CN106332173A - Distributed type node drift detection method and device - Google Patents

Abstract

The invention relates to a distributed type node drift detection method and a distributed type node drift detection device and belongs to the technical field of wireless sensor network. The distributed type node drift detection method includes: firstly, collecting RSSI (received signal strength indicator) data of communication of beacon nodes and the rest beacon nodes according to set time intervals; calculating RSSI variation of the neighboring sampled beacon nodes to acquire a time sequence G of the RSSI variation of the beacon nodes and then performing linear fitting on the time sequence G of the RSSI of the beacon nodes to acquire a fitted regression line of the timing sequence of the RSSI variation; finally, judging whether difference value between the fitted value on the fitted regression line of the timing sequence of the RSSI variation of the beacon nodes is greater than the set threshold and corresponding actual value is greater than a set threshold or not, if yes, determining that the corresponding beacon nodes shift. The beacon nodes possibly shift can be judged out automatically by calculating the RSSI variation at different time, simpleness is achieved, and high practicability is realized.

Description

A kind of distributed node drift detection method and device

Technical field

The present invention relates to a kind of distributed node drift detection method and device, belong to wireless sensor network technology neck Territory.

Background technology

Wireless sensor network be deployed in monitored area in the sensor node of substantial amounts of static or movement with self-organizing The network system constituted with the mode of multi-hop, cooperates ground perception, gathers and process network's coverage area between sensor node The information of middle monitoring object, and it is sent to observer.Wireless sensor network need not any fixing network support, has quickly The features such as expansion, strong, the operational lifetime length of survivability, have widely in complicated monitoring on a large scale and tracking task field Application prospect.

In actual applications, owing to the interstitial content of wireless sensor network is the hugest, and node is the most random Dispose, it is difficult to measure the position of each node one by one when disposing.And the position of node for monitoring information acquisition to close weight Want, main reason is that following 2 points: the most whether node location information is directly connected to the effectiveness of gathered data；Based on The premise that discovery, maintenance and the data of geographic routing protocol realization route forward is to obtain node location information.Obtain joint The direct method of some position is to use global positioning system (Global positioning system, GPS), but is affected by into The restriction of the factors such as basis, volume, power consumption, layout environment, actual applies each node all to configure gps receiver unrealistic, Therefore the research to node localization in wireless sensor networks becomes necessity.

According to location mechanism, Wireless Sensor Network Located Algorithm can be generally divided into location based on non-ranging technology and calculates Method and location algorithm based on range finding.Wherein location algorithm based on non-ranging technology is estimated according only to network-in-dialing relational implementation The fuzzy location of formula, little by such environmental effects, but positioning precision is relatively low, and higher to the density requirements of anchor node.And based on The location algorithm of range finding needs to measure the actual range between adjacent node or orientation calculates unknown node position, and positioning precision is relatively High.Therefore, generally use location algorithm based on range finding in the application scenario higher to node location precise requirements, and it is fixed Position precision depends greatly on the distance estimations between beaconing nodes and unknown node.

In traditional static wireless sensor network, beaconing nodes, as the basis of location, generally assumes that all beacons Node is all that transfixion and positioning performance keep stable, yet with there is various uncertain natural, anthropic factor or evil The seat offence of meaning, beaconing nodes is likely to occur unexpected mobile or positioning performance big ups and downs in actual applications, is referred to as For " drift ".For this type of situation, by periodic mode beaconing nodes reorientated and just can revise the location that drift causes Deviation.Beaconing nodes in actual monitoring region generally uses the method for pre-setting, and after deployment completes, beaconing nodes is to convergence Node uploads the packet comprising own node ID, position, and with unknown node communication.The most once beaconing nodes is sent out Raw drift, repositioning process will make position deviation spread further, affect the service quality of whole network.Therefore, research letter Mark node drift test problems, has very highland theory value and using value.

For node drifting problem test problems, Kuo etc. proposes beacon and moves detection algorithm (Beacon Movement Detection, BMD) for identifying that the position in network occurs the passive beaconing nodes changed, its basic thought is, at network Middle a BMD engine is set to collect RSSI (the received signal strength indication) information of whole network And process.The method can interpolate that out the movement of beaconing nodes in certain error tolerance.BMD model is substantially one The centralized algorithm solving np complete problem, and there is arithmetic speed and computing knot in the heuritic approach solving np complete problem The really contradiction between degree of accuracy.Ravi Garg etc. use to get rid of and provide bigger decline ladder during node location calculates The beaconing nodes of degree, improves location credibility, but does not accounts for the reference by location effect of ordinary node, be not suitable for beacon dilute The network dredged, and there is the problem that amount of calculation is bigger.

Summary of the invention

It is an object of the invention to provide a kind of distributed node drift detection method, detected solving the drift of current node The problem that journey operand is big.Present invention also offers a kind of distributed node shift testing apparatus simultaneously.

The present invention solves that above-mentioned technical problem provides a kind of distributed node drift detection method, this detection method bag Include following steps:

1) gather each beaconing nodes at set time intervals to communicate with remaining beaconing nodes RSSI data (S_k,t_i), (S_k,t_i) represent at t_i(i=1,2 ..., m) communicate with remaining beaconing nodes collection of RSSI data of moment beaconing nodes k is combined into S_k, K=1,2 ..., n, n are the sum of beaconing nodes, and m is sampling number；

2) neighbouring sample moment each beaconing nodes RSSI change degree is calculated, during to obtain the RSSI change degree of each beaconing nodes Between sequence G；

3) RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtain each beaconing nodes corresponding RSSI change degree time series fitted regression line；

4) judge match value on each beaconing nodes RSSI change degree time series fitted regression line and corresponding actual value it Between difference whether more than setting threshold value, if more than setting threshold value, being then that the beaconing nodes that its correspondence is described drifts about.

Further, described step 2) in neighbouring sample moment each beaconing nodes RSSI change degree be:

$G\[(S_k,t_i),(S_k,t_{i+1})\]=\underset{j=1}{\overset{n}{\Σ}}\left(1-\frac{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|}{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|+a_{kj}}\right)/n$

(S_k,t_i)_jFor t_iMoment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension, (S_k, t_i+1)_jFor t_i+1Moment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension；a_kjRepresent in jth dimension (S_k,t_i) and (S_k,t_i+1) absolute value of meansigma methods；G ∈ [0,1].

Further, described step 3) in obtain fitted regression line equation and be:

D '=β₀+β₁t

${\mathrm{\β}}_1=\frac{\underset{i=1}{\overset{n}{\Σ}}(t_i-\stackrel{\‾}{t})(d_i-\stackrel{\‾}{d})}{\underset{i=1}{\overset{n}{\Σ}}{(t_i-\stackrel{\‾}{t})}^2}$

${\mathrm{\β}}_0=\stackrel{\‾}{d}-{\mathrm{\β}}_1\stackrel{\‾}{t}$

Wherein t represents sampling time point, β₀Represent the intercept of this straight line, β₁Represent the slope of this straight line, when d ' represents sampling Between some match value corresponding to t；The meansigma methods of express time section,In express time section, the RSSI change degree of beaconing nodes is put down Average.

Further, described step 4) in set threshold value as 0.51.

Further, the method is additionally included in after every minor node differentiates and drift beaconing nodes is carried out location updating, and Directly regard it as unknown node to reorientate, and will estimate that position is as its new self-position.

Present invention also offers a kind of distributed node shift testing apparatus, this device include acquisition module, computing module, Fitting module and judge module,

Described acquisition module is for gathering each beaconing nodes and remaining beaconing nodes communication at set time intervals RSSI data (S_k,t_i), (S_k,t_i) represent at t_i(i=1,2 ..., m) moment beaconing nodes k communicates with remaining beaconing nodes RSSI The collection of data is combined into S_k, k=1,2 ..., n, n are the sum of beaconing nodes, and m is sampling number；

Described computing module is used for calculating neighbouring sample moment each beaconing nodes RSSI change degree, to obtain each beacon joint RSSI change degree time series G of point；

Described fitting module, for RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtains The RSSI change degree time series fitted regression line that each beaconing nodes is corresponding；

Described judge module is for judging the matching on each beaconing nodes RSSI change degree time series fitted regression line Whether the difference between value and corresponding actual value is more than setting threshold value, if more than setting threshold value, being then the beacon that its correspondence is described Node drifts about.

Further, described computing module calculated neighbouring sample moment each beaconing nodes RSSI change degree is:

$G\[(S_k,t_i),(S_k,t_{i+1})\]=\underset{j=1}{\overset{n}{\Σ}}\left(1-\frac{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|}{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|+a_{kj}}\right)/n$

(S_k,t_i)_jFor t_iMoment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension, (S_k, t_i+1)_jFor t_i+1Moment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension；a_kjRepresent in jth dimension (S_k,t_i) and (S_k,t_i+1) absolute value of meansigma methods；G∈[0,1].

Further, the fitted regression line equation that described fitting module uses is:

D '=β₀+β₁t

${\mathrm{\β}}_1=\frac{\underset{i=1}{\overset{n}{\Σ}}(t_i-\stackrel{\‾}{t})(d_i-\stackrel{\‾}{d})}{\underset{i=1}{\overset{n}{\Σ}}{(t_i-\stackrel{\‾}{t})}^2}$

${\mathrm{\β}}_0=\stackrel{\‾}{d}-{\mathrm{\β}}_1\stackrel{\‾}{t}$

Wherein t represents sampling time point, β₀Represent the intercept of this straight line, β₁Represent the slope of this straight line, when d ' represents sampling Between some match value corresponding to t；The meansigma methods of express time section,In express time section, the RSSI change degree of beaconing nodes is put down Average.

Further, described judge module sets threshold value as 0.51.

Further, this detection device also includes, after differentiating at every minor node, drift beaconing nodes is carried out position Update, and directly regard it as the module that unknown node carries out reorientating.

The invention has the beneficial effects as follows: the time interval that the present invention is first according to set gathers each beaconing nodes and believes with remaining Mark node communication RSSI data；Then neighbouring sample moment each beaconing nodes RSSI change degree is calculated, to obtain each beaconing nodes RSSI change degree time series G, and RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtains each The RSSI change degree time series fitted regression line that beaconing nodes is corresponding；Finally judge each beaconing nodes RSSI change degree time sequence Whether the difference between match value and corresponding actual value on row fitted regression line is more than setting threshold value, if more than setting threshold value, It is then that the beaconing nodes that its correspondence is described drifts about.The present invention sentences automatically by calculating the intensity of variation of RSSI the most in the same time Not may have occurred the beaconing nodes of drift, simple, there is higher practicality.

Accompanying drawing explanation

Fig. 1 is wireless sensor network modular concept figure；

Fig. 2 is beaconing nodes Drift Process figure；

Fig. 3-a is that threshold value affects schematic diagram to beaconing nodes drift detection method success rate；

Fig. 3-b is that threshold value affects schematic diagram to beaconing nodes drift detection method False Rate；

Fig. 4 is that element number affects schematic diagram to what beaconing nodes drift differentiated.

Detailed description of the invention

Below in conjunction with the accompanying drawings the detailed description of the invention of the present invention is described further.

The embodiment of the present invention a kind of distributed node drift detection method

Wireless signal is a kind of electromagnetic wave signal, it is considered to the received signal strength value (Received of isotropic spherical wave Signal strength indicator), unit is dBm, and it can be transmitted Absorption of Medium portion of energy in communication process, Intensity can decay with distance exponentially.Conventional radio signal propagation path loss model has: free space propagation model, two-wire Ground return model and log-distance path loss model model, wherein log-distance path loss model model is most widely used.Right Number distance path loss model is made up of two parts, and first is pass loss model, and this model can be predicted when distance is d Time received signal power, be expressed asUse connects paracentral distance d₀As reference, P_r(d₀) it is to be d at reference distance₀ The reception power at place, can measure acquisition or known.Relative to P_r(d₀) be calculated as follows:

$\frac{P_r\left(d_0\right)}{\stackrel{\‾}{P_r\left(d\right)}}={\left(\frac{d}{d_0}\right)}^{\mathrm{\β}}---\left(1\right)$

Wherein, β is path loss index (pass loss index), the empirical value generally got by reality measurement, reflects road The speed that footpath loss increases with distance.β depends primarily on the environment of radio signal propagation, the most aerial decay, reflection, many The complex jamming such as footpath effect.Pass loss model is generally using dB as measurement unit, and its expression formula is:

$\frac{\stackrel{\‾}{P_r\left(d\right)}}{P_r\left(d_0\right)}=-10\mathrm{\β}log\left(\frac{d}{d_0}\right)---\left(2\right)$

The Part II of log-distance path loss model model is the stochastic variable X meeting Gauss distribution_dB(0,σ²), reflect When distance one timing, due to noise jamming, cause the change receiving power.Therefore, log-distance path loss model model expression For

$\stackrel{\‾}{P_r\left(d\right)}=P_r\left(d_0\right)-10\mathrm{\β}log\left(\frac{d}{d_0}\right)+X_{dB}---\left(3\right)$

Now RSSI meets with actual value for expectation, and σ is the normal distribution of standard deviation.

It is specifically described as a example by a certain concrete wireless sensor network model below, the network of this wireless senser Model is as it is shown in figure 1, include the random placement one group of wireless sensor node S={S in 3D region (a × b × c)_i| i= 1,2 ..., M}, each node is isomorphism node, its Information Communication scope be one centered by self physical location, R is as radius Circle, i.e. the communication radius of node is the R diagonal L of region (R more than).All nodes are divided by its function in alignment system For beaconing nodes and unknown node.Front n node S₁(x₁,y₁)、S₂(x₂,y₂)、…、S_n(x_n,y_n) can be outside by GPS etc. Equipment or the actual arrangement known obtain self-position, in advance as beaconing nodes；Node S_i(x_i,y_i) (n < i≤M) at network Middle Location-Unknown and itself do not have special hardware device can obtain self information, as unknown node.The most each biography Sensor node has unique ID；Spacing wireless signal mode is preferable spheroid；All the sensors node isomorphism, electricity Identical with computing capability；All nodes are time synchronized, and can directly communicate.

In network, the process of beaconing nodes drift is as in figure 2 it is shown, after a period of time of location, beaconing nodes A there occurs drift, Neighborhood between node there occurs change the most therewith, but the information of beaconing nodes A ' broadcast location does not change.One The reliability of individual beaconing nodes positional information is described with the RSSI intensity of variation of remaining beaconing nodes with it, when change journey Spend the biggest, then it is the most violent with the relative motion of remaining beaconing nodes, more may have occurred drift.After network design completes, Each beaconing nodes is in communication with each other each other, according to node differentiate mechanism to self pass judgment on to weigh that there is drift in it can Energy property, if deviation is less than threshold value, is then labeled as the beaconing nodes that drifts about, is otherwise labeled as the beaconing nodes that drifts about.The present invention uses Beaconing nodes drift detection method based on RSSI change degree, the discrimination standard of beaconing nodes current location reliability rely on it with The RSSI of remaining beaconing nodes change degree within a period of time, change degree is the lowest, and the RSSI beacon joint without notable change is described Point is the most, and its reliability is the highest, otherwise change degree is the highest, and its reliability is the lowest.The detailed process of the method is as follows:

1. the time interval being first according to set gathers each beaconing nodes and communicates with remaining beaconing nodes RSSI data, t₁Time Carve the beaconing nodes in wireless sensor network and press a series of RSSI numbers that communicate with remaining beaconing nodes of ID number order sampling According to set, be designated as (S_k,t₁)=[r₁,r₂,…,r_n].Wherein n is beaconing nodes sum, corresponding to beaconing nodes self ID RSSI is designated as 1；(S_k,t₁),(S_k,,t₂),…,(S_k,,t_m) it is the beaconing nodes k collection with the RSSI data of Fixed Time Interval Close, (S_k,t_i), (S_k,t_i) represent at t_i(i=1,2 ..., m) moment beaconing nodes k communicates with remaining beaconing nodes RSSI data Collection be combined into S_k, k=1,2 ..., n, n are the sum of beaconing nodes, and m is sampling number.

2. calculate neighbouring sample moment each beaconing nodes RSSI change degree, during to obtain the RSSI change degree of each beaconing nodes Between sequence G.

The neighbouring sample moment computing formula of each beaconing nodes RSSI change degree is:

$G\&lsqb;(S_k,t_i),(S_k,t_{i+1})\&rsqb;=\underset{j=1}{\overset{n}{\&Sigma;}}\left(1-\frac{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|}{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|+a_{kj}}\right)/n---\left(4\right)$

(S_k,t_i)_jFor t_iMoment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension, (S_k, t_i+1)_jFor t_i+1Moment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension；a_kjRepresent in jth dimension (S_k,t_i) and (S_k,t_i+1) absolute value of meansigma methods；G∈[0,1].

3. RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtain each beaconing nodes corresponding RSSI change degree time series fitted regression line.

In wireless sensor network RSSI change degree time series G of beaconing nodes be considered as one with the sampling time between It is independent variable every t, changes the angle value G function as dependent variable with RSSI.If the RSSI change angle value of each time point is with linear gauge Rule is distributed in around straight line, and this straightway can be calculated by Linear Regression Model in One Unknown method, then be referred to as The RSSI change degree seasonal effect in time series unitary linear fit regression line.

Consider that regression function is the linear function of t, thus the unitary linear fit regression equation that the present invention uses be:

D '=β₀+β₁t (5)

${\mathrm{\&beta;}}_1=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}(t_i-\stackrel{\&OverBar;}{t})(d_i-\stackrel{\&OverBar;}{d})}{\underset{i=1}{\overset{n}{\&Sigma;}}{(t_i-\stackrel{\&OverBar;}{t})}^2}---\left(6\right)$

${\mathrm{\&beta;}}_0=\stackrel{\&OverBar;}{d}-{\mathrm{\&beta;}}_1\stackrel{\&OverBar;}{t}---\left(7\right)$

Wherein, t represents sampling time point, β₀Represent the intercept of this straight line, β₁Representing the slope of this straight line, d ' represents sampling Match value corresponding to time point t.The meansigma methods of express time section,The RSSI change degree of beaconing nodes in express time section Meansigma methods.Pass through β₁The size of value, it can be determined that the severe degree of data variation.

4. judge match value on each beaconing nodes RSSI change degree time series fitted regression line and corresponding actual value it Between difference whether more than setting threshold value, if more than setting threshold value, being then that the beaconing nodes that its correspondence is described drifts about.

Judge the match value on each beaconing nodes RSSI change degree time series fitted regression line and between corresponding actual value Difference whether more than setting threshold value, if more than setting threshold value, being then that the beaconing nodes that its correspondence is described drifts about, can floating Move beaconing nodes to get rid of from the beacon set participating in location.

As time goes on, wireless sensor network occurs drift beaconing nodes may get more and more, and can use Anchor node number can be fewer and feweri, the positioning precision of unknown node will be directly affected, for preventing the generation of this situation, often After minor node differentiates, reply drift beaconing nodes carries out location updating, can directly regard it as unknown node and carry out again determining Position, and using the position reorientated as its new self-position, for not occurring the beaconing nodes of drift to provide reference.Again fixed Position can use relatively common location mode, such as, triangle polyester fibre mode etc..

A kind of embodiment of the distributed node shift testing apparatus of the present invention

In the present embodiment, distributed node shift testing apparatus includes acquisition module, computing module, fitting module and judgement Module, acquisition module communicates with remaining beaconing nodes RSSI data for gathering each beaconing nodes at set time intervals (S_k,t_i), (S_k,t_i) represent at t_i(i=1,2 ..., m) moment beaconing nodes k communicates with remaining beaconing nodes the collection of RSSI data It is combined into S_k, k=1,2 ..., n, n are the sum of beaconing nodes, and m is sampling number；Computing module is used for calculating the neighbouring sample moment Each beaconing nodes RSSI change degree, to obtain RSSI change degree time series G of each beaconing nodes；Fitting module is for by each letter RSSI change degree time series G of mark node carries out linear fit, obtains the RSSI change degree time sequence that each beaconing nodes is corresponding Row fitted regression line；Judge module is for judging the match value on each beaconing nodes RSSI change degree time series fitted regression line With whether the difference between correspondence actual value is more than setting threshold value, if more than setting threshold value, then it is that the beacon that its correspondence is described saves Point drifts about.

In order to check the performance of the present invention, carrying out l-G simulation test below, simulating scenes sets as follows:

1) test 3D region is 100m × 100m × 50m；

2) node total number is 100, and the drift distance of beaconing nodes maximum is 20m；

3) unknown node and the transmitting signal intensity P of beaconing nodes_tFor 30dBm, reference distance d₀For 20m, launch antenna and increase Benefit G_t, receiving antenna gain G_rIt is 2 for 1dBi, path loss index n.

2 performance indications arranging measurement beaconing nodes drift differentiation are: success rate (Num (B_M∩B_MD)/Num(B_M)) and False Rate (Num ((U-B_M)∩B_MD)/Num(B_M)).Wherein, B_MRepresent the beaconing nodes set actually occurring drift, B_MDRepresent Being determined as the beaconing nodes set of drift, U represents the set of all beaconing nodes.Success rate be by correct decision for drift beacon The ratio of the number of node and actual drift beaconing nodes number, error rate be mistake be determined as drifting about beaconing nodes number with The ratio of actual drift interstitial content.

The standard weighing location algorithm accuracy is position error, and position error is defined as the located algorithm of unknown node Distance between estimation coordinate position and its real coordinate positionWherein estimate seat Mark is set to (x_e,y_e,z_e), real coordinate position is (x_i,y_i,z_i), during emulation, the location algorithm used is that triangle polyester fibre is calculated Method.

Differentiating whether a certain beaconing nodes drifts about is by analyzing its RSSI with remaining beaconing nodes when one section Interior degree of approximation solves, and between match value and the actual value on only RSSI change degree time series fitted regression line Difference more than threshold value, just judge that it there occurs drift, so it is most important to choose rational threshold value.By test of many times, threshold The scope of value, about 0.5, is chosen at the value near 0.5 and emulates, and the element number of RSSI data acquisition system is 10.From Fig. 3- It can be seen that along with the reduction of threshold value, the success rate of beaconing nodes drift distinguished number rises constantly in a and Fig. 3-b, but False Rate also increases constantly simultaneously.Threshold value is the least, and the change to RSSI is the most sensitive, easily by beaconing nodes less for drift Erroneous judgement, when threshold value is 0.51, success rate is 91.09%, and False Rate is 10.51%, and resultant effect is good, therefore, chooses The threshold epsilon of algorithm is 0.51.

The suitably element number n of RSSI data acquisition system contributes to shortening Riming time of algorithm, improves the accuracy differentiated. It is sensitive to the response fluctuation of RSSI change angle value that too small or excessive element number often leads to beaconing nodes drift distinguished number Or slow, increase the possibility of erroneous judgement, and reduce the robustness of algorithm.Element number is set from 5 to 15 consecutive variations, by becoming The comparison of power and False Rate reflects its impact on algorithm performance.Figure 4, it is seen that along with element number (sampled point Number) increase, the success rate of algorithm the most constantly rises, and False Rate constantly declines simultaneously, if but element number continue increase Greatly, the success rate of algorithm begins to decline, and False Rate also begins to rise simultaneously.Too small or excessive element number is all easily caused by mistake Sentence, it can be seen that choose the element number n of RSSI data acquisition system when being 9, show the most stable.Taking threshold epsilon is 0.51, RSSI The element number n of data acquisition system is 9, and the number of drift beaconing nodes is from 0 to 20, in the case of other conditions are constant, by this The beaconing nodes drift method of discrimination of bright proposition compares with BMD algorithm, and the success rate average obtaining 2 kinds of algorithms is respectively 92.23%, 80.05%, the meansigma methods of False Rate is respectively 9.95%, 12.47%, show the present invention than BMD algorithm in success Preferably performance is had in rate and False Rate.

Claims (10)

1. a distributed node drift detection method, it is characterised in that this detection method comprises the following steps:

1) gather each beaconing nodes at set time intervals to communicate with remaining beaconing nodes RSSI data (S_k,t_i), (S_k,t_i) Represent at t_i(i=1,2 ..., m) communicate with remaining beaconing nodes collection of RSSI data of moment beaconing nodes k is combined into S_k, k=1, 2 ..., n, n are the sum of beaconing nodes, and m is sampling number；

2) neighbouring sample moment each beaconing nodes RSSI change degree is calculated, to obtain the RSSI change degree time sequence of each beaconing nodes Row G；

3) RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtain the RSSI that each beaconing nodes is corresponding Change degree time series fitted regression line；

4) match value on each beaconing nodes RSSI change degree time series fitted regression line is judged and between corresponding actual value Whether difference is more than setting threshold value, if more than setting threshold value, being then that the beaconing nodes that its correspondence is described drifts about.

Distributed node drift detection method the most according to claim 1, it is characterised in that described step 2) in adjacent adopt Sample moment each beaconing nodes RSSI change degree is:

$G\&lsqb;(S_k,t_i),(S_k,t_{i+1})\&rsqb;=\underset{j=1}{\overset{n}{\&Sigma;}}\left(1-\frac{|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j|}{\left|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j\right|+a_{kj}}\right)/n$

(S_k,t_i)_jFor t_iMoment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension, (S_k,t_i+1)_jFor t_i+1Moment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension；a_kjRepresent the upper (S of jth dimension_k,t_i) and (S_k,t_i+1) absolute value of meansigma methods；G∈[0,1].

Distributed node drift detection method the most according to claim 1, it is characterised in that described step 3) in intended Conjunction regression line equation is:

D '=β₀+β₁t

${\mathrm{\&beta;}}_1=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}(t_i-\stackrel{\&OverBar;}{t})(d_i-\stackrel{\&OverBar;}{d})}{\underset{i=1}{\overset{n}{\&Sigma;}}{(t_i-\stackrel{\&OverBar;}{t})}^2}$

${\mathrm{\&beta;}}_0=\stackrel{\&OverBar;}{d}-{\mathrm{\&beta;}}_1\stackrel{\&OverBar;}{t}$

Wherein t represents sampling time point, β₀Represent the intercept of this straight line, β₁Representing the slope of this straight line, d ' represents sampling time point Match value corresponding to t；The meansigma methods of express time section,The RSSI change degree meansigma methods of beaconing nodes in express time section.

Distributed node drift detection method the most according to claim 1, it is characterised in that described step 4) in setting Threshold value is 0.51.

5. according to the distributed node drift detection method according to any one of claim 1-4, it is characterised in that the method is also It is included in after every minor node differentiates and drift beaconing nodes is carried out location updating, and directly regard it as unknown node and carry out again Location, and will estimate that position is as its new self-position.

6. a distributed node shift testing apparatus, it is characterised in that this device includes acquisition module, computing module, matching Module and judge module,

Described acquisition module communicates with remaining beaconing nodes RSSI for gathering each beaconing nodes at set time intervals Data (S_k,t_i), (S_k,t_i) represent at t_i(i=1,2 ..., m) moment beaconing nodes k communicates with remaining beaconing nodes RSSI data Collection be combined into S_k, k=1,2 ..., n, n are the sum of beaconing nodes, and m is sampling number；

Described computing module is used for calculating neighbouring sample moment each beaconing nodes RSSI change degree, to obtain each beaconing nodes RSSI change degree time series G；

Described fitting module, for RSSI change degree time series G of each beaconing nodes is carried out linear fit, obtains each letter The RSSI change degree time series fitted regression line that mark node is corresponding；

Described judge module for judge match value on each beaconing nodes RSSI change degree time series fitted regression line and Whether the difference between corresponding actual value is more than setting threshold value, if more than setting threshold value, being then the beaconing nodes that its correspondence is described Drift about.

Distributed node shift testing apparatus the most according to claim 6, it is characterised in that described computing module calculates To neighbouring sample moment each beaconing nodes RSSI change degree be:

$G\&lsqb;(S_k,t_i),(S_k,t_{i+1})\&rsqb;=\underset{j=1}{\overset{n}{\&Sigma;}}\left(1-\frac{|{(S_k,t_i)}_j-{(S_k,t_{i-1})}_j|}{\left|{(S_k,t_i)}_j-{(S_k,t_{i+1})}_j\right|+a_{kj}}\right)/n$

(S_k,t_i)_jFor t_iMoment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension, (S_k,t_i+1)_jFor t_i+1Moment beaconing nodes k communicates with remaining beaconing nodes RSSI data acquisition system S_kJth dimension；a_kjRepresent the upper (S of jth dimension_k,t_i) and (S_k,t_i+1) absolute value of meansigma methods；G∈[0,1].

Distributed node shift testing apparatus the most according to claim 6, it is characterised in that described fitting module uses Fitted regression line equation be:

D '=β₀+β₁t

${\mathrm{\&beta;}}_1=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}(t_i-\stackrel{\&OverBar;}{t})(d_i-\stackrel{\&OverBar;}{d})}{\underset{i=1}{\overset{n}{\&Sigma;}}{(t_i-\stackrel{\&OverBar;}{t})}^2}$

${\mathrm{\&beta;}}_0=\stackrel{\&OverBar;}{d}-{\mathrm{\&beta;}}_1\stackrel{\&OverBar;}{t}$

Wherein t represents sampling time point, β₀Represent the intercept of this straight line, β₁Representing the slope of this straight line, d ' represents sampling time point Match value corresponding to t；The meansigma methods of express time section,The RSSI change degree meansigma methods of beaconing nodes in express time section.

Distributed node shift testing apparatus the most according to claim 6, it is characterised in that setting in described judge module Determining threshold value is 0.51.

10. according to the distributed node shift testing apparatus according to any one of claim 6-9, it is characterised in that this detection Device also includes carrying out drift beaconing nodes after differentiating location updating at every minor node, and directly regards it as unknown joint Point carries out the module reorientated.