Disclosure of Invention
In order to overcome the defects, the invention aims to provide an accurate indoor positioning method for the field, so that the technical problems that the existing similar methods are easily interfered by other signals, and the positioning accuracy is poor due to errors in node distance measurement are solved. The purpose is realized by the following technical scheme.
An accurate indoor positioning method is characterized in that the indoor positioning method utilizes the correlation among objects to construct the coordinate distribution of corresponding points in space, and takes the actual distance of each object as the correlation to form a correlation matrix; then centralizing and decomposing the correlation matrix to obtain the relative position distribution of each point in the space; and finally, converting the relative coordinates of the nodes into absolute coordinates by using mirror image transformation, thereby realizing indoor positioning. The indoor positioning method provides an algorithm with high robustness, low energy consumption and high calculation time efficiency, and indoor positioning can be accurately realized without interference only by knowing the absolute coordinate positions of a small number of reference nodes in a space. Meanwhile, the indoor positioning method provides a technology for converting the space relative coordinates of the reference nodes into space absolute coordinates so as to realize accurate indoor positioning.
The specific flow of the indoor positioning method is as follows:
s100: in the conversion process of the relative coordinates, the wider the distribution of the reference nodes, the more stable the positioning performance, and in a two-dimensional space, the spatial relative coordinates of d +1 reference nodes forming the largest d +1 polygon area need to be determined; initializing k to 1, e to 0, determining the center coordinates p of the m reference nodes:
selecting a reference node a farthest from the center
i;
S200: d reference nodes are arbitrarily selected from the rest m-1 reference nodes, and all combinations are stored in C
(d-1)×rIn, C
(d-1)×rIn order to be a combination of the mathematics,
where r ═ (m-1) (m-2)/2, d is the spatial dimension in which localization is achieved;
s300: calculating ai,c1k,c2k,...,c(d-1)kArea s of d +1 polygon formed by reference nodeskIf s iskGreater than e, then e ═ skJ is k; if k equals r, continue the next step, otherwise k equals k +1 and repeat the step;
s400: let p be
ijRepresenting the correlation between objects i and j, X
n×mA coordinate matrix representing corresponding points of each object in the space, wherein n represents the number of the points, and m represents the dimension of the space; d
ij(X) represents the actual distance between the corresponding points of the objects i and j in the space, and then the actual distance is calculated by the formula
Calculating the actual distance and constructing a correlation matrix [ d ]
ij];
S500: for correlation matrix [ d ]ij]Centralizing and decomposing to obtain the distribution of the relative positions of each point in the space;
s600: order (y)
1,y
2,...,y
n) Representing the actual coordinates of the node, (x)
1,x
2,...,x
n) Representing the relative coordinates of the nodes obtained by the method, wherein n is the total number of the nodes, and y is
*(xvr + L) wherein ═ represents
R denotes a d x d orthogonal rotation matrix, d denotes a spatial dimension,
represents an offset; to determine the absolute position of the node for R and L, y
*Equivalent of xR + L is a system of equations
Solving the solution;
s700: and realizing indoor positioning according to the obtained absolute coordinates.
The distribution comprises the following specific steps:
s510: through a formula d'ij=dijU performs a correlation matrix [ d ]ij]The data centering, i.e., the data translation process, of (1), wherein u represents the mean of the data samples;
s520: by the formula [ dij]=u∑V*Proceed correlation matrix [ d ]ij]The singular value decomposition of (a) in the formula represents a conjugate transpose, where the columns of u form a set of pairs [ d ]ij]Is the basis vector of the orthogonal "inputs" or "analyses", these vectors being [ dij][dij]*The feature vector of (2); v's column constitutes a set of pairs [ dij]The basis vectors of the orthogonal "outputs" of (1), these vectors being [ d ]ij]*[dij]The feature vector of (2); according to the processed correlation matrix [ d ]ij]And dijAnd (X) calculating a formula to obtain the distribution of the relative positions of the points in the space.
The concrete steps of the solution are as follows:
s610: three reference nodes (p) with the largest area in two-dimensional space
1,p
2,p
3) Conversion to (q)
1,q
2,q
3) First, p is
1p
2Is translated so that p
1And q is
1Coincidence, p
2Move to p'
2=p
2+q
1-p
1(ii) a Then q is put
1p′
2Is rotated so that p'
2And q is
2Overlapping; let g be p
1-q
1Then there is
Wherein H is v ═ p
1p
2+q
1q
2)/||p
1p
2+q
1A mirror matrix of q | |, v being the unit column vector of the H matrix;
s620: by
And
and deriving R-H-I-vv from the uniqueness of the solution
T,
Is the solution of the system of equations, where the I identity matrix.
The indoor positioning method is feasible, the indoor positioning result is accurate, the positioning is convenient and fast, and the application range is wide; the method is suitable for being applied as an indoor positioning method of similar products and improvement of the similar positioning method.
Detailed Description
The construction and use of the invention will now be further described with reference to the accompanying drawings. The indoor positioning method utilizes the correlation (actual distance in positioning, namely the real distance between two points in space) between objects to construct the coordinate distribution of corresponding points in space, takes the actual distance of each object as the correlation to form a correlation matrix, and then centers and decomposes the correlation matrix to obtain the relative position distribution of each point in space; and finally, converting the relative coordinates of the nodes into absolute coordinates by using mirror image transformation, thereby realizing indoor positioning.
As shown in fig. 1, the specific flow of the indoor positioning method is as follows:
s100: in the conversion process of the relative coordinates, the wider the distribution of the reference nodes, the more stable the positioning performance, and in a two-dimensional space, the spatial relative coordinates of d +1 reference nodes forming the largest d +1 polygon area need to be determined; initializing k to 1, e to 0, determining the center coordinates p of the m reference nodes:
selecting a reference node a farthest from the center
i;
S200: d reference nodes are arbitrarily selected from the rest m-1 reference nodes, and all combinations are stored in C
(d-1)×rIn, C
(d-1)×rIn order to be a combination of the mathematics,
wherein r ═ 2 (m-1) and d is achievedThe spatial dimension of the location;
s300: calculating ai,c1k,c2k,...,c(d-1)kArea s of d +1 polygon formed by reference nodeskIf s iskGreater than e, then e ═ skJ is k; if k equals r, continue the next step, otherwise k equals k +1 and repeat the step;
s400: let p be
ijRepresenting the correlation between objects i and j, X
n×mA coordinate matrix representing corresponding points of each object in the space, wherein n represents the number of the points, and m represents the dimension of the space; d
ij(X) represents the actual distance between the corresponding points of the objects i and j in the space, and then the actual distance is calculated by the formula
Calculating the actual distance and constructing a correlation matrix [ d ]
ij];
S500: for correlation matrix [ d ]ij]Centralizing and decomposing to obtain the distribution of the relative positions of the points in the space.
The distribution comprises the following specific steps:
s510: through a formula d'ij=dijU performs a correlation matrix [ d ]ij]The data centering, i.e., the data translation process, of (1), wherein u represents the mean of the data samples;
s520: by the formula [ dij]=u∑V*Proceed correlation matrix [ d ]ij]The singular value decomposition of (a) in the formula represents a conjugate transpose, where the columns of u form a set of pairs [ d ]ij]Is the basis vector of the orthogonal "inputs" or "analyses", these vectors being [ dij][dij]*The feature vector of (2); v's column constitutes a set of pairs [ dij]The basis vectors of the orthogonal "outputs" of (1), these vectors being [ d ]ij]*[dij]The feature vector of (2); according to the processed correlation matrix [ d ]ij]And dij(X) obtaining the distribution of the relative positions of each point in the space by using a calculation formula;
s600: order (y)
1,y
2,...,y
n) Representing the actual coordinates of the node, (x)
1,x
2,...,x
n) Representing the relative coordinates of the nodes obtained by the method, wherein n is the total number of the nodes, and y is
*(xvr + L) wherein ═ represents
R denotes a d x d orthogonal rotation matrix, d denotes a spatial dimension,
represents an offset; to determine the absolute position of the node for R and L, y
*Equivalent of xR + L is a system of equations
And solving the solution.
The concrete steps of the solution are as follows:
s610: three reference nodes (p) with the largest area in two-dimensional space
1,p
2,p
3) Conversion to (q)
1,q
2,q
3) First, p is
1p
2Is translated so that p
1And q is
1Coincidence, p
2Move to p'
2=p
2+q
1-p
1(ii) a Then q is put
1p′
2Is rotated so that p'
2And q is
2Overlapping; let g be p
1-q
1Then there is
Wherein H is v ═ p
1p
2+q
1q
2)/||p
1p
2+q
1A mirror matrix of q | |, v being the unit column vector of the H matrix;
s620: by
And
and deriving R-H-I-vv from the uniqueness of the solution
T,
Is a solution of the system of equations, wherein the I identity matrix;
s700: and realizing indoor positioning according to the obtained absolute coordinates.
In summary, the indoor positioning technology is applied on a larger scale nowadays, and the algorithm provided by the invention can reduce the influence of reference node distribution on positioning accuracy, so that more stable performance is obtained, and the purpose of accurate indoor positioning is achieved. And aiming at large-scale indoor positioning, the algorithm provided by the invention has high-efficiency calculation, removes unnecessary reference nodes, ensures the accuracy and can save the expense of calculation time.