CN104333424B - A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device - Google Patents
A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device Download PDFInfo
- Publication number
- CN104333424B CN104333424B CN201410549873.8A CN201410549873A CN104333424B CN 104333424 B CN104333424 B CN 104333424B CN 201410549873 A CN201410549873 A CN 201410549873A CN 104333424 B CN104333424 B CN 104333424B
- Authority
- CN
- China
- Prior art keywords
- noise variance
- frequency spectrum
- authorized user
- state
- perception
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Abstract
The present invention is directed to the frequency spectrum detection problem under noise variance dynamic unknown situation in practical application, proposes that a kind of dynamical state space system model reflects its inherent mechanism first:Using authorized user's state and time-varying noise variance as two hidden states, two state single order Markovs and autoregression model is respectively adopted its dynamic migration characteristic is modeled.On this basis, design proposes a kind of new frequency spectrum sensing method device.This method is rooted in bayesian statistics inference theory, using marginalisation particle filter technology (see photo), can realize the Combined estimator to authorized user and noise variance.Especially, the present invention proposes marginalisation particle filter two-stage adaptive prediction coefficient updating method, makes full use of noise variance time-varying characteristics, and the accurate tracking for noise variance can be achieved.Using obtained noise variance information, the calculation device, which can be expanded, to be applied into (but not limited to) single node single antenna sensory perceptual system, and obtains good frequency spectrum perception performance.
Description
Technical field
The present invention is directed to frequency spectrum detection problem under the conditions of time-varying unknown noise variance, and design first proposes a kind of dynamical state
Space system model (Dynamic State-space Model, DSM), it is based respectively on two state Markov state probability and turns
Shifting formwork type and autoregression (Auto-regressive, AR) model, authorized user's state is hidden as two with time-varying noise variance
Tibetan state (Hidden States);On this basis, a kind of novel frequency spectrum detecting method is proposed.This method is based on Bayes
Theory, using marginalisation particle filter (Marginal Partical Filtering, MPF) technology, it can realize simultaneously to authorizing
The accurate detection of User Status and the real-time tracking of unknown noise variance, significantly improve the frequency under noise variance dynamic unknown condition
Compose perceptual performance.Belong to the communications field.
Background technology
With Modern wireless communication technology fast development, limited frequency spectrum resource and traditional static spectrum allocation schemes without
Method meets the needs of high transfer rate communication equipment.Cognitive radio (Cognitive Radio, CR) is as a kind of frequency spectrum intelligence
Technology of sharing, the availability of frequency spectrum can be greatly enhanced, turn into one of hot technology currently received significant attention.
As the basic technology of cognitive radio, frequency spectrum perception can be by authorizing frequency range to be periodically detected, finding
Frequency spectrum cavity-pocket, so as to not interfered while ensureing unauthorized user to authorizing frequency range to utilize to authorized user.In recent years
Come, a variety of frequency spectrum perception modes are suggested, and common method includes energy measuring (Energy Detection, ED), matching filter
Ripple detects (Matched Filter Detection, MFD) and cyclostationary characteristic detection (Cyclostationary
Detection).In actual applications, due to the uncertainty and time variation of noise, cognitive user is difficult that acquisition is accurately real-time
Noise variance, so as to cause existing detection method detection performance to be remarkably decreased.
In order to tackle above mentioned problem, the present invention proposes a kind of brand-new frequency spectrum sensing method.This method is based on bayesian theory,
Using marginalisation particle filter technology, in single node single antenna sensory perceptual system, can realize simultaneously to authorized user's dynamic work
Make the accurate detection of state and the real-time tracking of unknown noise variance, obtain good perceptual performance.
The content of the invention
Present invention firstly provides a kind of frequency spectrum detection dynamical state space mould being directed under noise variance time-varying unknown condition
Type.Regard authorized user's working condition and time-varying noise variance as two hiding system modes, the state horse of single order two is respectively adopted
Er Kefu models and AR Model Abstractions become migration characteristic at that time.Meanwhile observation, Ji Jiangjie are obtained using non-coherent reception scheme
Receipts sampled signal energy carries out accumulation summation in the time window of length-specific and obtains observation signal.On this basis, according to side
Edge particle filter technology and conjugate gradient descent method concept, design propose a kind of brand-new frequency spectrum sensing method.Particularly, present invention wound
New property proposes marginalisation particle filter two-stage adaptive prediction coefficient updating method, can be more accurately to time-varying noise
Variance is tracked.New departure greatly improves frequency spectrum perception performance on the premise of perception algorithm requirement of real-time is ensured, from
And provide a kind of scheme of great application potential with realizing for the design of distributed cognition wireless network.
In receiving terminal, including two modules of authorized user's state-detection and noise variance real-time tracking.
Authorized user's state-detection:According to current time receiving terminal observation signal and last moment Noise Variance Estimation
Value, sequence is carried out to the posterior probability of authorized user's working condition using particle filter (Particle Filtering, PF) technology
Renewal is passed through, so as to realize the detection to authorized user's state.
Noise variance real-time tracking:Exported according to current time observation signal and authorized user's state detection module, profit
With conjugate prior concept and marginalisation thought, it is proposed that marginalisation particle filter two-stage adaptive prediction coefficient updating method, it is right
Noise variance realizes real-time tracking.
It is an advantage of the invention that:
1) technical solution of the present invention is applied to the single node frequency spectrum perception under the uncertain wireless transmission environments of noise, is wireless
Solid foundation is established in cognitive radio technology application in mobile communication environment;
2) dynamical state space model that design proposes, HMM and autoregression model is respectively adopted to authorizing
User Status and the dynamic migration characteristic of noise variance are described, and can more effectively reflect frequency under wireless mobile communications environment
Spectrum perceives inherent mechanism;Observation signal is obtained using incoherent reception, reduces the complexity of scheme realization.
3) real-time tracking to noise variance, thus pole can be realized while new departure detects to authorized user's state
The earth improves the frequency spectrum perception performance under the uncertain wireless environment of noise, simultaneously for the real-time tracking of noise variance, lifting
Unauthorized user is advantageously implemented more effectively frequency spectrum access and shared to the cognitive ability of wireless environment;
4) invention makes full use of the priori transition probability information of authorized user's working condition, and skill is filtered using edge particles
Art, the non-stationary non-Gaussian feature for effectively overcoming observation signal (cumlative energy) to show, and avoid conventional particle filtering and exist
The problem of computation complexity occurred during reply higher-dimension detection is too high.
5) invention is directed to noise time-varying migration characteristic, proposes edge particles filtering two-stage adaptive prediction coefficient adjustment side
Method, more accurately time-varying noise variance can be tracked.
6) as incorrect noise increases, joint detection algorithm proposed by the present invention still has excellent robustness, because
And tool is had great advantage in actual applications.
Brief description of the drawings
Fig. 1 is frequency spectrum perception receiving end signal processing unit block diagram.
Fig. 2 is actual noise variance and its estimate comparison diagram
Fig. 3 is that new method frequency spectrum perception detects accuracy and traditional ED performance simulations comparison diagram.
Embodiment
The present invention sets up the frequency spectrum perception dynamical state space model under noise variance dynamic unknown condition, uses simultaneously
Edge particles filtering technique carries out Combined estimator to noise time-varying variance and authorized user's state.Below to dynamic system model and
Frequency spectrum perception process illustrates respectively.
1. shown in the frequency spectrum perception dynamical state space model such as formula (1) (2) (3) that the present invention establishes.
yn=(xn,zn) (3)
In above formula,Authorized user's state at n moment is represented, is shifted according to specific state transition function f ().
xnRepresent authorized user's transmission signal.Two kinds of hypothesis testings in frequency spectrum perception be present, that is, authorize frequency range idle and occupied, respectively
Use H0And H1Represent.When authorization user signal, which is not present, authorizes the frequency range free time, xn=0;When authorization user signal be present,
Authorization user signal energy is normalized, both obtains xn=1.In the present invention, the dynamic migration characteristic of authorized user is abstracted
For the state Markov model of single order two, i.e. authorized user's two states are mutually shifted with certain probability, transition probability matrix
(Transmission Probabilities Matrix, TPM) is as shown in formula (4)
Represent n moment noise variances.In the present invention, the time-varying migration characteristic of noise is abstracted as three rank AR models.I.e.:
enIt is 0 white Gaussian noise signal for average.While it is noted that noise variance time-varying cycle TσMore than sense
Know cycle Ts.For convenience of analysis, further simplify and assume Tσ=LTs, wherein L is the integer more than 1.
ynThe observation signal that cognitive user receives is represented, is the energy of sampled signal in length-specific observation time window
With time window length is set to M, as shown in formula (6).
Wherein, M=Ts×fspRepresent to perceive cycle TsInterior sampling number, fspFor sample frequency.zn=[zn,1,zn,2,…,
zn,M] the one-dimensional noise vector that M independent identically distributed Gauss sample values are formed is represented, interchannel noise is that average is that 0 variance is's
Additive white Gaussian noise (AWGN).Corresponding to authorized user's free time/work two states, observation signal y obeys the free degree respectively
For M center/non-central chi square distribution.
2. being based on above-mentioned dynamical state space model, the present invention further joins to authorized user's state and noise variance
Close estimation.From Bayes's angle, Combined estimator can be obtained by maximizing posterior probability, as shown in formula (7).
Obtaining combined estimation method proposed by the present invention by above formula includes two important components:It is to awarding first based on PF
Power User Status is detected, and on this basis, according to conjugate prior concept and marginalisation thought, noise variance is carried out real-time
Tracking.Described in detail below for above-mentioned two part.
1) authorized user's state sequence estimation based on particle filter
It can be obtained by formula (6), the acquisition of observation operates for nonlinear transformation, and its nonlinear and non-Gaussian characteristic will be undoubtedly to award
The sequential detection band of power User Status carrys out serious challenge.Thought is approached based on Monte Carlo (MC, Monte-Carlo) discrete digital
Particle filter technology, can effectively handle nonlinear and non-Gaussian signal detection problem.Particle filter mainly uses one group of band
There is weight w(i)Discrete particle x(i)To approach the Posterior distrbutionp p (x) of complexity, that is, there is p (x)=∑iw(i)δ(x-x(i)).Herein
On the basis of, n moment authorized users state reality is realized based on maximum a posteriori probability (Maximum a Posteriori, MAP) criterion
When estimate.In the specific implementation, PF, which mainly includes particle, generates and updates two important steps of respective weights.
It is to be directed to a specific distribution sampling process to generate on particle essential, and the distribution is different from Posterior distrbutionp probability, is claimed
For importance function.Optimal importance function is used in this programme, as shown in formula (8).
Particle weights caused by renewal, as shown in formula (9).
According to MAP criterions, authorized user's state is estimated.
2) the iteration renewal of noise variance
For the white Gaussian noise of Unknown Variance known to expectation, it is distributed using inverse gamma and defines its conjugate gradient descent method.Its
Bivariate Bayesian hierarchical approach can be write:
Then noise variance posterior probability equally meets that inverse gamma is distributed:
Distributed constant is updated as follows:
Wherein, Prediction Parameters computational methods are as follows:
λ represents predictive coefficient.Due to noise variance time-varying cycle TσMore than perception cycle Ts, Tσ=LTs.Therefore make an uproar for one
Comprising L perception cycle in sound variance period of change, i.e., per L, perception cycle, noise variance can produce according to AR models in formula (5)
Raw migration.For the system model characteristic, the present invention proposes a kind of novel two-stage adaptive prediction coefficient updating method.The party
Method can be according to the perception cycle in noise variance period position, adjustment predictive coefficient.Specifically, it is located at noise variance when the perception cycle
All moment in the end of term, i.e., when next perception periodic noise variance will produce migration, λ=λ1=0.995;Other perception cycles, λ=
λ2=0.960.
The final more new estimation that can obtain noise variance is calculated as shown in formula (15).
In summary, present invention design cognitive method flow chart is as shown in Figure 1.
Above-mentioned frequency spectrum sensing method is emulated, obtain the contrast of Noise Variance Estimation value and actual value as shown in Fig. 2
Frequency spectrum perception performance curve is as shown in Figure 3.
Actual time-varying noise variance isCorrespondingly, actual signal to noise ratio is SNRn,For unknown variable to be estimated.It is then initial
Actual signal to noise ratio can be denoted as SNR0, corresponding noise variance isOriginal hypothesis signal to noise ratio is denoted asCorrespondingly, noise variance
It is denoted as Snr obeys equally distributed random number for [- ε, ε] within the specific limits.
In Fig. 2, solid line represents noise variance actual value, and dotted line represents its estimate.Abscissa represents detecting period.
In Fig. 3, solid line and dotted line represent the detection performance of associated detecting method and traditional ED algorithms respectively.Abscissa is real
The initial signal to noise ratio snr in border0, ordinate is detection probability Pd.Initial signal to noise ratio floating parameter ε is set to 3,5 and 10.Can by figure
, it is evident that joint detection algorithm proposed by the present invention is significantly improved compared with ED algorithm detection performances.It is also noted that with making an uproar
The uncertain increase of sound, ED detection performances can great degeneration, and joint detection algorithm proposed by the present invention have it is excellent sane
Property, thus in actual applications also by more advantage.
Claims (4)
1. a kind of implementation method of frequency spectrum perception, the high-performance frequency spectrum perception in the case of time-varying unknown noise variance can be achieved;Its
It is characterised by:Based on the dynamical state space model proposed, effectively realize to authorized user's working condition and unknown noise
The Combined estimator of variance;
A kind of system model for deeply reflecting frequency spectrum perception mechanism under noise variance dynamic unknown situation, by authorized user's state and
Noise variance is respectively adopted two state single order Markov models and autoregression model describes its dynamic as system hidden state
Migration characteristic;Using incoherent reception, i.e., using the energy of sampled signal in special time window and as systematic perspective measured value;
Combined estimator is carried out to authorized user's working condition and unknown noise variance using observation signal:It is primarily based on Bayes's system
Meter Framework for Reasoning detects to authorized user's state, based on having obtained frequency spectrum detecting result and observation signal, after variance
Test distributed constant to be updated, then obtain Noise Variance Estimation value.
2. the implementation method of frequency spectrum perception according to claim 1, it is characterised in that:Using particle filter technology, one is utilized
The particle and its respective weights value of the sequential renewal of group approach authorization user signal posterior probability, and according to maximum posteriori criterion
The real-time Sequential Estimation of authorized user's state is obtained, overcomes the non-gaussian nonlinear characteristic of observation signal presentation, improves frequency
Spectrum detection accuracy.
3. the implementation method of frequency spectrum perception according to claim 1, it is characterised in that:Its conjugation is defined using the distribution of inverse gamma
Prior distribution, according to conjugate gradient descent method characteristic, noise variance posterior probability equally meets that inverse gamma is distributed, and then can be based on seeing
The estimate of signal and authorized user's working condition is surveyed, to be updated to the Posterior distrbutionp parameter of unknown noise variance, and
Estimate is updated using its statistical expection as the noise variance at current time.
4. the implementation method of frequency spectrum perception according to claim 3, it is characterised in that:It is proposed that two-stage adaptive prediction coefficient is adjusted
Adjusting method;This method is according to the perception cycle in noise variance period position, adjustment predictive coefficient λ;Specifically, when perception cycle position
In the moment in end of term noise variance week, i.e., when next perception periodic noise variance will produce migration, λ=λ1;Other perceive the cycle,
λ=λ2;Wherein, λ1< λ2。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410549873.8A CN104333424B (en) | 2014-10-16 | 2014-10-16 | A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410549873.8A CN104333424B (en) | 2014-10-16 | 2014-10-16 | A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104333424A CN104333424A (en) | 2015-02-04 |
CN104333424B true CN104333424B (en) | 2018-01-05 |
Family
ID=52408086
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410549873.8A Active CN104333424B (en) | 2014-10-16 | 2014-10-16 | A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104333424B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104867163A (en) * | 2015-05-28 | 2015-08-26 | 深圳大学 | Marginal distribution passing measurement-driven target tracking method and tracking system thereof |
CN108683469B (en) * | 2018-05-08 | 2023-05-23 | 国网江苏省电力有限公司信息通信分公司 | Power wireless private network spectrum noise prediction method and system |
CN110826019B (en) * | 2019-10-15 | 2023-03-14 | 电子科技大学 | Space spectrum state prediction method based on hidden Markov model |
CN111313985B (en) * | 2020-03-05 | 2022-05-13 | 北京振中电子技术有限公司 | Broadband power line carrier communication analog noise generation method and device and electronic equipment |
CN111625923B (en) * | 2020-04-16 | 2024-02-27 | 中国地质大学(武汉) | Antenna electromagnetic optimization method and system based on non-stationary Gaussian process model |
CN112383328B (en) * | 2020-10-13 | 2022-01-18 | 哈尔滨工业大学(深圳) | Improved matched filtering message transmission detection method based on probability cutting in communication system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101227252A (en) * | 2007-12-27 | 2008-07-23 | 复旦大学 | Multi pathway fading channel soft decision metric generating method of unknown noise information |
JP2011185864A (en) * | 2010-03-10 | 2011-09-22 | Alps Electric Co Ltd | Magnetic field detector |
CN103117817A (en) * | 2013-01-09 | 2013-05-22 | 北京邮电大学 | Spectrum sensing method and device under time varying fading channel |
CN103916969A (en) * | 2014-04-17 | 2014-07-09 | 北京邮电大学 | Combined authorized user perception and link state estimation method and device |
CN104168075A (en) * | 2014-08-28 | 2014-11-26 | 北京邮电大学 | Spectrum sensing method and device under condition of unknown noise variance |
-
2014
- 2014-10-16 CN CN201410549873.8A patent/CN104333424B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101227252A (en) * | 2007-12-27 | 2008-07-23 | 复旦大学 | Multi pathway fading channel soft decision metric generating method of unknown noise information |
JP2011185864A (en) * | 2010-03-10 | 2011-09-22 | Alps Electric Co Ltd | Magnetic field detector |
CN103117817A (en) * | 2013-01-09 | 2013-05-22 | 北京邮电大学 | Spectrum sensing method and device under time varying fading channel |
CN103916969A (en) * | 2014-04-17 | 2014-07-09 | 北京邮电大学 | Combined authorized user perception and link state estimation method and device |
CN104168075A (en) * | 2014-08-28 | 2014-11-26 | 北京邮电大学 | Spectrum sensing method and device under condition of unknown noise variance |
Also Published As
Publication number | Publication date |
---|---|
CN104333424A (en) | 2015-02-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104333424B (en) | A kind of frequency spectrum detection and unknown noise variance tracking method of estimation and device | |
Hsieh et al. | Deep learning-based indoor localization using received signal strength and channel state information | |
CN104168075B (en) | Frequency spectrum detecting method and device in the case of a kind of without knowledge of noise covariance | |
Denkovski et al. | Reliability of a radio environment map: Case of spatial interpolation techniques | |
CN104506378B (en) | A kind of device and method of prediction data flow | |
Li et al. | Wi-counter: smartphone-based people counter using crowdsourced wi-fi signal data | |
US20160356665A1 (en) | Pipeline monitoring systems and methods | |
US20160356666A1 (en) | Intelligent leakage detection system for pipelines | |
CN103117817B (en) | A kind of frequency spectrum detecting method under time-varying fading channels | |
CN103401625A (en) | Particle swarm optimization algorithm based cooperative spectrum sensing optimization method | |
Konings et al. | SpringLoc: A device-free localization technique for indoor positioning and tracking using adaptive RSSI spring relaxation | |
Zhou et al. | Target tracking in wireless sensor networks using adaptive measurement quantization | |
CN103916969B (en) | Combined authorization user perceives and Link State method of estimation and device | |
Shen et al. | Chip rate and pseudo‐noise sequence estimation for direct sequence spread spectrum signals | |
Nguyen et al. | Improved localization accuracy using machine learning: Predicting and refining RSS measurements | |
CN105634634B (en) | A kind of asynchronous channel cognitive method there are unknown timing | |
Huang et al. | Dynamic sampling rate adjustment for compressive spectrum sensing over cognitive radio network | |
Hou et al. | Indoor localization for Bluetooth low energy using wavelet and smoothing filter | |
Liao et al. | Distributed edge detection with composite hypothesis test in wireless sensor networks | |
CN108400826A (en) | A kind of frequency spectrum sensing method based on circulant matrix eigenvalue | |
Gufran et al. | FedHIL: Heterogeneity resilient federated learning for robust indoor localization with mobile devices | |
CN116634358A (en) | Terminal positioning method and device and nonvolatile storage medium | |
CN104469782B (en) | A kind of mobile authorization user frequency spectrum detection and its geographical position method of estimation and device | |
Sun et al. | Joint detection scheme for spectrum sensing over time‐variant flat fading channels | |
Gendron | Sparse broadband time varying acoustic response modeling and estimation from an undersampled vertical array with application to underwater communications |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |