Transformation between time domain and frequency domain based on nearly orthogonal filter banks

Abstract

A filter bank for signal decomposition is provided. The filter bank comprises a plurality of filter units each of which has one input and two outputs forming two paths whose transfer functions are complementary to each other, where the plurality of filter units are connected to form a tree structure.

Claims

1. A filter bank for signal decomposition, comprising: a plurality of filter units having one input and two outputs, wherein the two outputs comprise two paths that have complementary transfer functions, and wherein the plurality of filter units are connected to one another to form a tree structure, wherein the filter bank decomposes signals having N.sub.c sub-carrier signals, the filter bank having N.sub.s stages, stage s having 2.sup.s levels, wherein N.sub.s=log.sub.2N.sub.c, s is a stage number, and s[0, 1 . . . N.sub.s1], wherein an n.sup.th order impulse response coefficient of an s.sup.th stage, q.sup.th level filter unit, h.sub.s,q(n), is calculated by multiplying an n.sup.th order impulse response coefficient of s.sup.th stage, p.sup.th level filter unit, h.sub.s,p(n), and a rotation factor, wherein the rotation factor is a complex exponential factor, and wherein the rotation factor is $e^{j\frac{2}{N_c}\left(\tilde{p}-\tilde{q}\right)},$ wherein {tilde over (p)} is a first value of a first bit reversed version of a first N.sub.s1 bits binary encode of p, and {tilde over (q)} is a second value of a second bit reversed version of a second N.sub.s1 bits binary encode of q.

2. The filter bank of claim 1, wherein two outputs of an s.sup.th stage, l.sup.th level filter unit are respectively connected to inputs of an (s+1).sup.th stage, (2l).sup.th level filter unit and an (s+1).sup.th stage, (2l+1).sup.th level filter unit, where l[0, 1 . . . 2.sup.s1].

3. A filter bank for signal composition, comprising: a plurality of filter units having one output and two inputs, wherein the two inputs comprise two paths that have complementary transfer functions, and wherein the plurality of filter units are connected to one another to form a tree structure, wherein the filter bank composes signals having N.sub.c sub-carrier signals, the filter bank having N.sub.s stages, stage s having 2.sup.s levels, wherein N.sub.s=log.sub.2N.sub.c, s is a stage number, and s[0, 1 . . . N.sub.s1], wherein an n.sup.th order impulse response coefficient of an s.sup.th stage, q.sup.th level filter unit, h.sub.s,q(n), is calculated by multiplying an n.sup.th order impulse response coefficient of s.sup.th stage, p.sup.th level filter unit, h.sub.s,p(n), and a rotation factor, where the rotation factor is a complex exponential factor, and wherein the rotation factor is $e^{j\frac{2}{N_c}\left(\tilde{p}-\tilde{q}\right)},$ wherein {tilde over (p)} is a first value of a first bit reversed version of a first N.sub.s1 bits binary encode of p, and {tilde over (q)} is a second value of a second bit reversed version of a second N.sub.s1 bits binary encode of q.

4. The filter bank of claim 3, wherein two outputs of an s.sup.th stage, l.sup.th level filter unit are respectively connected to inputs of an (s+1).sup.th stage, (2l).sup.th level filter unit and an (s+1).sup.th stage, (2l+1).sup.th level filter unit, where l[0, 1 . . . 2.sup.s1].

5. A receiver, comprising: a first filter bank for decomposing signals, the signals containing N.sub.c sub-carrier signals and composed by a second filter bank of a transmitter, where both the first filter bank and the second filter bank have N.sub.s stages, and stage s of both the first filter bank and the second filter bank comprises 2.sup.s levels to form N.sub.c channels, where N.sub.s=log.sub.2N.sub.c, s is a stage number, and s[0, 1 . . . N.sub.s1], and where vector form transfer function {right arrow over (H)}.sub.r,p of channel p of the first filter bank is orthogonal to vector form transfer function {right arrow over (H)}.sub.t,q of channel q of the second filter bank.

6. The receiver of claim 5, wherein: when p=q, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p equals to 1; when |pq|=1, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p is less than a predetermined threshold; and otherwise, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p equals to 0, where [ ].sup.H is a conjugate transpose operation.

7. A signal composing method, comprising: feeding N.sub.c sub-carrier signals into N.sub.c inputs of a tree-structured filter bank, respectively, wherein the filter bank has a plurality of filter units, each filter unit having one output and two inputs, wherein the two inputs comprise two paths that have complementary transfer functions; and obtaining a composed signal containing the N.sub.c sub-carrier signals from an output of the filter bank, wherein the filter bank composes signals having N.sub.c sub-carrier signals, the filter bank having N.sub.s stages, stage s having 2.sup.s levels, wherein N.sub.s=log.sub.2N.sub.c, s is a stage number, and s[0, 1 . . . N.sub.s1], wherein an n.sup.th order impulse response coefficient of an s.sup.th stage, q.sup.th level filter unit, h.sub.s,q(n), is calculated by multiplying an n.sup.th order impulse response coefficient of s.sup.th stage, p.sup.th level filter unit, h.sub.s,p(n), and a rotation factor, where the rotation factor is a complex exponential factor, and wherein the rotation factor is $e^{j\frac{2}{N_c}\left(\tilde{p}-\tilde{q}\right)},$ wherein {tilde over (p)} is a first value of a first bit reversed version of a first N.sub.s1 bits binary encode of p, and {tilde over (q)} is a second value of a second bit reversed version of a second N.sub.s1 bits binary encode of q.

8. A signal decomposing method comprising: feeding a signal containing N.sub.c sub-carrier signals into a tree structured filter bank having one input and N.sub.c outputs, where the filter bank has a plurality of filter units having one input and two outputs, where the two outputs comprise two paths that have complementary transfer functions; and obtaining the N.sub.c sub-carrier signals from the N.sub.c outputs of the filter bank, respectively, wherein the filter bank decomposes signals having N.sub.c sub-carrier signals, the filter bank having N.sub.s stages, stage s having 2.sup.s levels, wherein N.sub.s=log.sub.2N.sub.c, s is a stage number, and s[0, 1 . . . N.sub.s1], wherein an n.sup.th order impulse response coefficient of an s.sup.th stage, q.sup.th level filter unit, h.sub.s,q(n), is calculated by multiplying an n.sup.th order impulse response coefficient of s.sup.th stage, p.sup.th level filter unit, h.sub.s,p(n), and a rotation factor, wherein the rotation factor is a complex exponential factor, and wherein the rotation factor is $e^{j\frac{2}{N_c}\left(\tilde{p}-\tilde{q}\right)},$ wherein {tilde over (p)} is a first value of a first bit reversed version of a first N.sub.s1 bits binary encode of p, and {tilde over (q)} is a second value of a second bit reversed version of a second N.sub.s1 bits binary encode of q.

9. A communication method, comprising: composing N.sub.c sub-carrier signals using a first tree structured filter bank having N.sub.c channels to obtain a composed signal containing the N.sub.c sub-carrier signals; and decomposing the composed signal using a second tree structured filter bank having N.sub.c channels to obtain the N.sub.c sub-carrier signals, where vector form transfer function {right arrow over (H)}.sub.r,p of channel p of the second filter bank is orthogonal to vector form transfer function {right arrow over (H)}.sub.t,q of channel q of the first filter bank.

10. The communication method of claim 9, wherein: when p=q, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p equals to 1; when |pq|=1, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p is less than a predetermined threshold; and otherwise, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p, equals to 0, where [ ].sup.H is a conjugate transpose operation.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The foregoing and other features of the present disclosure will become more fully apparent from the following description and appended claims, taken in conjunction with the accompanying drawings. Understanding that these drawings depict only several embodiments in accordance with the disclosure and are, therefore, not to be considered limiting of its scope, the disclosure will be described with additional specificity and detail through use of the accompanying drawings.

(2) FIG. 1 illustrates a schematic block diagram of a filter bank for signal decomposition in one embodiment.

(3) FIG. 2 illustrates a schematic block diagram of a filter unit of the filter bank in FIG. 1.

(4) FIG. 3 illustrates a schematic block diagram of a filter bank for signal composition in one embodiment.

(5) FIG. 4 illustrates a spectrum obtained in one experiment using a communication system of one embodiment.

(6) FIG. 5 illustrates an enlarged view of the spectrum in FIG. 4 and a spectrum of a conventional communication system based on FFT/IFFT.

(7) FIG. 6 illustrates a comparison between original signals and decoded signals, according to various embodiments of the present disclosure.

DETAILED DESCRIPTION

(8) In the following detailed description, reference is made to the accompanying drawings, which form a part hereof. In the drawings, similar symbols typically identify similar components, unless context dictates otherwise. The illustrative embodiments described in the detailed description, drawings, and claims are not meant to be limiting. Other embodiments may be utilized, and other changes may be made, without departing from the spirit or scope of the subject matter presented here. It will be readily understood that the aspects of the present disclosure, as generally described herein, and illustrated in the Figures, can be arranged, substituted, combined, and designed in a wide variety of different configurations, all of which are explicitly contemplated and make part of this disclosure.

(9) Referring to FIG. 1, a three stage filter bank 100 for decomposing signals containing eight sub-carrier signals is illustrated. The filter bank 100 includes three stages. The 0.sup.th stage includes one filter unit 101, the 1.sup.st stage includes two filter units 103 and 105, and the 2.sup.nd stage includes four filter units 107, 109, 111, and 113. Each of the filter units includes one input and two outputs which form two paths. The filter bank 100 as a whole includes one input and eight outputs, in other words, the filter bank 100 includes eight channels.

(10) A filter bank for decomposing signals having N.sub.c sub-carrier signals includes N.sub.s=log.sub.2N.sub.c stages, stage s includes 2.sup.s filter units/levels, where s stands for stage number.

(11) Referring to FIG. 2, the s.sup.th stage l.sup.th level filter unit 200 has an input 201 and two outputs 203 and 205, which form an upper path and a lower path. Given the frequency domain transfer function of the upper path is {tilde over (H)}.sub.s,l(z), then the frequency domain transfer function of the lower path shall be A{tilde over (H)}.sub.s,l(z), these two transfer functions are complementary to each other, where A represents a magnitude, z stands for z-transform i.e. z=e.sup.j, where j={square root over (1)}.

(12) Channel number c may be binary encoded, [c].sub.10=[B.sub.N.sub.s.sub.1B.sub.N.sub.s.sub.2 . . . B.sub.0].sub.2, where B.sub.N.sub.s.sub.1 is the most significant bit (MSB), and B.sub.0 is the least significant bit (LSB). For example, referring to FIG. 1, the channel number of channel 4 is four.

(13) Given the frequency domain transfer function of the s.sup.th stage 0.sup.th level filter unit is written as Equation (1),

(14) $\begin{array}{cc}{\tilde{H}}_{s,0}\left(z\right)=h_s\left(0\right)+\underset{n=1}{\overset{M_s-1}{.Math.}}{h}_s\left(n\right)z^{-n}& \mathrm{Equation}\left(1\right)\end{array}$
where M.sub.s1 represents order of transfer functions in s.sup.th stage, and h.sub.s(0), h.sub.s(1) . . . h.sub.s(n) are impulse response coefficients of the transfer function of s.sup.th stage 0.sup.th level filter unit, then the frequency domain transfer function of channel c in s.sup.th stage may be written as Equation (2),

(15) $\begin{array}{cc}{\tilde{H}}_c^s=h_s\left(0\right)+\underset{n=1}{\overset{M_s-1}{.Math.}}{\left(-1\right)}^{B_s}{h}_s\left(n\right)W_{N_c}^{\mathrm{nk}}{z}^{-n.Math.2^{N_s-s-1}}& \mathrm{Equation}\left(2\right)\end{array}$
where B.sub.s stands for the s.sup.th element/bit of the binary encode of the channel number c, N.sub.c stands for the sum of channels in the communication system, N.sub.s stands for the sum of stages in the signal decomposition system, for example, assuming N.sub.c=8, s=2, and c=6, the binary encode of c is 110, then B.sub.s is the 2.sup.nd element of 110 which is 1, where 0.sup.th element of a binary encode e.sub.2e.sub.1e.sub.0 is e0, 1.sup.st element of e.sub.2e.sub.1e.sub.0 is e.sub.1, and 2.sup.nd element of e.sub.2e.sub.1e.sub.0 is e.sub.2,

(16) $W_{N_c}^{\mathrm{nk}}=e^{-j\frac{2}{N_c}\mathrm{nk}},\mathrm{where}$$k=k_0.Math.2^{N_s-s-1},$
where k.sub.0 stands for the value of the least s bits of the binary encode of c. For example, assuming N.sub.s=3, s=2 and c=6, the binary encode of c is 110, the least s=2 bits of the binary encode of c is 10, and k.sub.0=2 in this example. When s=0, k.sub.0=0.

(17) For channel c, when its frequency domain transfer function in each stage is obtained, the channel transfer function {tilde over (H)}z.sub.c in the frequency domain may be written as:
{tilde over (H)}z.sub.c=H.sub.1.Math.{tilde over (H)}.sub.c.sup.0.Math.{tilde over (H)}.sub.c.sup.1. . . .Math.H.sub.c.sup.N.sup.s.sup.1Equation (3),
where H.sub.1 may be defined as:

(18) $\begin{array}{cc}{H}_1=H_0=\frac{1}{\sqrt{{.Math.h_c\left(0\right).Math.}^2+{.Math.h_c\left(1\right).Math.}^2\mathrm{.Math.}{.Math.h_c\left(M_c-1\right).Math.}^2}}& \mathrm{Equation}\left(4\right)\end{array}$
where h.sub.c(n) is a coefficient of transfer function, n[0, 1 . . . M.sub.c1], where M.sub.c1 is order of the transfer function of channel c.

(19) Referring to FIG. 3, a three stage filter bank 300 for composing signals having eight sub-carrier signals is illustrated. A signal composed using the filter bank 300 can be decomposed using the filter bank 100. The filter bank 300 also includes three stages. The 0.sup.th stage includes one filter unit 301, the 1.sup.st stage includes two filter units 303 and 305, and the 2.sup.nd stage includes four filter units 307, 309, 311, and 313. Each of the filter units includes one output and two inputs which form two paths. The filter bank 300 as a whole includes one output and eight inputs, in other words, the filter bank 300 also includes eight channels.

(20) A filter bank for composing N.sub.c sub-carrier signals into one signal containing the N.sub.c sub-carrier signals includes N.sub.s=log.sub.2N.sub.c stages, stage s includes 2.sup.s filter units/levels, and each filter unit includes two inputs which form two paths whose transfer functions are complementary to each other. Its structure is substantially inverse to that of a filter bank for decomposing signals composed by it.

(21) Assuming the frequency domain transfer function of channel c in the filter bank 100 may be written as:
{tilde over (H)}z.sub.c=.Math.[h.sub.c(0)+.sub.n=1.sup.M.sup.c.sup.1h.sub.c(n)z.sup.n]Equation (5),
where may be defined as:

(22) $\begin{array}{cc}=\frac{1}{\sqrt{{.Math.h_c\left(0\right).Math.}^2+{.Math.h_c\left(1\right).Math.}^2\mathrm{.Math.}{.Math.h_c\left(M_c-1\right).Math.}^2}}.& \mathrm{Equation}\left(6\right)\end{array}$

(23) For simplicity, the transfer function of channel c in the filter bank 100 may be re-written in vector form as:
{right arrow over (H)}.sub.r,c=.Math.[h.sub.c(0),h.sub.c(1) . . . h.sub.c(M.sub.c1)].sup.TEquation (7)
where [ ].sup.T stands for transpose operation.

(24) The transfer function of channel c in the filter bank 300 may be re-written in vector form as:
{right arrow over (H)}.sub.t,c={right arrow over (H)}*.sub.r,c=.Math.[h.sub.c(0),h.sub.c(1) . . . h.sub.c(M.sub.c1)].sup.HEquation (8),
where [ ]* stands for conjugate operation, and [ ].sup.H stands for conjugate transpose operation. As a result, the following Equation (9) may be obtained:
{right arrow over (H)}.sup.H.sub.t,c.Math.{right arrow over (H)}.sub.r,c=1Equation (9).

(25) In a signal composition system of a transmitter, if a symbol X.sub.c is fed to a channel c having a transfer function of {right arrow over (H)}.sub.t,c, then a symbol X.sub.c.Math.{right arrow over (H)}.sub.t,c may be generated by the channel c. Since the transmitted symbol X is constituted by symbols generated by all channels, the transmitted symbol X may be written as:
X=X.sub.1.Math.{right arrow over (H)}.sub.t,1+X.sub.2.Math.{right arrow over (H)}.sub.t,2. . . X.sub.N.sub.c.sub.1.Math.{right arrow over (H)}.sub.t,N.sub.c.sub.1Equation (10).

(26) In a signal decomposition system of a receiver, for a received symbol X, a channel c having a transfer function of {right arrow over (H)}.sub.r,c may generate a symbol {tilde over (X)}.sub.c according to Equation (11):
{tilde over (X)}.sub.c=X.sup.T.Math.{right arrow over (H)}.sub.r,cEquation (11).

(27) According to Equations (9) and (10), Equation (12) may be obtained:

(28) $\begin{array}{cc}\begin{array}{c}{\tilde{X}}_c=X^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}\\ ={\left[\begin{array}{c}{X}_1.Math.{\stackrel{.fwdarw.}{H}}_{t,1}+\mathrm{.Math.}+X_c.Math.{\stackrel{.fwdarw.}{H}}_{t,c}+\mathrm{.Math.}\\ X_{N_c-1}.Math.{\stackrel{.fwdarw.}{H}}_{t,N_c-1}\end{array}\right]}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}\\ =X_1.Math.{\stackrel{.fwdarw.}{H}}_{t,1}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}+\mathrm{.Math.}+X_c.Math.{\stackrel{.fwdarw.}{H}}_{t,c}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}+\mathrm{.Math.}\\ X_{N_c-1}.Math.{\stackrel{.fwdarw.}{H}}_{t,N_c-1}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}\\ =X_1.Math.{\stackrel{.fwdarw.}{H}}_{t,1}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}+\mathrm{.Math.}+X_c+\mathrm{.Math.}{X}_{N_c-1}.Math.\\ {\stackrel{.fwdarw.}{H}}_{t,N_c-1}^T.Math.{\stackrel{.fwdarw.}{H}}_{r,c}.\end{array}& \mathrm{Equation}\left(12\right)\end{array}$

(29) Then Equation (13) may be obtained:
{tilde over (X)}.sub.cX.sub.c=X.sub.1.Math.{right arrow over (H)}.sub.t,1.sup.T.Math.{right arrow over (H)}.sub.r,c+ . . . +X.sub.c1.Math.{right arrow over (H)}.sub.t,c1.sup.T.Math.{right arrow over (H)}.sub.r,c+X.sub.c+1.Math.{right arrow over (H)}.sub.t,c+1.sup.T.Math.{right arrow over (H)}.sub.r,c+ . . . +X.sub.N.sub.c.sub.1.Math.{right arrow over (H)}.sub.t,N.sub.c.sub.1.sup.T.Math.{right arrow over (H)}.sub.r,cEquation (13),
where the items on the right of the equation may be called interference items.

(30) To guarantee that {tilde over (X)}.sub.cX.sub.c is equal to zero, vector {right arrow over (H)}.sub.r,p of the receiver shall be orthogonal to vector {right arrow over (H)}.sub.t,q of the transmitter. However, in practice, perfect orthogonality is very difficult to achieve. If nearly orthogonality is achieved, symbols can also be decomposed correctly.

EXAMPLE

(31) A communication system having 64 sub-carriers based on filter banks of the present application was designed, and FIG. 4 illustrates a spectrum of the communication system.

(32) Referring to FIG. 5, an enlarged view of the spectrum of the communication system based on filter banks and a spectrum of conventional FFT/IFFT method is shown. It can be seen that the communication system has the following characteristics: flat-pass band, narrow transition band, small interference between adjacent sub-carriers and large attenuation in the stop-band etc.

(33) Given that {right arrow over (H)}.sub.r,p is the vector of the p.sup.th channel of the receiver and {right arrow over (H)}.sub.t,q is the vector of the q.sup.th channel of the transmitter. In this example, results of multiplication of the two arbitrary vectors are listed below:

(34) ${\left[{\stackrel{.fwdarw.}{H}}_{t,q}\right]}^H.Math.{\stackrel{.fwdarw.}{H}}_{r,p}=\left\{\begin{array}{c}1,p=q\\ 0.0362,.Math.p-q.Math.=1\\ 0,\mathrm{others}\end{array}\right\}.$

(35) Since when p=q, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p is substantially equal to 1; when |pq|=1, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p is less than 0.0362 which is small enough to be negligible; otherwise the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p equals to zero, then {right arrow over (H)}.sub.t,q and {right arrow over (H)}.sub.r,p may be regarded as nearly orthogonal. In this example, 1 and 0.0362 is the result of normalization.

(36) In other words, as long as the above conditions are met, the receiver can decode symbols correctly. To decompose sub-carrier signals correctly, when |pq|=1, the result of [{right arrow over (H)}.sub.t,q].sup.H.Math.{right arrow over (H)}.sub.r,p shall be less than a certain threshold, and the threshold may be determined based on how the signal containing the sub-carrier signals is modulated in the transmitter.

(37) Referring to FIG. 6, differences between original symbols and decoded symbols are shown, where original symbols are represented using symbol o, and decoded symbols are represented using symbol *. It can be seen that the symbols were correctly decoded.

(38) There is little distinction left between hardware and software implementations of aspects of systems; the use of hardware or software is generally a design choice representing cost vs. efficiency tradeoffs. For example, if an implementer determines that speed and accuracy are paramount, the implementer may opt for a mainly hardware and/or firmware vehicle; if flexibility is paramount, the implementer may opt for a mainly software implementation; or, yet again alternatively, the implementer may opt for some combination of hardware, software, and/or firmware.

(39) While various aspects and embodiments have been disclosed herein, other aspects and embodiments will be apparent to those skilled in the art. The various aspects and embodiments disclosed herein are for purposes of illustration and are not intended to be limiting, with the true scope and spirit being indicated by the following claims.