CN103812815A - Method for achieving multi-carrier wireless broadband signal modulation based on frequency spectrum inverse transformation - Google Patents

Abstract

The invention provides a method for achieving multi-carrier wireless broadband signal modulation based on frequency spectrum inverse transformation. The method for achieving the multi-carrier wireless broadband signal modulation based on the frequency spectrum inverse transformation comprises the following steps of firstly obtaining a power spectrum sample of a broadband signal according to a broadband signal template or a measuring meter; secondly, extracting and obtaining amplitude spectrum data from power spectrum features of the signal; thirdly, adding randomized phase information into amplitude spectra to obtain amplitude vector spectra; fourthly, performing IFFT (inverse fast fourier transform) on the amplitude vector spectra to obtain a series of random signals of a finite length; fifthly, performing time domain randomization processing on a finite-length time domain random sequence to obtain an aperiodic infinite-length time domain sequence and finally outputting the filtering-processed aperiodic infinite-length time domain sequence to a high-speed DAC (Digital-to-Analog Converter) to construct an expected time domain random signal. The method for achieving the multi-carrier wireless broadband signal modulation based on the frequency spectrum inverse transformation well solves the problem of over-high signal PAPR (Peak-to-Average Power Ratio) caused by a traditional multi-carrier DDS (direct digital synthesis) method and the defects that a signal bandwidth is limited by processing speed of a hardware platform, the number of multi-carriers is limited by hardware platform resources and significance bit width losses of synthesized signals occur during multi-carrier DDS can be eliminated.

Description

A kind of method that realizes the modulation of multi-carrier-wave wireless broadband signal based on frequency spectrum inverse transformation

Technical field

Patent of the present invention belongs to a kind of data modulation technique being applied on cordless communication network, belongs to electronic technology field.

Background technology

From 1897, the transmission that successfully realizes for the first time radio signal starts, wireless communication technology was through the development of more than 200 years, the particularly fast development of nearly 30 years, wide influence is to the mankind's scientific research, the every aspect of productive life, and extend to computer science, aerospace flight technology, automatically control and the research of other each ambits.Along with the development of informationized society, radio communication is broadband, and digitized developing direction becomes clear day by day.

Broadband is one of the important directions of wireless communication technology development.When within 1987, second generation wireless transmitting system GSM comes out, the maximum data transfer rate of support is 9.6Kbps, and during to 3G (Third Generation) Moblie, this speed can reach 2Mbps, even higher.In recent years, high rate communication systems it development is very fast, can see that a large amount of wide-band communication systems has all adopted multi-carrier modulation technology scheme.The basic thought of multi-carrier modulation is that transmission bit stream is divided into multiple sub-bit streams, then is modulated on different subcarriers and transmits.With regard to each carrier wave in multi-carrier modulation, the operation principle of its modulation and the operation principle of single-carrier modulated are basic identical, as QAM, QPSK etc.When wait transmit multi-group data combination of transmitted time, for demodulation that can be correct at receiving terminal, need to carry out quadrature modulation for the different subcarrier of frequency of every group of data allocations, finally all modulation signals are superimposed and be transmitted, so just form multi-carrier modulation.

Digitlization is another main direction of wireless communication technology development.Computer technology, the development of microelectric technique, and large scale integrated circuit, developing rapidly of Digital Signal Processing, has expedited the emergence of this emerging technology of software radio.The basic thought of software radio is that wideband A/D, D/A conversion is positioned as close to radio-frequency antenna, realize data processing by software, carrier modulation etc. in the past need to be by bulky in a large number, the function that the analog electronic equipment of complex structure completes, makes system can work in flexibly different frequency range and different standards.

The formation of a communication system is mainly divided into following three parts: transmitting apparatus, receiving equipment, also has transmission medium.For wireless communication system, this transmission medium is electromagnetic wave.Because electromagnetic wave is subject to the impact of various interference in the communication process in space, disturb in order to reduce these impact that signal is received, simultaneously in order to improve the utilance of wireless channel, need to modulate information at transmitting apparatus end.Traditional transmitter signal handling process is: digital baseband signal is converted to Simulation with I/Q signal after DAC conversion, output to low-pass filtering, then carry out mixing with the orthogonal carrier wave of two-way respectively, then superpose again, then carry out intermediate frequency Modulation and radio frequency mixing, then output to power amplifier, and radio-frequency filter, finally go out by antenna transmission.Utilize software and radio technique, we can realize digitlization intermediate frequency process, i.e. the i/q signal of base band, advanced row digital filtering, then utilizes Direct Digital Frequency Synthesizer Technology (DDS), carries out Digital Modulation respectively with orthogonal two-way digital intermediate frequency carrier signal, after having modulated, output to DAC, be converted to modulated-analog signal, next pass through successively analog intermediate frequency filter, radio-frequency (RF) mixer, radio-frequency (RF) power amplification, output filter, finally arrives antenna, settling signal transmitting.Digitlization MF transmitter has reduced the simulation link in signal processing greatly, and the noise that front end is introduced is less, and circuit is more succinct, and the volume of total system is less, constructs more flexibly, and adaptability is stronger.Therefore, this system is widely used in the every field of radio communication at present, as mobile communication base station, and satellite communication system, radar communications system etc.

In digitlization intermediate frequency WiMAX modulating system, Direct Digital frequency synthesis (DDS) technology is at present comparatively ripe waveform generation technique, this technology can produce good stability, broader bandwidth, the signal that frequency resolution is higher, and conversion speed is fast, modulation system control is flexible.But in multicarrier system, if adopt the DDS synthetic technology of digital structure, the resource of hardware device is proposed to higher requirement, be limited in addition the maximum clock frequency of digital circuit, be difficult to produce very high-frequency signal.On side circuit, the highest output bandwidth only can reach 500MHz at present, and because spurious level and the harmonic wave level of digital circuit are higher, actual working band is well below the highest output bandwidth.So direct digital frequency synthesis technology is not well positioned to meet the application requirements of WiMAX modulating system under a lot of environment.

For some complicated signal form, comprise clutter, noise etc., these signals belong to uncertain signal, show some statistical nature, and these features are often present in power spectrum characteristic.This provides a new approach for we carry out signal reconstruction, because by various supporting instruments, we can more conveniently obtain their statistics power spectrum characteristic.By software and radio technique, combined digital signal treatment theory, on general hardware platform, can calculate the time domain waveform data of signal by fast fourier algorithm, then carry out signal conversion by high-speed DAC device, coordinate necessary filtering processing, just can realize precision high, real-time is good, the signal reconstruction that relative bandwidth is large.

Summary of the invention

For existing methodical deficiency, the present invention, take software and radio technique as basis, is intended to less implementation complexity and hardware platform complexity, realize high accuracy, the broadband signal reconstruct of high real-time.

Technical scheme of the present invention is as follows:

A method that realizes the modulation of multi-carrier-wave wireless broadband signal based on frequency spectrum inverse transformation, described method comprises following steps:

The 1st step: the amplitude spectrum of calculating signal

$\left|X\left(\mathrm{n\Δf}\right)\right|=\sqrt{(S\left(\mathrm{n\Δf}\right)\mathrm{\Δf}}$

Wherein S (n Δ f), n=0, ± 1, ± 2 ..., be the discrete power spectral function of signal, Δ f is the spectral resolution of discrete power spectrum.

The 2nd step: by the amplitude spectrum of random signal | and X (n Δ f) | be superimposed with and there is equally distributed random phase spectrum φ (n Δ f), obtains the vector frequency spectrum that comprises phase information

$X\left(\mathrm{n\Δf}\right)=\left|X\left(\mathrm{n\Δf}\right)\right|\·e^{\mathrm{j\φ}\left(\mathrm{n\Δf}\right)}$

The 3rd step: calculate inverse fourier transform, obtain sequence

$x\left(\mathrm{n\Δt}\right)=\frac{1}{N}\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}X\left(\mathrm{k\Δf}\right)\exp (j2\mathrm{\πkn}/N)$

The 4th step: with random tap j retracing sequence, the sequence after rearrangement is

x _i[0]＝x[j]，x _i[1]＝x[j+1]，…，x _i[N-j-1]＝x[N-1]，

x _i[N-j]＝x[0]，x _i[N-j+1]＝x[1]，…，x _i[N-1]＝x[j-1]

The 5th step: retracing sequence is weighted to stack with semisinusoidal window

$x_d\left(m\right)=\underset{i=0}{\overset{k}{\mathrm{\Σ}}}w(\mathrm{iH}+m)x_i(\mathrm{iH}+m),m=\mathrm{0,1,2},\·\·\·$

Wherein, window function is

h=N/2.

The 6th step: constantly repeat above step, obtain time-domain-random-sequence.

The 7th step: carry out digital-to-analogue conversion, generate the time-domain signal of expecting.

The beneficial effect that the present invention produces mainly contains following 4 points:

1. can overcome traditional multi-carrier modulation and cause the too high problem of peak-to-average power ratio (PAPR): peak-to-average power ratio is the peak power of transmitter output and the ratio of average power.Concerning single carrier modulation system, PAPR value is mainly determined by the rolloff-factor of shaping filter.And for the modulating system of multicarrier, because signal is synthetic by a series of separate modulated carriers, according to central-limit theorem, in the time that number of signals N is larger, there is larger peak value in the result of multiple subcarriers stacks sometimes.In the time that peak-to-average power ratio is very large, will have a strong impact on the efficiency of rear class power amplifier, increase the difficulty of power amplifier design.

2. can overcome under traditional multi-carrier modulation, the resampling difficulty causing due to base band data speed difference: in order to realize the up-conversion in broadband, in system, need multiple separate, the inconsistent baseband signal of bandwidth is carried out resampling, unified on identical spreading rate for up-conversion is prepared.Under normal circumstances, the speed of resampling is defined as the common multiple of the baseband signal bandwidth of need modulation, but when there is following situation, for example, 1 baseband signal is GSM signal, signal rate is 270.883KHz, another baseband signal is TD-SCDMA signal, and signal rate is 1.28MHz, only these 2 baseband signals, want on FPGA hardware circuit, to carry out resampling, it is more difficult implementing.

3. can overcome under traditional multi-carrier modulation, in the time that carrier number is larger, numbers pile up causes the too low problem of effective bit wide of composite signal after processing: for the digital modulation system of multicarrier, digitized effective bit wide of supposing every road signal is 14bit, in the time that the multiply digital signals to after carrier modulation carries out wide-band stack, effectively bit wide must increase, in order to adapt to the conversion of DAC, before DAC, also need to do suitable intercepting outputing to, this will inevitably impact signal, particularly, in the time that number of signals is larger, this impact is also just larger.

4. can overcome under traditional multi-carrier modulation, for realizing DDS, the problem that the implementation complexity causing is too high, in multicarrier system, need to provide reference frequency for each way carrier wave, need to consume a large amount of resources of chip and realize digital oscillator.In the time that carrier number increases, resource consumption is linear increasing also, and this has proposed very high requirement to hardware handles platform.After carrier number acquires a certain degree, implement very difficult.

Accompanying drawing explanation

Fig. 1 is the multi-carrier-wave wireless broadband signal modulator approach flow chart illustrating based on frequency spectrum inverse transformation.

Embodiment

For above-mentioned purpose of the present invention, feature and advantage can be become apparent more, below in conjunction with the drawings and specific embodiments, the application is described in further detail.

With reference to accompanying drawing 1, show the signal modulating method realization flow schematic diagram of the present invention in multi-carrier broadband wireless communication system.Comprise:

Step 1: obtain random signal discrete power spectral function, calculate signal discrete amplitude spectrum.

The power spectrum function of random signal, we can pass through power spectrum template, or various supplementary instrument instrument is easier obtained.

If the discrete power spectral density of given random signal be expressed as S (n Δ f), n=0, ± 1, ± 2 ..., the spectral resolution that wherein Δ f is discrete spectrum, to its extraction of square root, can obtain amplitude spectrum

$\left|X\left(\mathrm{n\Δf}\right)\right|=\sqrt{\left(S\right(\mathrm{n\Δf})\mathrm{\Δf}}$

Step 2: be the amplitude spectrum equally distributed phase spectrum that superposes.

According to signal reconstruction uniqueness principle, only there is the amplitude spectrum function cannot unique reconstruct original signal, need to fill into corresponding phase information for it.When the value at the phase angle filling into has randomness and meets [0,2 π] on be uniformly distributed or when Gaussian Profile, according to the law of large numbers, the I of the time domain data that obtains through inverse discrete fourier transform { x (n Δ t) }, Q two paths of signals all meets Gaussian Profile, its mould value | x (n Δ t) | } Rayleigh distributed.Like this, by the amplitude spectrum of random signal | and X (n Δ f) | be superimposed with and there is equally distributed random phase spectrum φ (n Δ f), just can obtain the vector frequency spectrum that comprises phase information

$X\left(\mathrm{n\Δf}\right)=\left|X\left(\mathrm{n\Δf}\right)\right|\·e^{\mathrm{j\φ}\left(\mathrm{n\Δf}\right)}$

Step 3: carry out inverse discrete fourier transform, obtain time domain sequences.

According to the requirement of Fourier inversion, the amplitude of the vector sequence zero padding in above formula, to N value, is then done to inverse discrete fourier transform to this sequence, obtain time-domain signal sequence

$x\left(\mathrm{n\Δt}\right)=\frac{1}{N}\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}X\left(\mathrm{k\Δf}\right)\exp (j2\mathrm{\πkn}/N)$

Step 4: time domain sequences is reset.

The time-domain signal x (n) that IFFT by above formula generates is actually a pseudo-random signal, and on the one hand, it meets the feature of periodic signal; On the other hand, in one-period T, its amplitude distribution presents the characteristics of amplitude distribution of random signal.Random signal in Practical Project is aperiodic true random signal mostly, in order to overcome the deficiency of pseudo-random signal, must carry out time-domain randomization processing to pseudo-random signal, makes the actual signal that approaches of echo signal maximum possible.Carrying out time-domain randomization processing, will be first that the pseudo random sequence x (n) that N is ordered carries out random tap to length, and reset successively.

The x obtaining in step 3 (n) is called to auxiliary sequence, by x _i(n) be called i retracing sequence, first, a random element first element as retracing sequence of choosing in auxiliary sequence, establishes the result of randomly drawing and is

x _i[0]＝x[j]，i＝0，1，…

In formula, j ∈ [0.N-1].Retracing sequence x _iother element in [n] and auxiliary sequence x[n] in the corresponding relation of element be respectively

x _i[0]＝x[j]，x _i[1]＝x[j+1]，…，x _i[N-j-1]＝x[N-1]，

x _i[N-j]＝x[0]，x _i[N-j+1]＝x[1]，…，x _i[N-1]＝x[j-1]

In above formula, how the value of random tap sequence number j is calculated so, first establishes p[j] be known and obedience (0,1) equally distributed random number sequence, the computing formula of so random tap sequence number j is

In formula,

for rounding symbol downwards.

Step 5, is weighted stack to retracing sequence with semisinusoidal window.

Obviously sequence x _i[n] meets the requirement of amplitude spectrum distribution character within the cycle, and still, because ending and the beginning of next group data of each group signal data often have " kick " phenomenon, the every frame sequence of simple combination will inevitably make signal discontinuous, thereby causes spectrum leakage large.Adopt the method for windowing stack can effectively avoid this problem, windowing is actually carries out smoothing processing to the boundary point of sequence with window function, so that can guarantee the continuity of signal when sequence is carried out to periodic extension at boundary point.After guaranteeing, one group of number initial value and last group of number end value equate (being generally 0), and the object of stack is that the mean-square value in order to guarantee arbitrary moment signal is constant.If burst is { x _i(n) } ,=1,2,3 ..., window function is expressed as w (n), and H represents the interval that superposes, the frame number that i is retracing sequence, k represents the number of retracing sequence, the later signal indication that superposes is:

$x_d\left(m\right)=\underset{i=0}{\overset{k}{\mathrm{\Σ}}}w(\mathrm{iH}+m)x_i(\mathrm{iH}+m),m=\mathrm{0,1,2},\·\·\·$

Obviously constant in order to meet the amplitude characteristic of signal, require window function to meet following equation:

$\underset{i=0}{\overset{k}{\mathrm{\Σ}}}w(\mathrm{iH}+m)=1$

In order to meet this requirement, in actual design, get semisinusoidal window,

$w\left(t\right)=\sin \mathrm{\π}\frac{t}{T},0\≤t\≤T$

Discretization is expressed as

$w\left(n\right)=\sin \mathrm{\π}\frac{n}{N},0\≤n\≤N-1$

Here the desirable H=N/2 in interval that superposes, i.e. 50% stack, needs wrong 2 sequences after resetting half period to superpose.

Step 6, constantly repeats above-mentioned steps, if the operation aperiodicity that sequence is carried out to random tap, the so last sequence x generating _d(m) be exactly true random sequence, and, x _d(m) will keep characteristics of amplitude distribution and the auto-power spectrum shape feature of x (n), the random sequence finally obtaining is

$x_d\left(m\right)=\underset{i=0}{\overset{k}{\mathrm{\Σ}}}w(i\frac{N}{2}+m)x_i(i\frac{N}{2}+m),m=\mathrm{0,1},2,\·\·\·$

Step 7, outputs to digital to analog converter conversion by sequence, and adds suitable filtering and process and just can obtain time-domain signal corresponding to former power spectrum signal.

In theory as long as the operation aperiodicity that sequence is carried out to random tap, the so last sequence generating is exactly true random sequence, and in actual implementation procedure, random tap operations can not be random, but the sequence of random tap value is approximately long, the true random sequence x of generation _d(m) longer.Use T _mrepresent that random sequence continues duration, F _mrepresent the speed that sequence is carried out digital-to-analogue conversion, N _jrepresent the Cycle Length of random tap sequence, obviously have following relation to set up:

$T_m=\frac{N\·N_j}{2F_m}$

Because in the time obtaining the power spectrum sample of random signal, will upgrade once at interval of regular hour sample, as long as therefore guarantee that random tap j no periodic just can meet formation sequence x within the Sample Refreshment cycle _d(m) time domain randomness.

Claims (5)

1. a method that realizes the modulation of multi-carrier-wave wireless broadband signal based on frequency spectrum inverse transformation, by input power spectrum signal is carried out to data processing, obtains endless time domain random signal, and the method includes the steps of:

(1) the power spectrum sqrt of signal is obtained to amplitude spectrum data;

(2), for amplitude spectrum is superimposed with equally distributed random phase spectrum, obtain the amplitude vector frequency spectrum that comprises phase information;

(3) amplitude of the vector spectrum is carried out to inverse fast fourier transform (IFFT), obtain a series of time-limited time-domain-random-sequences;

(4) time domain sequences is reset;

(5) with semisinusoidal window, retracing sequence is weighted to stack

(6) repeating above step obtains the unlimited random sequence of time domain and enters

(7) unlimited time domain random sequence is carried out to digital to analog conversion and obtain time domain random signal.

2. the method for claim 1, is characterized in that the phase spectrum of stack meets being uniformly distributed on [0,2 π].

3. the method for claim 1, is characterized in that amplitude of the vector spectrum to carry out inverse fast fourier transform.

4. the method for claim 1, is characterized in that time domain pseudo random sequence carries out random tap rearrangement.

5. the method as described in claim 1 and 4, is characterized in that limited long random sequence to carry out half cosine window function weighted superposition.

Images