CN105915299A - Time-frequency two-dimensional LMBP neural network based frequency spectrum prediction method in ISM frequency range - Google Patents
Time-frequency two-dimensional LMBP neural network based frequency spectrum prediction method in ISM frequency range Download PDFInfo
- Publication number
- CN105915299A CN105915299A CN201610149087.8A CN201610149087A CN105915299A CN 105915299 A CN105915299 A CN 105915299A CN 201610149087 A CN201610149087 A CN 201610149087A CN 105915299 A CN105915299 A CN 105915299A
- Authority
- CN
- China
- Prior art keywords
- frequency
- time
- dimensional
- lmbp
- csi
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/373—Predicting channel quality or other radio frequency [RF] parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
- H04B17/3913—Predictive models, e.g. based on neural network models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a time-frequency two-dimensional LMBP neural network based frequency spectrum prediction method in an ISM frequency range. The method comprises the following steps: step one, calculation of ISM frequency range correlation, i.e., obtaining correlation between a time domain and a frequency domain of the ISM frequency range through actual measurement, quantification and correlation analysis of the ISM frequency range; step two, based on the time-frequency correlation of the ISM frequency range, constructing a time-frequency two-dimensional LMBP neural network for realizing prediction of the ISM frequency range; step three, by taking actually measured data as a time-frequency input vector and an object vector of the neural network, realizing iteration training of the time-frequency two-dimensional LMBP neural network by use of Newton's method learning rule, and obtaining an optimal solution of a parameter vector u formed by a neural network internodal weight w and a threshold b; and step four, realizing frequency spectrum prediction of the ISM frequency range through the trained two-dimensional LMBP neural network. According to the invention, based on the calculation of the time-frequency correlation of the ISM frequency range, the frequency spectrum prediction of the ISM frequency range is realized through the time-frequency two-dimensional LMBP neural network. The method, compared to a time-domain LMBP neural network and a Markov algorithm, has the advantages of high prediction precision and short training convergence time.
Description
Technical field
The invention belongs to spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in cognitive radio frequency spectrum prediction field, specifically ISM band.
Background technology
The fast development of Internet of Things industry causes the internet of things equipment being operated in ISM (2.4GHz) frequency range day by day to increase, the problem of co-channel interference of this frequency range is on the rise, it is unauthorized frequency range in view of ISM band, small-power network (such as ZigBee) is easily fallen into oblivion by the interference of similar WLAN this kind of high power equipment, cause internodal data frame losing, the paralysis of the most whole network.Know that the occupied information of frequency range is a kind of effective way solving ISM band equipment room compatibility coexistence problems in advance by spectrum prediction algorithm.
nullFrom Kumar Acharya PA,Singh S,Document " Reliable Open Spectrum Communications through Proactive Spectrum " (the PROCEEDINGS OF FIRST INTERNATIONAL WORKSHOP ON TECHNOLOGY AND POLICY FOR ACCESSING SPECTRUM of Zheng H,AUGUST,2006) propose first to be applied to forecasting mechanism to deduce time-domain spectral hole go out now and persistent period after,All kinds of prediction algorithms are widely used in spectrum prediction.
nullPrior art discloses Sixing Yin,Dawei Chen,Qian Zhang,Document " Mining spectrum usage data:a large-scale spectrum measurement study " (the IEEE TRANSACTIONS ON MOBILE COMPUTING of Mingyan Liu and Shufang Li,VOL.11,NO.6,JUNE 2012),This article proposes a kind of detailed frequency spectrum data measuring method,Channel idle data to 20MHz 3GHz frequency range、The channel utilization of every kind of independent wireless service and the time-frequency dependency of channel have carried out statistical analysis,And propose 2D frequent pattern mining algorithm based on frequency spectrum dependency with this and realize the prediction of channel availability,But this algorithm take from frequency spectrum rule is excavated prediction rule minimum training time a length of 2 hours,Otherwise it is difficult to ensure that precision of prediction and the loss of algorithm,This algorithm does not meets, for the restriction of minimum training duration, the feature that ISM band channel time variation is bigger.
nullPrior art discloses Chen Binhua document " the spectrum prediction algorithm research in cognitive radio system " (master thesis. Beijing University of Post & Telecommunication,2011),This article time domain related features based on GSM frequency range,It is respectively adopted single order Markov chain and time domain LMBP neutral net to realize GSM900、The spectrum prediction of GSM1800 frequency range,And by setting up the spectral model of a multichannel association,Realize the prediction of multi channel joint spectrum,But the method only accounts for the GSM frequency range time slot dependency usable value for spectrum prediction,The frequency domain correlation properties of frequency range are not verified,And use single order Markov chain and time domain LMBP neutral net to realize the spectrum prediction of ISM band,The training convergence time of algorithm and precision of prediction are difficult to reach ideal equilibrium.
From correlational study, the basic skills of spectrum prediction algorithm is frequency spectrum measured data based on concrete frequency range, knows the frequency spectrum correlation properties of concrete frequency range, takies from frequency spectrum and excavates feasible prediction rule rule, proposes the improvement to pre existing method of determining and calculating with this.The present invention obtains the time-frequency correlation properties of ISM band based on measured data, proposes a kind of time-frequency two-dimensional LMBP neutral net, is realized the spectrum prediction of ISM band by the parallel training of time-frequency input vector.
Summary of the invention
For above existing deficiency, it is proposed that a kind of method.Technical scheme is as follows: a kind of spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band, and it includes step:
Step 1: collect ISM band measured data, calculates the time-frequency dependency between ISM band time domain and frequency domain;
Step 2: time-frequency dependency based on ISM band, builds two-dimensional prediction matrix, realizes the common training of time domain and frequency domain in a parallel fashion, obtain ground floor network output vector Y in ground floor network1t,Y2t, by ground floor network output vector Y1t,Y2tAs the input vector of second layer network, obtain final predictive value Yt, i.e. build the spectrum prediction of time-frequency two-dimensional LMBP neural fusion ISM band;
Step 3: utilize the measured data of the ISM band obtained in step 1 as training sequence, by adjusting formula, with error function as condition, complete the repetitive exercise of time-frequency two-dimensional LMBP neutral net, obtain the optimal solution of parameter vector u, obtain the weight vector w and threshold vector b of neutral net with this;
Step 4: build time domain and frequency domain input vector matrix Xt, Xf: by Xt, XfThe time-frequency two-dimensional LMBP neutral net built by step 2, obtains output vector Y=[Y1,Y2,Y3,...,Ym]T, YmIt is CSI (tm,cm)
Predictive value, complete spectrum prediction.
Further, described step 1 and time domain X of step 21=[CSI (t-1, c), CSI (t-2, c) ..., CSI (t-Δ t, c)]TWith frequency domain X2=[CSI (t, c ± 1), CSI (t, c ± 2) ..., CSI (t, c ± Δ f)]T, (t-Δ t c) represents that the moment, (channel condition information of t-Δ t) channel c, (t, c ± Δ f) represented the moment t channel (channel condition information of c ± Δ f) to CSI to shown CSI.
Further, the adjustment formula of described step 3 is:
Adjust formulauk+1Represent the parameter vector that next step iteration obtains, ukRepresent the parameter vector of current iteration, J (uk) represent e (uk) Jacobian matrix,Represent Tiny increment dt unit matrix, e (uk) represent error vector, adjust formula with error function F (u)≤ε as condition, ε represents default target error.
Further, in step 1, collecting after ISM band frequency spectrum measured data, the dependency R of this frequency range time domain (Δ t), the dependency R of frequency domain (Δ f) is calculated as follows:
Wherein, CSI (t, quantitative formula c) is:
(t c) represents measured power value and the channel condition information of moment t lower channel c respectively with CSI;
The degree of association of two 0-1 sequencesIt is defined as follows:
This formula is usually used in assessing the dependency of two binary sequences, wherein I (A) is discriminant function, if A value is true, then I (A)=1, otherwise I (A)=0, by R, ((correlation curve of Δ f) obtains ISM band adjacent to the dependency between time slot and neighbouring frequency for Δ t), R.
Further, in step 2, frequency domain prediction point is added in neutral net input vector, build two-dimensional prediction matrix, ground floor network realizes time domain X in a parallel fashion1=[CSI (t-1, c), CSI (t-2, c) ..., CSI (t-Δ t, c)]TWith frequency domain X2=[CSI (t, c ± 1), CSI (t, c ± 2) ..., CSI (t, c ± Δ f)]TCommon training, obtain output vector Y1t,Y2t, by Y1t,Y2tAs the input vector of second layer network, obtain final predictive value Yt, the method combined with time domain and frequency domain builds time-frequency two-dimensional LMBP neutral net and realizes the spectrum prediction of ISM band.
Further, step 3 particularly as follows:, using the measured data of ISM band as training sequence input time-frequency two-dimensional LMBP neutral net in, obtain output vector Y=[Y1,Y2,Y3,...,Ym]T, and with object vector constitute error function:
Wherein, the parameter vector that u is made up of neural network weight w and threshold values b:
The Newton method learning rules of parameter vector u:WhereinFor Hessian inverse of a matrix matrix;gkFor the gradient of F (u),J (u) is the Jacobian matrix of e (u):
Further, in step 4 particularly as follows: build time domain and frequency domain input vector matrix X with ISM band measured datat, Xf:
By Xt, XfAs neutral net input vector, the time-frequency two-dimensional LMBP neutral net having reached optimal value by weight vector w and threshold vector b, obtain output vector Y=[Y1,Y2,Y3,...,Ym]T, YmIt is CSI (tm,cm) predictive value.
Advantages of the present invention and having the beneficial effect that:
Present invention employs time-frequency correlation properties based on ISM band, build a kind of time-frequency two-dimensional LMBP neutral net, frequency domain parameter is added in input vector, build two-dimensional prediction matrix, based on adjacent frequency point (Δ f=1, Δ f=2) high correlation, part time domain input vector is replaced with the frequency domain input vector that dependency is higher, on the basis of ensureing precision of prediction, reduce the computation complexity of neutral net repetitive exercise, and then shorten the training convergence time of network, it is short that ISM band spectrum prediction method the most proposed by the invention has training convergence time, the advantage that precision of prediction is high.
Accompanying drawing explanation
Fig. 1 is spectrum prediction method flow diagram based on time-frequency two-dimensional LMBP neutral net in ISM (2.4GHz) frequency range that the present invention provides preferred embodiment to propose;
Fig. 2 be the present invention obtain according to measured data 2.4GHz frequency range neighbour time/correlogram frequently;
Fig. 3 is present invention two dimension LMBP Prediction Accuracy curve chart when different time-frequency input vectors;
Fig. 4 is the present invention precision of prediction curve comparison figure with time domain LMBP neutral net and Markov algorithm;
Table 1 is the present invention contrast (N=9) with the training convergence in mean time of time domain LMBP neutral net and Markov algorithm.
Detailed description of the invention
Below in conjunction with accompanying drawing, the invention will be further described:
As shown in Figure 1, the present embodiment is the training program of time-frequency two-dimensional LMBP neutral net, input vector Data Source is in the actual measurement quantized data of ISM band, neutral net time domain input Δ t is respectively 1,2,3, ..., 10, frequency domain input Δ f is respectively 0,1,2, output vector N=16, the target error of error function F (u) is 0.01, reach, using F (u), the condition that target error completes as network training, obtain the optimal solution of the parameter vector u being made up of neural network weight and threshold value.
The first step: calculating ISM band time domain and frequency domain correlation:
Wherein, CSI (t, quantitative formula c) is:
(t c) represents measured power value and the channel condition information of moment t lower channel c respectively with CSI.
The degree of association of two 0-1 sequencesIt is defined as follows:
This formula is usually used in assessing the dependency of two binary sequences, and wherein I (A) is discriminant function, if A value is true, then and I (A)=1, otherwise I (A)=0.By R, ((correlation curve of Δ f) understands, and ISM frequency spectrum has height correlation characteristic adjacent to time slot for Δ t), R;Neighbouring frequency has stronger correlation properties, and in Δ f=± 1 and Δ f=± 2 adjacent to frequency, dependency is respectively 0.87 and 0.86.
Second step: based on ISM band time-frequency dependency conclusion, builds time-frequency two-dimensional LMBP neutral net and realizes the spectrum prediction of ISM band;
3rd step: the measured data of ISM band is obtained output vector Y=[Y as training sequence input time-frequency two-dimensional LMBP neutral net1,Y2,Y3,...,Ym]T, and with object vector constitute error function:
Wherein, the neural network parameter vector that u is made up of weights and threshold values:
The Newton method training rules of parameter vector u:WhereinFor Hessian inverse of a matrix matrix, gkGradient for F (u).Wherein, J (u) is the Jacobian matrix of e (u):
Owing to F (u) has the form of error of sum square, HkCan approximate expression be:For ensureingReversible for positive definite, add a Tiny increment dtThe adjustment formula thus obtaining u is:Using the measured data of ISM band as training sequence, as F (u) > ε, with adjust formula be iterated training;As F (u)≤ε, terminate training, obtain the optimal solution of parameter vector u, namely obtain the weight vector w and threshold vector b of neutral net.
4th step: the time-frequency two-dimensional LMBP neutral net constituted with parameter vector u, to realize the spectrum prediction of ISM band, builds time domain and frequency domain input vector matrix X with ISM band measured datat, Xf:
By Xt, XfAs neutral net input vector, the time-frequency two-dimensional LMBP neutral net having reached optimal value by weight vector w and threshold vector b, obtain output vector Y=[Y1,Y2,Y3,...,Ym]T, YmIt is CSI (tm,cm) predictive value.By output vector Y with actual measurement object vector Z=[Z1,Z2,Z3,...,Zm]TContrast, it is thus achieved that the precision of prediction of time-frequency two-dimensional LMBP neutral net.
In the present embodiment, Fig. 2 gives based on the calculated time domain of ISM band measured data, frequency domain correlation properties curve chart;Fig. 2 give based on different LMBP neutral net time-frequency input vector combinations (Δ t=1,2,3 ..., 10, Δ f=0,1,2) the precision of prediction figure that obtains;Fig. 3 give based on no input vector number (N=1,2,3 ..., 14), time-frequency two-dimensional LMBP neutral net (Δ f=2) is with tradition LMBP neutral net and the precision of prediction comparison diagram of Markov algorithm.From Figure 2 it can be seen that ISM frequency spectrum time domain has height correlation characteristic in short-term, tend to 0.85 along with Δ t increases relativity of time domain;Frequency domain is respectively 0.87 and 0.86 at Δ f=± 1 and the frequency of Δ f=± 2, dependency, has higher spectrum prediction dependency.As seen from Figure 3, the precision of prediction of time domain LMBP neutral net was incremented by before Δ t=5, was basically stable at less than 0.92 after Δ t=5;In the time-frequency two-dimensional LMBP neutral net being made up of Δ f=1, Δ f=2, Δ f=3, as Δ t < 5, it was predicted that precision relatively time domain LMBP improves a lot;When after Δ t > 5, it was predicted that precision tends towards stability, wherein the precision of prediction of Δ f=2 is optimum, stable about 95%.As seen from Figure 3, the precision of prediction of time domain LMBP neutral net and Markov algorithm is stable 86% and about 91% after N=5, i.e. can not improve precision of prediction by increasing input vector;And time-frequency two-dimensional LMBP neutral net (Δ f=2) compares first two method under conditions of increasing by 4 input vectors on year-on-year basis, its precision of prediction is compared time domain LMBP algorithm and is improved about 4%, compare Markov algorithm and improve about 9%, and precision tends towards stability after N=9 equally.From table 1, under conditions of target error is identical, the network training convergence time required for time-frequency two-dimensional LMBP neutral net is shorter, with error target 10-2As a example by, the training convergence time of time-frequency two-dimensional LMBP neutral net fast 1.9 times than time domain LMBP, fast 3.8 times than Markov algorithm.Understanding institute's extracting method in conjunction with Fig. 1, Fig. 2, Fig. 3, to compare time domain LMBP Forecasting Methodology and Markov Forecasting Methodology precision of prediction under the conditions of same input vector more excellent, training convergence time is shorter, time-frequency two-dimensional LMBP neutral net is using Δ t=5 and Δ f=2 as input vector, the spectrum prediction precision of 95% can be realized under conditions of input vector number N=9, under conditions of not increasing computation complexity, it is effectively improved the precision of prediction of ISM band.Institute's extracting method can know the occupied information of frequency range in advance by spectrum prediction, effectively solves ISM band equipment room compatibility coexistence problems, it is thus achieved that spectrum prediction method more excellent in ISM band.
Table 1
Above the specific embodiment of the present invention is described.It is to be appreciated that the invention is not limited in above-mentioned particular implementation, those skilled in the art can make various deformation or amendment within the scope of the claims, and this has no effect on the flesh and blood of the present invention.
Claims (7)
1. spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in an ISM band, it is characterised in that include step:
Step 1: collect ISM band measured data, calculates the time-frequency dependency between ISM band time domain and frequency domain;
Step 2: time-frequency dependency based on ISM band, builds two-dimensional prediction matrix, realizes the common training of time domain and frequency domain in a parallel fashion, obtain output vector Y of ground floor network in ground floor network1t,Y2t, by output vector Y of ground floor network1t,Y2tAs the input vector of second layer network, obtain final predictive value Yt, i.e. build the spectrum prediction of time-frequency two-dimensional LMBP neural fusion ISM band;
Step 3: utilize the measured data of the ISM band obtained in step 1 as training sequence, by adjusting formula, with error function as condition, complete the repetitive exercise of time-frequency two-dimensional LMBP neutral net, obtain the optimal solution of parameter vector u, obtain the weight vector w and threshold vector b of neutral net with this;
Step 4: build time domain and frequency domain input vector matrix Xt,Xf: by Xt,XfThe time-frequency two-dimensional LMBP neutral net built by step 2, obtains output vector Y=[Y1,Y2,Y3,...,Ym]T, YmIt is CSI (tm,cm) predictive value, complete spectrum prediction.
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 1, it is characterised in that described step 1 and time domain X of step 21=[CSI (t-1, c), CSI (t-2, c) ..., CSI (t-Δ t, c)]TWith frequency domain X2=[CSI (t, c ± 1), CSI (t, c ± 2) ..., CSI (t, c ± Δ f)]T, (t-Δ t c) represents that the moment, (channel condition information of t-Δ t) channel c, (t, c ± Δ f) represented the moment t channel (channel condition information of c ± Δ f) to CSI to shown CSI.
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 1 and 2, it is characterised in that the adjustment formula of described step 3 is:
Adjust formulauk+1Represent the parameter vector that next step iteration obtains, ukRepresent the parameter vector of current iteration, J (uk) represent e (uk) Jacobian matrix,Represent Tiny increment dt unit matrix, e (uk) represent error vector, adjust formula with error function F (u)≤ε as condition, ε represents default target error.
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 3, it is characterised in that
In step 1, collecting after ISM band frequency spectrum measured data, the dependency R of this frequency range time domain (Δ t), the dependency R of frequency domain (Δ f) is calculated as follows:
Wherein, CSI (t, quantitative formula c) is:
(t, c) (t c) represents measured power value and the channel condition information of moment t lower channel c to R respectively with CSI;
The degree of association of two 0-1 sequencesIt is defined as follows:
This formula is usually used in assessing the dependency of two binary sequences, wherein I (A) is discriminant function, if A value is true, then I (A)=1, otherwise I (A)=0, by R, ((correlation curve of Δ f) obtains ISM band adjacent to the dependency between time slot and neighbouring frequency for Δ t), R.
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 4, it is characterized in that, in step 2, frequency domain prediction point is added in neutral net input vector, build two-dimensional prediction matrix, ground floor network realizes time domain X in a parallel fashion1=[CSI (t-1, c), CSI (t-2, c) ..., CSI (t-Δ t, c)]TWith frequency domain X2=[CSI (t, c ± 1), CSI (t, c ± 2) ..., CSI (t, c ± Δ f)]TCommon training, obtain output vector Y1t,Y2t, by Y1t,Y2tAs the input vector of second layer network, obtain final predictive value Yt, the method combined with time domain and frequency domain builds time-frequency two-dimensional LMBP neutral net and realizes the spectrum prediction of ISM band.
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 5, it is characterized in that, step 3 particularly as follows:, using the measured data of ISM band as training sequence input time-frequency two-dimensional LMBP neutral net in, obtain output vector Y=[Y1,Y2,Y3,...,Ym]T, and with object vector constitute error function:
Wherein, the parameter vector that u is made up of neural network weight w and threshold values b:
The Newton method learning rules of parameter vector u:WhereinFor Hessian inverse of a matrix matrix;gkFor the gradient of F (u),J(uk) it is the Jacobian matrix of e (u).
Spectrum prediction method based on time-frequency two-dimensional LMBP neutral net in ISM band the most according to claim 5, it is characterised in that in step 4 particularly as follows: build time domain and frequency domain input vector matrix X with ISM band measured datat, Xf:
By Xt, XfAs neutral net input vector, the time-frequency two-dimensional LMBP neutral net having reached optimal value by weight vector w and threshold vector b, obtain output vector Y=[Y1,Y2,Y3,...,Ym]T, YmIt is CSI (tm,cm) predictive value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610149087.8A CN105915299B (en) | 2016-03-16 | 2016-03-16 | Spectrum prediction method based on time-frequency two-dimensional LMBP neural networks in ISM band |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610149087.8A CN105915299B (en) | 2016-03-16 | 2016-03-16 | Spectrum prediction method based on time-frequency two-dimensional LMBP neural networks in ISM band |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105915299A true CN105915299A (en) | 2016-08-31 |
CN105915299B CN105915299B (en) | 2018-08-14 |
Family
ID=56745058
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610149087.8A Active CN105915299B (en) | 2016-03-16 | 2016-03-16 | Spectrum prediction method based on time-frequency two-dimensional LMBP neural networks in ISM band |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105915299B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20210150286A1 (en) * | 2019-11-20 | 2021-05-20 | Rohde & Schwarz Gmbh & Co. Kg | Method and system for detecting and/or classifying a wanted signal |
CN113541700A (en) * | 2017-05-03 | 2021-10-22 | 弗吉尼亚科技知识产权有限公司 | Method, system and apparatus for learning radio signals using a radio signal converter |
WO2021237423A1 (en) * | 2020-05-25 | 2021-12-02 | Oppo广东移动通信有限公司 | Channel state information transmission methods, electronic device, and storage medium |
CN116527177A (en) * | 2023-04-28 | 2023-08-01 | 哈尔滨工程大学 | Three-dimensional spectrum prediction method based on correlation analysis and graph convolution network |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103209417A (en) * | 2013-03-05 | 2013-07-17 | 北京邮电大学 | Method and device for predicting spectrum occupancy state based on neural network |
-
2016
- 2016-03-16 CN CN201610149087.8A patent/CN105915299B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103209417A (en) * | 2013-03-05 | 2013-07-17 | 北京邮电大学 | Method and device for predicting spectrum occupancy state based on neural network |
Non-Patent Citations (3)
Title |
---|
吴建绒等: "基于K-RBF神经网络的认知无线电频谱预测", 《电视技术》 * |
邢晓双: "认知无线电网络中的频谱预测技术研究", 《中国优秀博士论文全文数据库 信息科技辑》 * |
陈斌华: "认知无线电系统中的频谱预测算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113541700A (en) * | 2017-05-03 | 2021-10-22 | 弗吉尼亚科技知识产权有限公司 | Method, system and apparatus for learning radio signals using a radio signal converter |
CN113541700B (en) * | 2017-05-03 | 2022-09-30 | 弗吉尼亚科技知识产权有限公司 | Method, system and apparatus for learning radio signals using a radio signal converter |
US11468317B2 (en) | 2017-05-03 | 2022-10-11 | Virginia Tech Intellectual Properties, Inc. | Learning radio signals using radio signal transformers |
US20210150286A1 (en) * | 2019-11-20 | 2021-05-20 | Rohde & Schwarz Gmbh & Co. Kg | Method and system for detecting and/or classifying a wanted signal |
WO2021237423A1 (en) * | 2020-05-25 | 2021-12-02 | Oppo广东移动通信有限公司 | Channel state information transmission methods, electronic device, and storage medium |
CN116527177A (en) * | 2023-04-28 | 2023-08-01 | 哈尔滨工程大学 | Three-dimensional spectrum prediction method based on correlation analysis and graph convolution network |
CN116527177B (en) * | 2023-04-28 | 2024-03-12 | 哈尔滨工程大学 | Three-dimensional spectrum prediction method based on correlation analysis and graph convolution network |
Also Published As
Publication number | Publication date |
---|---|
CN105915299B (en) | 2018-08-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105634787B (en) | Appraisal procedure, prediction technique and the device and system of network key index | |
CN105915299A (en) | Time-frequency two-dimensional LMBP neural network based frequency spectrum prediction method in ISM frequency range | |
CN110730046B (en) | Cross-frequency-band spectrum prediction method based on deep migration learning | |
CN105653502B (en) | A kind of communication base station electromagnetic radiation correlation analysis based on genetic algorithm | |
Angjelicinoski et al. | Comparative analysis of spatial interpolation methods for creating radio environment maps | |
CN108876021B (en) | Medium-and-long-term runoff forecasting method and system | |
CN110138475A (en) | A kind of adaptive threshold channel occupation status prediction technique based on LSTM neural network | |
CN105024951B (en) | A kind of power delay spectrum PDP methods of estimation and device | |
Olive | Asymptotically optimal regression prediction intervals and prediction regions for multivariate data | |
CN110635858B (en) | Autoregressive model channel prediction method based on quantum computation | |
CN105929340A (en) | Method for estimating battery SOC based on ARIMA | |
CN109344993B (en) | River channel flood peak water level forecasting method based on conditional probability distribution | |
CN103491551B (en) | A kind of weighting cooperative frequency spectrum sensing method of feature based vector | |
Peng et al. | A small scale forecasting algorithm for network traffic based on relevant local least squares support vector machine regression model | |
CN103916969A (en) | Combined authorized user perception and link state estimation method and device | |
CN104640137B (en) | A kind of QoE optimization methods accessing selection based on wireless ubiquitous network | |
CN106372348A (en) | Vector fitting model order reduction method based on error control in linear system | |
CN103886391A (en) | Method and device for predicating service volume | |
CN107167658B (en) | A kind of jamproof electric system fundamental frequency of high-precision and Method for Phase Difference Measurement | |
CN103139828A (en) | Broadband spectrum sensing device and method | |
WO2023169589A1 (en) | Predictive channel modeling method based on adversarial network and long short-term memory network | |
CN104601264A (en) | Multi-antenna spectrum sensing method applicable to high-dimension finite sample conditions | |
Nagamatsu et al. | Padé approximation for coverage probability in cellular networks | |
CN106295903A (en) | The efficiency Forecasting Methodology of WSN data Lossy Compression Algorithm based on linear fit | |
CN109495197A (en) | A kind of adaptive wideband cooperation compression frequency spectrum sensing method |
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 |