的前提下, 还可以作出若干改进和润饰, 这 些改进和润饰也应视为本发明的保护范围。

Claims

U 一种时变系统中预存储的滤波器系数组的选择方法, 其特征在于, 包 括:
步骤 A, 对所述 B寸变系统中定义 B寸变场景的每个参数进行量化, 获得由 量化参数组所定义的量化场景, 每个量化参数组包括有所述每个参数量化后 得到的量化参数;
步骤 B, 根据所获得的量化场景和预定的滤波器结构, 确定每一量化场 景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波器对所有量化 场景分别进行滤波所引起的滤波性能损失;
步骤(:, 根据所述滤波性能损失, 选择出数量大于预存储的滤波器系数 组的第一组数的待选择的量化场景, 将该待选择的量化场景对应的滤波器系 数组作为待选择的滤波器系数组;
步骤 D, 通过遍历方式, 从待选择的滤波器系数组中选择出针对所有量 化场景具有最佳滤波性能的、 组数等于所述第一组数的滤波器系数组, 作为 所述预存储的滤波器系数组。
2、 如权利要求 I 所述的选择方法, 其特征在于, 所述步骤 D之后还包 步骤 E, 针对每一量化场景, 通过遍历方式, 获得采用所述预存储的滤 波器系数组中的每一滤波器系数组对应的滤波器, 对该量化场景分别进行滤 波所能够获得的最小滤波性能损失, 进而建立该最小性能损失对应的滤波器 系数组与该量化场景之间的映射关系。
3、 如权利要求 I所述的选择方法, 其特征在于, 所述步骤 A具体包括: 根据所述时变系统中定义时变场景的每个参数的取值范围, 对每个参数 分别进行量化, 得到量化后的量化参数;
根据量化后的量化参数, 确定由量化参数组所定义的量化场景, 其中, 每个量化参数组唯一对应于一个量化场景。
4、 如权利要求 3所述的选择方法, 其特征在于, 所述步骤 B具体包括: 对于预定的滤波器结构, 根据每个量化场景对应的量化参数组, 确定每 个量化场景对应的滤波器系数组;
根据每个量化场景对应的滤波器系数组, 确定每个量化场景对应的滤波 器;
针对每一量化场景, 确定采 ^任一量化场景对应的滤波器对该每一量化 场景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每一量化场景 进行滤波所引起的滤波性能损失。
5、 如权利要求 4所述的选择方法, 其特征在于,
所述确定采用任一量化场景对应的滤波器对该每一量化场景进行滤波, 相对于采用该每一量化场景对应的滤波器对该每一量化场景进行滤波所引起 的滤波性能损失, 具体是通过滤波仿真或根据滤波器系数生成过程通过理论 计算得到的。
6、 如权利要求 1所述的选择方法, 其特征在于, 所述步骤 C具体包括: 生成一滤波性能损失的损失矩阵, 所述损失矩阵的元素 ^表示第 /个量 化场景对应的滤波器对第 个量化场景进行滤波,相对于第 J个量化场景对应 的滤波器对第 J个量化场景进行滤波所引起的性能损失;
对选择基础进行一次以上的选择处理, 直至选择基础中符合预定的性能 损失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体包括: 从损 失矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述列号集合 的值相等的所有行上的元素, 得到本次选择处理的选择基础; 在本次选择处 理的选择基础上, 选择出具有最多的、 符合预定的性能损失容忍门限要求的 元素的行, 将该行的行号增加到行号集合中, 同^将该行上符合所述性能损 失容忍门限要求的所有元素的列号增加到列号集合中;
在所述行号集合中行号的数量大于预存储的滤波器系数组的组数时, 确 定所述行号集合中所有行号所对应的量化场景的集合, 进而根据所述量化场 景的集合中所有量化场景的量化参数组, 确定对应的滤波器系数组, 得到一 组待选择的滤波器系数组。
7、 如权利要求 6所述的选择方法, 其特征在于,
在所述行号集合中行号的数量小于或等于预存储的滤波器系数组的组数 时, 减小所述预定的性能损失容忍门限后, 返回所述对选择基础进行一次以 上的选择处理, 直至选择基础中符合预定的性能损失容忍门限要求的元素的 数目为零的步骤。
8、 如权利要求〗所述的选择方法, 其特征在于, 所述步骤 D具体包括: 根据预存储的滤波器系数组的组数, 从所述待选择滤波器系数组选择出 预存储的滤波器系数组的所有可能的组合;
针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波器系数组 对应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能损失, 计 算该每一组合针对所有量化场景所能获得的最小性能损失的和值;
选择出最小的所述和值所对应的组合, 作为预存储的滤波器系数组。
9、 一种时变系统中预存储的滤波器系数组的选择装置, 其特征在于, 包 括:
场景量化器,用于对所述 B寸变系统中定义 B寸变场景的每个参数进行量化, 获得由量化参数组所定义的量化场景, 每个量化参数组包括有所述每个参数 量化后得到的量化参数;
性能损失生成器, 用于根据所获得的量化场景和预定的滤波器结构, 确 定每一量化场景对应的滤波器系数组, 以及采用具有该滤波器系数组的滤波 器对所有量化场景分别进行滤波所引起的滤波性能损失;
滤波器系数组初始选择器, 用于根据所述滤波性能损失, 选择出数量大 于预存储的滤波器系数组的第一组数的待选择的量化场景, 将该待选择的量 化场景对应的滤波器系数组作为待选择的滤波器系数组;
滤波器系数组二次选择器, ^于通过遍历方式, 从待选择的滤波器系数 组中选择出针对所有量化场景具有最佳滤波性能的、 组数等于所述第一组数 的滤波器系数组, 作为所述预存储的滤波器系数组。
10、 如权利要求 9所述的选择装置, 其特征在于, 还包括- 映射表生成器, 用于针对每一量化场景, 通过遍历方式, 获得采用所述 预存储的滤波器系数组中的每一滤波器系数组对应的滤波器, 对该量化场景 分别进行滤波所能够获得的最小滤波性能损失, 进而建立该最小性能损失对 应的滤波器系数组与该量化场景之间的映射关系。
11、 如权利要求 9所述的选择装置, 其特征在于, 所述场景量化器具体用于:
根据所述时变系统中定义时变场景的每个参数的取值范围, 对每个参数 分别进行量化, 得到量化后的量化参数;
根据量化后的量化参数, 确定由量化参数组所定义的量化场景, 其中, 每个量化参数组唯一对应于一个量化场景。
12、 如权利要求 11所述的选择装置, 其特征在于,
所述性能损失生成器具体用于:
对于预定的滤波器结构, 根据每个量化场景对应的量化参数组, 确定每 个量化场景对应的滤波器系数组;
根据每个量化场景对应的滤波器系数组, 确定每个量化场景对应的滤波 器;
针对每一量化场景, 确定采 ^任一量化场景对应的滤波器对该每一量化 场景进行滤波, 相对于采 ^该每一量化场景对应的滤波器对该每一量化场景 进行滤波所引起的滤波性能损失。
13、 如权利要求 12所述的选择装置, 其特征在于,
优选地, 所述性能损失生成器进一步用于通过滤波仿真或根据滤波器系 数生成过程通过理论 算, 得到所述滤波性能损失。
14、 如权利要求 9所述的选择装置, 其特征在于,
所述滤波器系数组初始选择器具体用于:
生成一滤波性能损失的损失矩阵, 所述损失矩阵的元素 ^表示第 i个量 化场景对应的滤波器对第 个量化场景进行滤波,相对于第 /个量化场景对应 的滤波器对第 个量化场景进行滤波所引起的性能损失;
对选择基础进行一次以上的选择处理, 直至选择基础中符合预定的性能 损失容忍门限要求的元素的数目为零, 其中, 每次选择处理具体包括: 从损 失矩阵中剔除列号集合所涉及的所有列上的元素, 以及剔除与所述列号集合 的值相等的所有行上的元素, 得到本次选择处理的选择基础; 在本次选择处 理的选择基础上, 选择出具有最多的、 符合预定的性能损失容忍门限要求的 元素的行, 将该行的行号增加到行号集合中, 同时将该行上符合所述性能损 失容忍门限要求的所有元素的列号增加到列号集合中; 在所述行号集合中行号的数量大于预存储的滤波器系数组的组数时, 确 定所述行号集合中所有行号所对应的量化场景的集合, 进而根据所述量化场 景的集合中所有量化场景的量化参数组, 确定对应的滤波器系数组, 得到一 组待选择的滤波器系数组。
1 5、 如权利要求 14所述的选择装置, 其特征在于,
所述滤波器系数组初始选择器进一步用于在所述行号集合中行号的数量 小于或等于预存储的滤波器系数组的组数时, 减小所述预定的性能损失容忍 门限后, 返回所述对选择基础迸行一次以上的选择处理, 直至选择基础中符 合预定的性能损失容忍门限要求的元素的数目为零的步骤。
16、 如权利要求 9所述的选择装置, 其特征在于,
所述滤波器系数组二次选择器具体用于:
根据预存储的滤波器系数组的组数, 从所述待选择滤波器系数组选择出 预存储的滤波器系数组的所有可能的组合;
针对每一组合, 通过遍历方式, 确定该每一组合中的每一滤波器系数组 对应的滤波器对每一量化场景分别进行滤波所能够获得的最小性能损失, 计 算该每一组合针对所有量化场景所能获得的最小性能损失的和值;
选择出最小的所述和值所对应的组合, 作为预存储的滤波器系数组。
PCT/CN2011/084904 2010-12-31 2011-12-29 时变系统中预存储滤波器系数组的选择方法及装置 WO2012089140A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201010624297.0A CN102158199B (zh) 2010-12-31 2010-12-31 时变系统中预存储滤波器系数组的选择方法及装置
CN201010624297.0 2010-12-31

Publications (1)

Publication Number Publication Date
WO2012089140A1 true WO2012089140A1 (zh) 2012-07-05

Family

ID=44439425

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/084904 WO2012089140A1 (zh) 2010-12-31 2011-12-29 时变系统中预存储滤波器系数组的选择方法及装置

Country Status (2)

Country Link
CN (1) CN102158199B (zh)
WO (1) WO2012089140A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107733830A (zh) * 2016-08-12 2018-02-23 中兴通讯股份有限公司 一种多载波信号产生的方法、装置及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102158199B (zh) * 2010-12-31 2014-02-19 意法·爱立信半导体(北京)有限公司 时变系统中预存储滤波器系数组的选择方法及装置
CN106330133B (zh) * 2016-08-11 2019-03-26 哈尔滨工业大学 一种时变数字滤波器的实现方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101315772A (zh) * 2008-07-17 2008-12-03 上海交通大学 基于维纳滤波的语音混响消减方法
CN101702696A (zh) * 2009-11-25 2010-05-05 北京天碁科技有限公司 信道估计的实现方法和装置
CN102158199A (zh) * 2010-12-31 2011-08-17 意法·爱立信半导体(北京)有限公司 时变系统中预存储滤波器系数组的选择方法及装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3964092B2 (ja) * 2000-02-17 2007-08-22 アルパイン株式会社 オーディオ用適応イコライザ及びフィルタ係数の決定方法
US20060128326A1 (en) * 2004-12-09 2006-06-15 Interdigital Technology Corporation Low complexity adaptive channel estimation
US7239115B2 (en) * 2005-04-04 2007-07-03 Power-One, Inc. Digital pulse width modulation controller with preset filter coefficients

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101315772A (zh) * 2008-07-17 2008-12-03 上海交通大学 基于维纳滤波的语音混响消减方法
CN101702696A (zh) * 2009-11-25 2010-05-05 北京天碁科技有限公司 信道估计的实现方法和装置
CN102158199A (zh) * 2010-12-31 2011-08-17 意法·爱立信半导体(北京)有限公司 时变系统中预存储滤波器系数组的选择方法及装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107733830A (zh) * 2016-08-12 2018-02-23 中兴通讯股份有限公司 一种多载波信号产生的方法、装置及系统

Also Published As

Publication number Publication date
CN102158199B (zh) 2014-02-19
CN102158199A (zh) 2011-08-17

Similar Documents

Publication Publication Date Title
KR20150079479A (ko) 폴라 프로세싱 메커니즘을 포함하는 컴퓨팅 시스템 및 그것의 동작 방법
RU2341030C2 (ru) Многомасштабная беспроводная связь
WO2018196786A1 (zh) Polar码的速率匹配方法及装置
EP2873155A1 (en) Method and device for performance evaluation of forward error correction (fec) codes
WO2012089140A1 (zh) 时变系统中预存储滤波器系数组的选择方法及装置
CN116032702B (zh) 自适应信道估计方法、装置、计算机设备和存储介质
TW201635736A (zh) 一種信號檢測方法及裝置
TW201025955A (en) Transmitter and the method therefor
CN103595736B (zh) 视频监控系统中的访问请求处理方法和装置
KR20220107287A (ko) 데이터 처리 장치, 방법, 기지국 및 저장 매체
TWI422182B (zh) 基於幾何方法欠定多輸入多輸出系統之高效率解碼器及其解碼方法
WO2021232566A1 (zh) 扩频信号发送方法、扩频信号接收方法、装置、设备及介质
WO2018126788A1 (zh) 准循环低密度奇偶校验编码方法、装置和存储介质
KR102067050B1 (ko) 스파스 코드 기반 엠비언트 백스캐터 시스템에서 액세스 포인트의 통신 방법 및 그 시스템
CN101355367B (zh) 干扰消除的方法和装置
JP6559789B2 (ja) Ofdmシステムにおけるpapr低減のためのツリー探索トーン予約
WO2022052933A1 (zh) 一种信号处理方法及相关设备
WO2014127663A1 (zh) 一种插值滤波方法及插值滤波器
CN113794479A (zh) 极化调节卷积码的生成方法及相关设备
CN109962875B (zh) 信号处理方法、装置及存储介质
WO2013097782A1 (zh) 一种视频传输、接收方法及装置
CN112671704B (zh) 一种攻击感知的mMTC切片资源分配方法、装置及电子设备
CN114285711B (zh) 缩放信息传播方法及其在定点fft的vlsi实现中的应用
CN101930736A (zh) 基于子带滤波框架的解码器的音频均衡方法
WO2014097583A1 (ja) 周波数オフセット補償装置および周波数オフセット補償方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11854519

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC

122 Ep: pct application non-entry in european phase

Ref document number: 11854519

Country of ref document: EP

Kind code of ref document: A1