CN103745489A - Method for constructing base station signal field intensity map based on compressed sensing - Google Patents

Abstract

The invention relates to the technical field of compressed sensing and the technical field of wireless communication, such as mobile communication and wireless communication, in particular to a method for constructing a base station signal field intensity map based on compressed sensing in mobile communication networks such as Wi-Fi (wireless fidelity) wireless local area networks and cellular communication. The method is based on the sparse decomposition and compressed sensing technology of a redundant dictionary, and through constructing a special observing matrix, a high-resolution field intensity distribution map is constructed on the basis of a low-resolution single-frame actual measuring field intensity map. Through the method provided by the invention, the wireless signal field intensity map can be effectively subjected to recovery reconstruction, the work load of measuring the signal field intensity values in the off-line stage in the application such as wireless network planning and wireless field intensity positioning can be actually reduced, and the construction efficiency and the accuracy of the signal intensity map can be improved.

Description

A kind of method of setting up base station signal field intensity map based on compressed sensing

Technical field

The present invention relates to compressed sensing technical field and wireless communication technology field, in mobile communication, radio communication, specially refer to the method for setting up base station signal field intensity map in the mobile communications networks such as Wi-Fi WLAN (wireless local area network), cellular communication based on compressed sensing.

Background technology

In mobile communications network, the field intensity of base station signal distribution situation is geographically schemed on the spot doughtily, is very valuable data in radio communication operation, in the business such as programming and distribution of the service based on geographic position, wireless network, has important application.Yet wireless signal field map is but difficult to obtain in simple mode.This is mainly that communication environments by wireless signal complexity causes.In mobile communications network, base station is very complicated to the travel path between receiver, and the processes such as barrier reflection, refraction, diffraction and multipath transmisstion from simple line-of-sight propagation to various complexity all may run into.So the signal strength in wireless transmission distributes and can not accurately predict, the difficulty that the randomness of wireless transmission has caused wireless field density to obtain.

Common wireless signal field map is to obtain by surveying or survey the method that adds interpolation.While directly using the method for actual measurement, obtain high-resolution map and can only obtain by a large amount of field survey work, so workload is huge, can only be used in the region that area is less; First the field intensity map in larger region needs to obtain by actual measurement the field intensity data of some, then reaches higher resolution by the method for interpolation.According to the quality of interpolation method, the actual measurement workload needing and the final accuracy of map obtaining have very large difference, and existing method still needs larger actual measurement workload.Therefore, be necessary to further investigate the constructing technology of field intensity map, reduce the resolution of surveying workload and improving map.Advanced field intensity map structuring technology can reduce actual measurement workload at double, saves time and manpower and materials.According to the mathematical method of using in field intensity map structuring process, field intensity Map building method can be divided into classical method of interpolation, propagation model computing method, super-resolution rebuilding method etc.

Classical method of interpolation is the image interpolation algorithm of using for reference in image processing field, utilizes the known of close position point or actual measurement field intensity value, produces the field intensity value of unknown position point with certain interpolating function.These algorithms comprise proximal point algorithm, bilinear interpolation etc.Although classical interpolation method is easily gone fast, reconstruction effect is unsatisfactory, and high-frequency information is lost serious, and resolution is difficult for raising.

Propagation model computing method is propagation (decay) model that makes full use of wireless signal, comprises deterministic models and empirical model, and the field intensity data of unknown point are calculated.Yet, complicacy due to actual propagation environment, be the propagation condition that determinacy propagation model or experience propagation model are all difficult to simulate exactly actual signal, the signal strength map therefore calculating according to propagation model still has larger error, and applicable is limited in scope.

Super-resolution rebuilding method is different with method of interpolation.Super-resolution rebuilding can be according to a width or a few width same scene but the field intensity map of different angles, reconstructs the field intensity map of a width more clear (being that resolution is higher) according to certain mathematical programming.The key of Super-resolution Reconstruction is to obtain the low resolution copy of image itself and the corresponding relation between high resolving power copy, utilizes the reconstruction rule of this pass series structure more more accurate than common interpolating function.Super-resolution research at present can be divided into three main category: based on interpolation, based on reconstruction and the method based on study.Wherein the method based on study is the focus direction of super-resolution algorithms research in recent years, its basic ideas are by given training atlas, calculate image block and the training plan of test sample book and concentrate the neighborhood relationships between image block set, and construct optimum weights constraint, obtain priori and approach the High Resolution Ground Map of test sample book.When the information providing when high-resolution data does not meet high resolving power demand, the method based on study can obtain more map high-frequency information, thereby tool has great advantage.

The existing Super-resolution Reconstruction method based on study is mainly used in image processing field, is not also applied to the report of wireless signal field map structuring in open source information.The Super-resolution Reconstruction algorithm based on study that image processing field is used mainly contains document " Image Super-Resolution as Sparse Representation of Raw Image Patches " (Jianchao Yang, IEEE Conferenceon Computer Vision and Pattern Recognition, 2008) report in, the method cardinal principle is the Its Sparse Decomposition model in redundant dictionary based on picture signal, utilize the high resolving power copy of image under same redundant dictionary, to there is the identical feature of sparse coefficient with low resolution copy, the low-resolution image that observation is obtained carries out Its Sparse Decomposition, recycling is decomposed the sparse coefficient and the high resolving power dictionary that obtain and is carried out Super-resolution Reconstruction.

Said method is also not exclusively applicable to the structure of wireless signal field map.In wireless signal field, field strength attenuation process is relevant with factors such as propagation distance, travel paths, and processes by image mode these factors that gives no thought to.On the other hand, the compressed sensing technology based on Its Sparse Decomposition technology, needs to introduce observing matrix, can consider the signal attenuation constituent element that some are actual in signal rejuvenation, therefore has certain room for improvement.

Summary of the invention

The object of the invention is to overcome the above-mentioned deficiency of prior art, provide a kind of method of setting up base station signal field intensity map based on compressed sensing to construct the High Resolution Ground Map that base station radio signal strength distributes.

For achieving the above object, the invention provides following technical scheme:

A method of setting up base station signal field intensity map based on compressed sensing, comprises the following steps:

(1), according to the resolution of needed final map, determine the quantity of the location point in this field intensity map, the quantity of establishing location point is n, wherein through actual measurement location point number be m, n should meet m<n<100m;

(2), according to the field intensity numerical value of an existing m eyeball in this map, utilize the common interpolation algorithm of Krieger (kriging), set up the field intensity value of all unmeasured location points, obtain preliminary field strength distribution interpolation map, if the matrix form that this preliminary field intensity map is corresponding is C, corresponding by column vector form is

(3) the field intensity value of, establishing all m eyeball in map Matrix C is P ₁, P ₂..., P _i..., P _m.Define a radius of influence R, and think that an eyeball i is subject to the impact of the field intensity value of other location points in round property region that radius is R, claim these " affecting a little " that point is position i, calculate " affecting a little " quantity of each eyeball, use variable K _ithe quantity that represents the impact point of each eyeball i;

(4), by preliminary field intensity Matrix C, constructed the down-sampling matrix A of corresponding eyeball field intensity value, the element in A is the weighing factor coefficient of field intensity value of the impact point of corresponding each eyeball i, these coefficients are calculated by following formula:

${\mathrm{\&alpha;}}_{\mathrm{ik}}=\frac{1}{K_i}\frac{P_i}{P_{\mathrm{ik}}},k=\mathrm{1,2},......,K_i$

α wherein _ikk that is eyeball i affects the field intensity of point with respect to the weight of the field intensity of eyeball i, P _ithe field intensity of eyeball i position, P _ikit is the field intensity of k the impact point position of eyeball i.The field intensity value P of eyeball i _ican be written as K _ithe weighted sum of the field intensity value of individual impact point:

$P_i={\mathrm{\&alpha;}}_{i1}{P}_{i1}+{\mathrm{\&alpha;}}_{i2}{P}_{i2}+...{\mathrm{\&alpha;}}_{\mathrm{ik}}{P}_{\mathrm{ik}}+...+{\mathrm{\&alpha;}}_{{\mathrm{iK}}_i}{P}_{{\mathrm{iK}}_i}$

To whole investigation region, the field intensity value of all eyeballs can be write as matrix product form:

$\stackrel{\&RightArrow;}{M}=A\stackrel{\&RightArrow;}{C}$

Wherein:

for the vector of all eyeball field intensity values formations, $\stackrel{\&RightArrow;}{M}={\left[\begin{array}{cccc}P1& P2& \&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;& \mathrm{Pm}\end{array}\right]}^T;$

for the preliminary field intensity interpolation map Matrix C that obtained by step (2) by column vector form;

A is the sampling weight coefficient α by all impact points of aforesaid each eyeball of correspondence _ikthe sampling matrix forming;

(5), using the sampling matrix A that sets up in step (4) as observing matrix, set up compressed sensing model:

$\stackrel{\&RightArrow;}{M}=\mathrm{AB}\stackrel{\&RightArrow;}{S},$

Wherein,

for eyeball field intensity vector, A is observing matrix (sampling matrix), and B is by the former word bank of field intensity map Its Sparse Decomposition of KSVD algorithm construction, for field intensity map Its Sparse Decomposition coefficient to be solved;

(6), by base, follow the trail of BP restructing algorithm, the sparse coefficient in solution procedure (5)

(7), according to the sparse coefficient obtaining reconstruct high-resolution field intensity map vector

$\stackrel{\&RightArrow;}{X}=B\stackrel{\&RightArrow;}{S}$

According to this vector , then write as matrix form X, obtain final field intensity high-resolution map.

The present invention is based on Its Sparse Decomposition and the compressed sensing technology of redundant dictionary, by constructing special observing matrix, on the single width actual measurement field intensity map basis of low resolution, construct high-resolution field strength distribution map.By method provided by the invention, can effectively to wireless signal field map, carry out restoration and reconstruction, conscientiously minimizing, in the workload of off-line phase measuring-signal field intensity value in the application such as wireless network planning, wireless field density location, improves efficiency and the accuracy of signal intensity map structuring.

Compared with prior art, beneficial effect of the present invention:

1. the present invention is applied to compressed sensing technology in the building process of field intensity map of wireless base station signal, can realize the signal strength map of the sampled point structure high-resolution (being higher density) of using low resolution (more sparse), thereby can reduce actual measurement workload, improve map structuring efficiency.

2. the present invention has utilized the compressed sensing principle of interpolation technique and image simultaneously, by first using Krieger interpolation process, utilized the distance factor of signal attenuation process, by compressed sensing process of reconstruction, utilized the former word bank by similar wireless signal Map building again simultaneously, thereby in process of reconstruction, used more actual signal information, rather than simply wireless signal map is regarded as to picture signal, its result more approaches true field intensity signal than general process of image interpolation, has higher degree of accuracy.

Accompanying drawing explanation

Fig. 1 is schematic flow sheet of the present invention.

Fig. 2 is an example of actual measurement field intensity map.

Fig. 3 is for carrying out the preliminary field intensity map of Kriging interpolation to Fig. 2.

Fig. 4 is that the final high-resolution field recovering is schemed doughtily.

Embodiment

Below in conjunction with the drawings and specific embodiments, the present invention is further illustrated.

In order to make object, technical scheme and the advantage of invention clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.

Embodiment 1

The present embodiment has been introduced a kind of method of setting up base station signal field intensity map based on compressed sensing, as shown in Figure 1, comprises the following steps:

(1), first collect some existing base station signal high-resolution fields and scheme doughtily, utilize KSVD algorithm to carry out sparse redundant dictionary decomposition, obtain the redundant dictionary storehouse of corresponding wireless signal field map.

(2), the field intensity map example of take in Fig. 2 is example, setting final recovery field intensity map resolution is 50 * 50.Fig. 2 is carried out to Ordinary Kriging Interpolation (Kriging) algorithm interpolation, obtain 50 * 50 preliminary interpolation map, as shown in Figure 3, its image array is expressed as C, and what map matrix was corresponding is expressed as by column vector form

(3) the field intensity value of, establishing all 625 eyeballs in map Matrix C is P ₁, P ₂..., P _i..., P ₆₂₅.Definition radius of influence R=1.5, the unit of this radius is the row or column in Matrix C, and thinks that an eyeball i is subject to the impact of the field intensity value of other location points within the scope of radius R, claims these " affecting a little " that point is position i.Calculate " affecting a little " quantity of each eyeball.Use variable K _ithe quantity that represents the impact point of each eyeball i, in this example, each eyeball has 8 to affect a little.

(4), by preliminary field intensity Matrix C, constructed the down-sampling matrix A of corresponding eyeball field intensity value, the element in A is the weighing factor coefficient of field intensity value of the impact point of corresponding each eyeball i, these coefficients are calculated by following formula:

${\mathrm{\&alpha;}}_{\mathrm{ik}}=\frac{1}{8}\frac{P_i}{P_{\mathrm{ik}}},k=\mathrm{1,2},......,8$

α wherein _ikk that is eyeball i affects the field intensity of point with respect to the weight of the field intensity of eyeball i, P _ithe field intensity of eyeball i position, P _ikit is the field intensity of k the impact point position of eyeball i.The field intensity value P of eyeball i _ican be written as K _ithe weighted sum of the field intensity value of individual impact point:

$P_i={\mathrm{\&alpha;}}_{i1}{P}_{i1}+{\mathrm{\&alpha;}}_{i2}{P}_{i2}+...+{\mathrm{\&alpha;}}_{\mathrm{ik}}{P}_{\mathrm{ik}}+...+{\mathrm{\&alpha;}}_{i8}{P}_{i8}$

To whole investigation region, the field intensity value of all eyeballs can be write as matrix product form:

$\stackrel{\&RightArrow;}{M}=A\stackrel{\&RightArrow;}{C}$

Wherein:

for the vector of all eyeball field intensity values formations, $\stackrel{\&RightArrow;}{M}={\left[\begin{array}{cccc}{P}_1& P_2& \&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;& P_{625}\end{array}\right]}^T;$

for the preliminary field intensity interpolation map Matrix C that obtained by step (2) by column vector form;

A is the sampling weight coefficient α by all impact points of aforesaid each eyeball of correspondence _ikthe sampling matrix forming, matrix A has following form:

$A=$

$\left[\begin{array}{cccccccccccccccccccc}{\mathrm{\&alpha;}}_{11}& {\mathrm{\&alpha;}}_{12}& {\mathrm{\&alpha;}}_{13}& 0& ...& 0& {\mathrm{\&alpha;}}_{14}& 0& {\mathrm{\&alpha;}}_{15}& 0& ...& 0& {\mathrm{\&alpha;}}_{16}& {\mathrm{\&alpha;}}_{17}& {\mathrm{\&alpha;}}_{18}& 0& ...& ...& ...& 0\\ 0& 0& {\mathrm{\&alpha;}}_{21}& {\mathrm{\&alpha;}}_{22}& {\mathrm{\&alpha;}}_{23}& 0& ...& 0& {\mathrm{\&alpha;}}_{24}& 0& {\mathrm{\&alpha;}}_{25}& 0& ...& 0& {\mathrm{\&alpha;}}_{26}& {\mathrm{\&alpha;}}_{27}& {\mathrm{\&alpha;}}_{28}& ...& ...& 0\\ ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& .\\ ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& ..& 0\\ 0& ..& ..& 0& 0& {\mathrm{\&alpha;}}_{6251}& {\mathrm{\&alpha;}}_{6252}& {\mathrm{\&alpha;}}_{6253}& 0& ...& 0& {\mathrm{\&alpha;}}_{6254}& 0& {\mathrm{\&alpha;}}_{6255}& 0& ..& 0& {\mathrm{\&alpha;}}_{6256}& {\mathrm{\&alpha;}}_{6257}& {\mathrm{\&alpha;}}_{6258}\end{array}\right]$

(5), using the sampling matrix A that sets up in step (4) as observing matrix, set up compressed sensing model:

$\stackrel{\&RightArrow;}{M}=\mathrm{AB}\stackrel{\&RightArrow;}{S},$

Wherein, for eyeball field intensity vector, A is observing matrix (sampling matrix), and B is by the former word bank of field intensity map Its Sparse Decomposition of KSVD algorithm construction, for field intensity map Its Sparse Decomposition coefficient vector to be solved;

(6), by base, follow the trail of BP restructing algorithm, the sparse coefficient in solution procedure (5)

by the sparse constraint item to following, solve and obtain sparse coefficient

vector:

$\underset{\stackrel{\&RightArrow;}{S}}{\min }{\left|\right|\stackrel{\&RightArrow;}{M}-\mathrm{AB}\stackrel{\&RightArrow;}{S}\left|\right|}_2^2+\mathrm{\&lambda;}{\left|\right|\stackrel{\&RightArrow;}{S}\left|\right|}_0$

In above formula

a, B, the same step of definition (5), λ is regularization parameter,

represent L2 norm, || || ₀represent L0 norm.

(7), according to the sparse coefficient obtaining reconstruct high-resolution field intensity map vector

$\stackrel{\&RightArrow;}{X}=B\stackrel{\&RightArrow;}{S}$

According to this vector

write again as matrix form X, obtained final field intensity high-resolution map as accompanying drawing 4.

The present embodiment is applied to compressed sensing technology on the super-resolution rebuilding of wireless signal field map, by constructing a kind of weight sampling matrix as the observing matrix in compressive sensing theory, the deamplification forming process and the signal actual attenuation process that make to measure chess matrix analogue are more approaching, thereby realize on a small amount of measured signal of foundation basis, improve the resolution of reconstruction signal map, the sampled point of having realized use low resolution (more sparse) builds the signal strength map of high-resolution (being higher density), thereby can reduce actual measurement workload, improve map structuring efficiency.The present embodiment adopts the Its Sparse Decomposition principle of image in addition, by first using Krieger interpolation process, utilized the distance factor of signal attenuation process, by compressed sensing process of reconstruction, utilized the former word bank by similar wireless signal Map building again simultaneously, thereby in process of reconstruction, used more actual signal information, rather than simply wireless signal map is regarded as to picture signal, its result more approaches true field intensity signal than general process of image interpolation, there is higher degree of accuracy, signal strength map after assurance restoration and reconstruction is compared with traditional approach better effects if.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any modifications of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.

Claims (1)

1. based on compressed sensing, set up a method for base station signal field intensity map, it is characterized in that: comprise the following steps:

(1), according to the resolution of needed final map, determine the quantity of the location point in this field intensity map, the quantity of establishing location point is n, wherein through actual measurement location point number be m, n should meet m<n<100m;

(2), according to the field intensity numerical value of an existing m eyeball in this map, utilize the common interpolation algorithm of Krieger (kriging), set up the field intensity value of all unmeasured location points, obtain preliminary field strength distribution interpolation map, if the matrix form that this preliminary field intensity map is corresponding is C, corresponding by column vector form is

;

(3) the field intensity value of, establishing all m eyeball in map Matrix C is P ₁, P ₂..., P _i..., P _m, define a radius of influence R, and think that an eyeball i is subject to the impact of the field intensity value of other location points in round property region that radius is R, claim these " affecting a little " that point is position i, calculate " affecting a little " quantity of each eyeball, use variable K _ithe quantity that represents the impact point of each eyeball i;

(4), by preliminary field intensity Matrix C, constructed the down-sampling matrix A of corresponding eyeball field intensity value, the element in A is the weighing factor coefficient of field intensity value of the impact point of corresponding each eyeball i, these coefficients are calculated by following formula:

${\mathrm{\&alpha;}}_{\mathrm{ik}}=\frac{1}{K_i}\frac{P_i}{P_{\mathrm{ik}}},k=\mathrm{1,2},......,K_i$

α wherein _ikk that is eyeball i affects the field intensity of point with respect to the weight of the field intensity of eyeball i, P _ithe field intensity of eyeball i position, P _ikthe field intensity of k the impact point position of eyeball i, the field intensity value P of eyeball i _ican be written as K _ithe weighted sum of the field intensity value of individual impact point:

$P_i={\mathrm{\&alpha;}}_{i1}{P}_{i1}+{\mathrm{\&alpha;}}_{i2}{P}_{i2}+...{\mathrm{\&alpha;}}_{\mathrm{ik}}{P}_{\mathrm{ik}}+...+{\mathrm{\&alpha;}}_{{\mathrm{iK}}_i}{P}_{{\mathrm{iK}}_i}$

To whole investigation region, the field intensity value of all eyeballs can be write as matrix product form:

$\stackrel{\&RightArrow;}{M}=A\stackrel{\&RightArrow;}{C}$

Wherein:

for the vector of all eyeball field intensity values formations, $\stackrel{\&RightArrow;}{M}={\left[\begin{array}{cccc}P1& P2& \&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;& \mathrm{Pm}\end{array}\right]}^T;$

for the preliminary field intensity interpolation map Matrix C that obtained by step (2) by column vector form;

A is the sampling weight coefficient α by all impact points of aforesaid each eyeball of correspondence _ikthe sampling matrix forming;

(5), using the sampling matrix A that sets up in step (4) as observing matrix, set up compressed sensing model:

$\stackrel{\&RightArrow;}{M}=\mathrm{AB}\stackrel{\&RightArrow;}{S},$

Wherein,

for eyeball field intensity vector, A is observing matrix (sampling matrix), and B is by the former word bank of field intensity map Its Sparse Decomposition of KSVD algorithm construction, for field intensity map Its Sparse Decomposition coefficient to be solved;

(6), by base, follow the trail of BP restructing algorithm, the sparse coefficient in solution procedure (5)

(7), according to the sparse coefficient obtaining , reconstruct high-resolution field intensity map vector

$\stackrel{\&RightArrow;}{X}=B\stackrel{\&RightArrow;}{S}$

According to this vector

, then write as matrix form X, obtain final field intensity high-resolution map.

Images