CN106454750A - Multi-region indoor safety positioning method based on compressed sensing technology - Google Patents

Abstract

The present invention provides a multi-region indoor safe positioning method based on compressed sensing technology. The positioning method utilizes the theory of compressed sensing and includes three stages: offline, safety monitoring and online; 1) within the set area where the node to be positioned is located, Deploy the anchor nodes according to an equilateral triangle, and construct the RSSI security key matrix of the node to be located according to the distance between the node to be located and the anchor node; use the actual RSSI value received by the node to be located and the correspondence in the RSSI security key matrix Values are compared to remove malicious anchor nodes; the node to be positioned receives the perceived anchor node information in real time to draw the overlapping area, and the signal recovery algorithm is used to restore the orthogonalized sparse signal to perform accurate position estimation and obtain the location of the node to be positioned positioning information. By establishing an adaptive random number dictionary, malicious positioning information is eliminated, so as to ensure the security of positioning.

Description

A Multi-region Indoor Safe Positioning Method Based on Compressed Sensing Technology

technical field

The invention belongs to the field of wireless sensor networks, in particular to a multi-region indoor safe positioning method based on compressed sensing technology.

Background technique

Wireless sensor network indoor positioning has developed rapidly in recent years, including the application of intelligent sensing technology, embedded computing technology and communication technology to improve the accuracy of indoor positioning and broaden the application range of indoor positioning.

Indoor positioning has been successfully applied in various fields in recent years, including museum navigation, warehouse navigation, parking lot navigation, etc. However, the current research in this field is mainly focused on how to improve the positioning accuracy and energy efficiency, and insufficient consideration is given to the security of wireless sensor networks in the process of indoor positioning. Since wireless sensor nodes are deployed openly, they are easily captured by attackers, and the physical properties of the signals based on positioning are easily tampered by the outside world. And because the signal transmission process is carried out in a completely open environment, this leads to the node's signal being easily intercepted or tampered with during transmission. The indoor positioning process of the wireless sensor network suffers from internal or external attacks, and the wrong positioning results may lead to the failure of network functions and errors in monitoring results, thereby destroying the effectiveness of network applications. Therefore, how to realize the safe positioning of nodes in the application of wireless sensor network with possible hostility is a problem that must be solved.

Contents of the invention

The technical problem to be solved by the present invention is to provide a multi-region indoor safe positioning method based on compressed sensing technology for the security of wireless sensor network indoor positioning, and to eliminate malicious positioning information on the basis of ensuring positioning accuracy, thereby Ensure the security of the positioning environment.

A multi-region indoor safe positioning method based on compressed sensing technology, including an offline stage, a safety monitoring stage and an online stage, the specific steps are as follows:

1) Offline stage

In the set area where the node to be located is located, the anchor nodes are deployed according to an equilateral triangle, and the RSSI security key matrix of the node to be located is constructed according to the distance between the node to be located and the anchor node;

2) Safety monitoring stage

Compare the actual RSSI value received by the node to be located with the corresponding value in the RSSI security key matrix to eliminate malicious anchor nodes;

3) Online stage

The node to be positioned receives the perceived anchor node information in real time, draws a circular area with the perceived anchor node as the center of the circle, and obtains the overlapping area formed by the circular areas corresponding to all anchor nodes;

The grid is selected according to the overlapping area to construct the online measurement matrix, and the online observation matrix is constructed according to the anchor node information perceived by the node to be positioned; the orthogonal pre-operation is performed on the online measurement matrix and the online observation matrix to obtain sparse signals, and the signal recovery algorithm is used to The sparse signal after orthogonalization is restored, the precise position estimation is performed, and the positioning information of the node to be positioned is obtained.

The elements in the online measurement matrix Ψ are in turn the offline Its dimension is in Indicates the number of grid nodes with an association degree of 1;

The degree of association of grid nodes located in the overlapping area is 1;

Indicates the offline RSS average value sampled by the j-th grid node at the i-th anchor node;

Indicates that the j-th grid node receives the τ-th RSS sample value of the i-th anchor node; if the information of the anchor node is not collected, let

The online observation matrix is Φ,

Φ is an M×L-dimensional matrix, each row of Φ is a 1×L vector, all elements satisfy φ _i,j ∈{0,1}, and each row has only one element that is 1, where the element of 1 is The column number indicates which of the L anchor nodes is selected; if φ _{h, g} = 1, h = 1, 2..., M, g = 1, 2..., L, it indicates the hth anchor selected The node is the gth among the L anchor nodes, and the anchor node is the selected anchor node, according to w _i ×RSS _i,j , i=1,2,...,L,j=1,2,... , the value of N, select a non-zero anchor node;

RSS _{i, j} represent the online RSS average value of the j grid node sampled at the i anchor node, w _i represents the weight of the i anchor node, That is, the contribution value of each anchor node in the process of collecting RSS at all grid points, and Numk indicates whether the anchor node is perceived on the kth grid node, that is, when a test sensor node is placed on the grid node, the anchor node Whether the node is within the communication range of the test sensor node, Numk∈{0,1}.

Further, the RSSI security key matrix of the node to be located is Num:

Among them, num _i,j represents the random number sent by the jth anchor node received by the i-th node to be positioned, Indicates the threshold value of the RSSI value RSS _{i, j} sent by the i-th node to be located received by the j-th anchor node, if but The value is The value range of k is 0-2;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is r ₁ ;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is _r2 ;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is r ₃ ;

r ₁ represents the distance between any two adjacent anchor nodes, and r ₃ is the maximum range radius that the node to be positioned can receive RSSI information. When the distance between the node to be positioned and the anchor node exceeds r ₃ , it will be positioned The RSSI value received by the node is set to zero, r ₂ is the average value of r ₁ and r ₃ ;

The value range of i is 1-L, and L represents the number of nodes to be located; the value range of j table is 1-N, and N represents the number of anchor nodes.

Further, the actual RSSI value received by the node to be positioned is compared with the corresponding value in the RSSI security key matrix, and the specific process of removing malicious anchor nodes is as follows:

The i-th node to be positioned receives the random number num _i,j sent by the j-th anchor node, and the perceived RSSI value is RSS _i,j , and the threshold corresponding to RSS _i,j is

Determine whether the random number num _{i, j} satisfies the formula:

If it is satisfied, it is determined that the location information of the jth anchor node perceived by the ith node to be positioned is correct;

If it is not satisfied or the corresponding random number is not received, it is judged that the positioning information sent by the j-th anchor node received by the i-th node to be positioned is malicious, and will not be used in the positioning process of the i-th node to be positioned. The positioning information of the jth anchor node, and delete the positioning information sent by the malicious anchor node.

Further, when drawing a circular area with the perceived anchor node as the center of the circle, the radius of the circle is determined according to the following process:

The threshold obtained according to the RSS _i,j received by the node to be positioned Obtain the radius level r _i,j under the corresponding threshold, and obtain the circular radius of the corresponding anchor node according to the radius level;

like The value is Then the radius level r _i,j corresponds to r ₁ ;

like The value is Then the radius level r _i,j corresponds to r ₂ ;

like The value is Then the radius level r _i,j corresponds to r ₃ .

Beneficial effect

The wireless sensor network indoor safe positioning method based on compressed sensing technology of the present invention utilizes the theory of compressed sensing and includes three stages of offline, safety monitoring and online. By establishing an adaptive random number dictionary, malicious positioning information is eliminated, so as to ensure the security of positioning. Compared with traditional positioning methods, the advantages of this patent are:

1) A multi-area superposition mechanism is designed, and an adaptive random number key algorithm based on the transmission radius of the anchor node is proposed to eliminate malicious positioning information in the indoor positioning environment;

2) A step of preprocessing is carried out in the online stage, and the offline RSS information is weighted to balance the influence of each anchor node;

3) The CS algorithm is introduced to improve the efficiency of positioning;

Using multi-area coverage and CS algorithm to ensure the accuracy of positioning, emphatically introducing a real-time verification mechanism to ensure the security of the positioning environment.

Description of drawings

Figure 1 is a scene diagram of positioning;

Fig. 2 is the flow chart of positioning method of the present invention;

Figure 3 shows the average positioning error of different algorithms in an attacked environment;

Figure 4 shows the positioning errors of different security positioning algorithms under different attacks.

detailed description

The present invention will be described in further detail below in conjunction with accompanying drawing and specific embodiment:

Example 1:

A multi-region indoor safe positioning method based on compressed sensing technology, including an offline stage, a safety monitoring stage and an online stage, the specific steps are as follows:

1. Offline stage

In the set area where the node to be located is located, the anchor nodes are deployed according to an equilateral triangle, and the RSSI security key matrix of the node to be located is constructed according to the distance between the node to be located and the anchor node;

1) Establish offline measurement matrix:

The total number of grid nodes in the positioning area is N, and the number of anchor nodes is L; on each grid, q times of sampling are performed to collect the RSS (received signal strength) values of all anchor nodes sensed by grid points, Obtain an initial L×N dimensional offline measurement matrix

in Indicates the offline RSS average value sampled by the j-th grid node at the i-th anchor node;

Indicates that the grid node j has received the τth RSS sampling value of the anchor node i; if the information of the anchor node is not collected, let

2) Weighted anchor nodes:

Weight matrix W=[w ₁ ,w ₂ ,…,w _L ] ^T , where Indicates the weight of the i-th anchor node, that is, the contribution value of each anchor node in the RSS collection process of all grid nodes, and Numk indicates whether the anchor node is perceived on the k-th grid node, that is, when the grid node When the node places a test sensor node, whether the anchor node is within the communication range of the test sensor node, Numk∈{0,1};

3) Establish offline fingerprint library:

The unbiased estimate of the sampled value of the anchor node i received by the reference node j is expressed as For each reference node j, the unbiased estimation matrix is expressed as Δ _j = [Δ _1,j ,Δ _2,j ,…,Δ _L,j ] ^T ; the offline fingerprint library is expressed as where (x _j ,y _j ) is the coordinate of reference node j.

2. Safety monitoring stage

Compare the actual RSSI value received by the node to be located with the corresponding value in the RSSI security key matrix to eliminate malicious anchor nodes;

The RSSI security key matrix of the node to be positioned is Num:

Among them, num _{i, j} represents the random number sent by the jth anchor node received by the i-th node to be positioned, Indicates the threshold value of the RSSI value RSS _i,j sent by the i-th node to be located received by the j-th anchor node, if but The value is The value range of k is 0-2;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is r ₁ ;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is _r2 ;

Indicates the RSSI threshold received by the node to be positioned when the distance from the anchor node is r ₃ ;

r ₁ represents the distance between any two adjacent anchor nodes, and r ₃ is the maximum range radius that the node to be positioned can receive RSSI information. When the distance between the node to be positioned and the anchor node exceeds r ₃ , it will be positioned The RSSI value received by the node is set to zero, and r ₂ is the average value of r ₁ and r ₃ ;

The value range of i is 1-L, and L represents the number of nodes to be located; the value range of j table is 1-N, and N represents the number of anchor nodes.

The actual RSSI value received by the node to be positioned is compared with the corresponding value in the RSSI security key matrix, and the specific process of removing malicious anchor nodes is as follows:

The i-th node to be positioned receives the random number num _i,j sent by the j-th anchor node, and the perceived RSSI value is RSS _i,j , and the threshold corresponding to RSS _i,j is

Determine whether the random number num _{i, j} satisfies the formula:

If it is satisfied, it is determined that the location information of the jth anchor node perceived by the ith node to be positioned is correct;

If it is not satisfied or the corresponding random number is not received, it is judged that the positioning information sent by the j-th anchor node received by the i-th node to be positioned is malicious, and will not be used in the positioning process of the i-th node to be positioned. The positioning information of the jth anchor node, and delete the positioning information sent by the malicious anchor node.

3. Online stage

The node to be positioned receives the perceived anchor node information in real time, draws a circular area with the perceived anchor node as the center of the circle, and obtains the overlapping area formed by the circular areas corresponding to all anchor nodes;

The grid is selected according to the overlapping area to construct the online measurement matrix, and the online observation matrix is constructed according to the anchor node information perceived by the node to be positioned; the orthogonal pre-operation is performed on the online measurement matrix and the online observation matrix to obtain sparse signals, and the signal recovery algorithm is used to The sparse signal after orthogonalization is restored, the precise position estimation is performed, and the positioning information of the node to be positioned is obtained.

The elements in the online measurement matrix Ψ are in turn the offline Its dimension is in Indicates the number of grid nodes with an association degree of 1;

The degree of association of grid nodes located in the overlapping area is 1;

Indicates the offline RSS average value sampled by the j-th grid node at the i-th anchor node;

Indicates that the j-th grid node receives the τ-th RSS sample value of the i-th anchor node; if the information of the anchor node is not collected, let

The online observation matrix is Φ,

Φ is an M×L-dimensional matrix, each row of Φ is a 1×L vector, all elements satisfy φ _i,j ∈{0,1}, and each row has only one element that is 1, where the element of 1 is The column number indicates which of the L anchor nodes is selected; if φ _h , _g ＝1, h＝1,2...,M, g＝1,2...,L, it indicates the hth anchor selected The node is the gth among the L anchor nodes, and the anchor node is the selected anchor node, according to w _i ×RSS _i,j , i=1,2,...,L,j=1,2,... , the value of N, select a non-zero anchor node;

RSS _{i, j} represent the online RSS average value of the j grid node sampled at the i anchor node, w _i represents the weight of the i anchor node, That is, the contribution value of each anchor node in the process of collecting RSS at all grid points, and Numk indicates whether the anchor node is perceived on the kth grid node, that is, when a test sensor node is placed on the grid node, the anchor node Whether the node is within the communication range of the test sensor node, Numk∈{0,1}.

Firstly, the orthogonalization pre-operation is performed on Φ and Ψ, and then the signal recovery algorithm is used to perform the sparse signal Restore; the signal recovery is processed by the l ₁ -minimum paradigm in the convex optimization algorithm, and the restored sparse signal is obtained

pass Weighted with the position coordinates of the node, the position of the node to be positioned is estimated more accurately, the expression is:

Among them, px and py respectively represent the abscissa and ordinate corresponding to the final position of the node to be positioned; x _k , y _k and Represent the x-coordinate and y-coordinate of the kth grid and the probability that the node to be located appears in the grid, respectively.

In the steps of orthogonalization pre-operation and signal recovery, assuming that T is the orthogonalization pre-operation of the measured value y, y is orthogonalized to obtain the measured value after orthogonalization y′=Ty; definition P=ΦΨ, Q=orth(P ^T ) ^T , orth(A) represents the normalized orthogonalization operation on the matrix P, is the pseudo-inverse matrix of P, then the localization problem is described as the following l ₁ -minimum normal form model:

Among them, θ is a vector representing a set of positions where the node to be located may appear, Indicates a specific value of θ that satisfies the condition of the equation for a specific error value ε′, Z is the target matrix function used in the calculation, Z=y, and y is the measurement vector, which refers to the perception collected by the node to be positioned in the online phase The RSS values of all the anchor nodes arrived, ε' is the set error, and the value is in the interval [0,0.1]; calculate the inequality z-Qθ≤ε'.

Building an incidence matrix can be seen as a clustering problem. A multi-region overlay algorithm is designed to solve this problem. The idea of the multi-area overlay algorithm is to use the overlay technology to narrow the positioning area. The region overlay algorithm has many advantages in solving clustering problems. Area overlays are non-iterative. In this way, the positioning speed is faster than the general clustering algorithm; the area superposition algorithm is easy to implement. The RSS information obtained dynamically in the online stage can better reflect the current environmental conditions than the offline database, so the positioning is more accurate; the area superposition algorithm has a better performance in shrinking the positioning area in the online stage. Therefore, this algorithm is more efficient than traditional clustering algorithms.

In clustering thinking, finding the smallest and most accurate localization area is the most critical issue. The region overlay algorithm uses the overlap mechanism to obtain the optimal region. The basic idea of overlapping is to superimpose all the communication radius areas of the candidate anchor nodes in the communication area. Candidate nodes are selected through the optimal function. In this algorithm, candidate nodes have higher online RSS value and offline weight.

According to the received online RSSI matrix, the labels of the anchor nodes with non-zero RSSI values are obtained, and the corresponding transmission radius matrix is fused according to these labels to obtain the position estimation value of the node to be located. Finally, the area is superimposed and covered according to the obtained radius value. Then the sparse matrix is generated by orthogonalization, and then the sparse signal is recovered by using compressed sensing and L1-minimum paradigm, and the coordinates of the nodes with positioning are obtained.

FIG. 1 is a diagram of an application scenario of the present invention. A total of 5 nodes to be located are deployed in the positioning area on the field, indicated by red dots. The 25 anchor nodes are represented by blue triangles. 4 malicious nodes, represented by yellow triangles. Each node to be positioned perceives the surrounding anchor node information, and downloads the relevant information of the anchor node from the database. Each anchor node uses relevant information and positioning methods to determine its own position. Malicious nodes mislead positioning by sending wrong RSS information.

Fig. 2 is a flowchart of the positioning method of the present invention. Positioning is divided into offline, safety monitoring and online phases. The initial database is built in the offline phase. Then weight the anchor nodes perceived from the current stage, so that each anchor node reaches a balance in the positioning process. Then establish a security key, set the sending power of the anchor node, and give random numbers in different ranges according to the sending power. Anchor nodes send random numbers while sending RSS information. Each node to be positioned can receive a different random number as a key for security verification in the online price segment. At this stage, the corresponding radius value is given according to the random number set received by the node to be positioned. Verify that the random number is equal to the received RSS information. Eliminate malicious positioning information. Finally, the area is superimposed and covered according to the obtained radius value. Then the sparse matrix is generated by orthogonalization, and then the sparse signal is recovered by using compressed sensing and L1-minimum paradigm, and the coordinates of the nodes with positioning are obtained.

Figure 3 shows the positioning errors of several algorithms when they are under attack. The abscissa represents the number of malicious anchor nodes, and the ordinate represents the average positioning error. The CS_NSL algorithm is a single-key verification algorithm. The anchor node broadcasts the same key to all positioning areas while sending positioning information. When the key received by the node to be located is not the initial key, it proves that the positioning information is malicious. CS_SL is a multi-area grid indoor positioning algorithm based on compressed sensing. It uses grids to divide the area into small areas one by one to improve the accuracy of positioning. At the same time, the method of compressed sensing is used to select anchor nodes with relatively strong activity for positioning, and the interference of some malicious nodes is eliminated. Among them, when the number of malicious anchor nodes gradually increases, the positioning error of CS_SL increases sharply, and when the number of malicious anchor nodes exceeds two, the positioning error of CS_NSL also increases significantly. However, the positioning error of the multi-region key verification algorithm proposed in this patent hardly changes. Because the verification algorithm of the present invention is dynamically verified in real time, almost all malicious positioning information can be verified and eliminated.

Fig. 4 is the average positioning error of the CS_NSL algorithm and the algorithm of the present invention when subjected to external attacks and internal attacks, wherein Hongfan attack and forgery attack are selected as representatives of internal and external attacks to verify the security of the algorithm. In the figure, '-*-' represents the positioning method (CSMR_SL) of the present invention, which is compared with other two algorithms CS_SL and CS_NSL. Both attack results show that the algorithm described in the present invention has a better ability to defend against attacks. External attacks cannot obtain the key, but can continuously send wrong keys to guess the security key. Internal attacks can destroy the positioning process by capturing anchor nodes and sending wrong positioning information. When the key is relatively single, an external attack can guess the key of the algorithm by continuously trying to send the key, making the security algorithm invalid. However, the key based on the transmission radius value changes in real time, making it almost impossible to guess. In the internal attack, according to the key generation algorithm. If the received random number key does not exceed the radius threshold used for key generation, the location information is normal. Compared with the traditional single-key algorithm, the method of the invention divides the previous coverage range into several sub-ranges and performs one-by-one verification, which greatly improves the effectiveness of the algorithm.

Claims (4)

1. a kind of multizone indoor security localization method based on compressed sensing technology, it is characterised in that including off-line phase, peace Full monitoring stage and on-line stage, comprise the following steps that：

1) off-line phase

In the setting regions of node place to be positioned, anchor node is disposed according to equilateral triangle, and according to nodal distance to be positioned The distance between anchor node, builds the RSSI safe key matrix of node to be positioned；

2) the safety monitoring stage

It is compared with the respective value in RSSI safe key matrix using the actual RSSI value of node to be positioned reception, rejects and dislike Meaning anchor node；

3) on-line stage

The anchor node information that node real-time reception to be positioned is perceived, the anchor node for having perceived is drawn border circular areas for the center of circle, Obtain the overlapping region that the corresponding border circular areas of all anchor nodes are formed；

Grid is chosen according to overlapping region to build on-line measurement matrix, according to the anchor node information structure that node perceived to be positioned is arrived It build line observing matrix in；Pre-operation being orthogonalized to on-line measurement matrix and online observation matrix and obtains sparse signal, utilizes Signal recovery algorithms are reduced to the sparse signal after orthogonalization, carry out accurate position estimation, obtain node to be positioned Location information.

2. method according to claim 1, it is characterised in that the RSSI safe key matrix of the node to be positioned is Num：

Wherein, num_i,jRepresent the random number that j-th anchor node that i-th node to be positioned is received is sent out, Represent that i-th node to be positioned receives the RSSI value that j-th anchor node sends RSS_i,jThreshold value, ifThenValue beThe span of k is 0-2；

Represent that apart from anchor node be r₁When, the RSSI threshold value that node to be positioned is received；

Represent that apart from anchor node be r₂When, the RSSI threshold value that node to be positioned is received；

Represent that apart from anchor node be r₃When, the RSSI threshold value that node to be positioned is received；

r₁Represent the distance between adjacent anchor node of any two, r₃The maximum of RSSI information can be received for node to be positioned Scope radius, when node to be positioned has exceeded r with the distance of anchor node₃The RSSI value for just receiving node to be positioned is put Zero, r₂For r₁And r₃Meansigma methodss；

The span of i represents the number of node to be positioned for 1-L, L；The span of j table represents the individual of anchor node for 1-N, N Number.

3. method according to claim 2, it is characterised in that the actual RSSI value for being received using node to be positioned with Respective value in RSSI safe key matrix is compared, and the detailed process for rejecting malice anchor node is as follows：

It is num that i-th node to be positioned receives the random number that j-th anchor node send_i,j, and the RSSI value for perceiving is RSS_i,j, RSS_i,jCorresponding threshold value is

Judge random number num_i,jWhether formula is met：

If it is satisfied, then judge i-th node perceived to be positioned to the location information of j-th anchor node be correct；

If being unsatisfactory for or corresponding random number not being received, j-th anchor section that i-th node to be positioned is received is judged The location information of transmission is malice, is not used the positioning of j-th anchor node in the position fixing process of i-th node to be positioned Information, deletes the location information that the malice anchor node sends.

4. method according to claim 3, it is characterised in that described is that circle is drawn in the center of circle by the anchor node for having perceived During domain, circular radius are determined according to procedure below：

According to the RSS that node to be positioned is received_i,jThe threshold value of acquirementObtain radius grade r under corresponding threshold value_i,j, according to The circular radius of corresponding anchor node are obtained according to radius grade；

IfValue isThen radius grade r_i,jCorresponding radius is r₁；

IfValue isThen radius grade r_i,jCorresponding radius is r₂；

IfValue isThen radius grade r_i,jCorresponding radius is r₃.