CN106993273A - Based on distance weighted and genetic optimization DV Hop localization methods - Google Patents

Abstract

The present invention provides a kind of based on distance weighted and genetic optimization DV Hop localization methods, main to include setting up signal intensity with distance relation carrying out the calculating of least-squares algorithm position, and new optimization method.Row constraint is entered to minimum hop count by the RSSI averages between anchor node first；Then jumped using the minimum hop count between anchor node to average away from weighting；Finally optimize the location estimation result of least-squares algorithm using Revised genetic algorithum, improve positioning precision.

Description

Based on distance weighted and genetic optimization DV-Hop localization methods

Technical field

It is especially a kind of based on distance weighted and genetic optimization the present invention relates to wireless sensor network technology field DV-Hop localization methods.

Background technology

With Internet of Things research and application deeply, wireless sensor network (Wireless Sensor Networks, WSN) The impetus of key is served wherein.Wireless sensor network has many applications in the field such as military, civilian, including Environmental monitoring, web portal security, medical diagnosis, battlefield surveillance, disaster relief and assisted living etc..Many applications and position in life Relevant, wireless sensor network is easy to deployment, that scalability is high, the low feature of cost makes it have in positioning indoors is very big Advantage.

Location technology is one of main support technology of wireless sensor network, by the extensive concern of researcher.It is existing Some wireless sensor network positioning can be divided mainly into the positioning based on ranging and the location algorithm without ranging, its main distinction It is whether need range information.Location algorithm interior joint based on ranging obtains range information, positioning accurate using ranging technology Degree is higher but needs extra equipment, and the connected relation that the location algorithm without ranging is only relied between adjacent node is determined Position, without the support of infrastructure network, positioning precision is relatively low.

If can be with direct communication between two nodes, then the distance between the two nodes are certain or equal to both Between maximum communication distance (R).Niculescu and Nath et al. propose DV-Hop algorithms, are communicated using between two nodes Hop count carrys out the distance between decision node relation.In wireless sensor network, due to the limitation of communication distance, two communication sections Direct communication link is not present between point, its communication can just be realized using other via nodes.Under normal circumstances, in order to Ensure on communication efficiency, the multilink between two nodes, most of routing algorithms can all select the few link of relay point.By In the hardware limitation of wireless device, in the case where ranging is impossible or inaccurate, the method for estimation based on hop count is to base In effective supplement of distance ranging.Any ranging hardware need not be relied on based on hop count positioning, be a kind of localization method of lightweight, But position error is larger.In order to improve the accuracy of positioning, generally require substantial amounts of reference mode to realize node and node The distance between estimation, so increasing network burden.

The implementation process of DV-Hop location algorithms is broadly divided into three phases：

(1) minimum hop count of network is obtained.It can be realized with a kind of distributed method：Each anchor node flood land is wide Broadcast the information { (x of oneself_i,y_i),h_i, wherein [x_i,y_i] represent anchor node i physical coordinates, h_iRecording distance anchor node i jump Number.h_iInitial value be 1, every time forwarding increase by 1.So when transit node receives information, the h of the message_iRepresent the now section The minimum hop count of point and anchor node i.The acquisition of hop count is comparatively accurate in denser network as shown in Figure 1, but Fig. 2 is dilute Dredge in type network, Node distribution is excessively sparse, hop count with jump away from be no longer simple linear corresponding relation so that jump away from error Become big.

(2) average distance often jumped is calculated.Assuming that the minimum hop count between anchor node j to anchor node i is h_ij, then they it Between average jump away from for：

As the hop count hop between two nodes_ijWhen known, d can be passed through_hop=d_ij×hop_ijObtain node i and node j The distance between.As shown in figure 3, distance is excessive between working as nodes so that euclidean distance between node pair error increases, causes to calculate and tie Fruit deviation accumulation.

(3) calculate node position.By previously obtained jump away from using trilateration or least-squares algorithm come really Its fixed position.

Assuming that (x, y) is the position of unknown node, (x_i,y_i) be anchor node position, d_iAnchor node is arrived for unknown node Distance；N represents anchor node quantity；The location model of least-squares algorithm is：

It is rewritten into matrix form：

It is abbreviated as：

Hx=b

Wherein,

The minimum variance solution of this equation is:

X=(H^TH)^-1HTb (4)

The position (x, y) of unknown node can be tried to achieve.

The content of the invention

The invention aims to solve the problem of wireless sensor network positioning precision is low, and propose it is a kind of be based on away from From the DV-Hop localization methods of weighted sum genetic optimization, the technical solution adopted by the present invention is：

It is a kind of based on distance weighted and genetic optimization DV-Hop localization methods, comprise the following steps：

(1) initialize network, each anchor node in the way of flooding to surrounding broadcast itself ID and receive unknown node RSSI；RSSI is signal intensity；

(2) according to the RSSI value of receiving, unknown node is corrected to the minimum of each anchor node with the average hop-by-hop of signal intensity Hop count；

(3) using the minimum hop count after correction come weighted average jump away from；

(4) the distance between anchor node is arrived away from unknown node is obtained according to the average jump of unknown node to anchor node；

(5) on the premise of position calculating is met, the target location of unknown node is calculated using least-squares algorithm；

(6) positioning result of unknown node is finally optimized using improved adaptive GA-IAGA.

Further, step (2) is specifically included：

The signal intensity that each anchor node is received is designated as：RSSI₁,RSSI₂,…,RSSI_n, n is anchor node quantity；To institute Some RSSI averageThen the signal intensity received is divided into two groups：

Average is less than or equal to according to signal intensity between above formula (8) nodeBe designated as 1 jump；Conversely, under utilizing Formula (9) minimum hop count is modified to

In formula (9), RSSI_ijRepresent the signal intensity between two adjacent nodes i and j；

Hop-by-hop carries out the final unknown node that obtains to the minimum hop count minhop of each anchor node_ij。

Further, step (3) is specifically included：

Assuming that the minimum hop count between unknown node i and each anchor node j is minh_ij, minh_ijPass through the side of step (2) Method is obtained；K represents the anchor node nearest apart from unknown node, the average jump of Part I away from：

C represent the average jump of network anchor node away from；

Part II averagely jump away from for

Wherein, h_kjFor the anchor node nearest apart from unknown node to hop count, d between other anchor nodes_kjFor apart from unknown node Nearest anchor node is to distance between other anchor nodes；w_kjFor range error weights；Convolution (10) and (11) obtain unknown node To anchor node average jump away from

Further, step (4) is specifically included：The distance between unknown node and anchor node are：

Wherein, d_xjFor the distance between unknown node to anchor node j,For being averaged for unknown node to anchor node Jump away from minhop_xjFor acquisition unknown node to minimum hop count between anchor node, obtained by method in step (2).

Further, in step (5), on condition that unknown node could be calculated when effective anchor node number is more than or equal to 3 Position.

Further, it is specific to carry out in step (6)：

The crossover probability and mutation probability of classical genetic algorithm are optimized, changes conventional fitness function and is allowed to applicable In wireless sensor network indoor positioning；Revised genetic algorithum key element and parameter setting are as follows：

1) chromosome and initial population：For the ease of genetic algorithm for solving, required unknown node coordinate (x, y) is regarded as The space of Solve problems, i.e. chromosome s；The coordinate transformation of unknown node coordinate (x, y) is

S=(v₁,v₂)(14)

2) fitness function：Revised genetic algorithum is reciprocal as fitness function using unknown node position error, adapts to Spending function is

3) crossover operator and mutation operator are respectively as shown in formula (17) and formula (18)：

Wherein, s_iAnd s_jIt is the parent chromosome of two random pairs in population, s_i' and s'_jIt is s_iAnd s_jBy intersecting behaviour The new chromosome generated after work, v_iRepresent new chromosome s two genes generated via previous step crossover operation, γ_iIt is dye Random number in the span of each gene of colour solid, v_i' represent the new chromosome s that is generated via previous step crossover operation The newest chromosome s' generated again after mutation operation two genes, y_iFor the random number between [0,1], P_cFor genetic algorithm Crossover probability, P_mFor the mutation probability of genetic algorithm；

Difference mproved crossover operators, crossover probability and mutation probability in mutation operator：

In formula, f_maxFor the maximum adaptation value in per generation population, f_avFor the average adaptive value of per generation population, f' is to intersect Two individuals in larger adaptive value, f is the larger adaptive value in two individuals to be made a variation, P_c1, P_c2, P_m1And P_m2For Constant less than 1；

4) selection opertor：Each chromosome in population is all made after intersecting, making a variation with formula (16) fitness function For foundation；Follow formula (19) and (20) in intersection and mutation process to intersect and mutation probability, selection fitness function value is big Chromosome be retained on chromosome position.

The advantage of the invention is that：

(1) the generally each anchor node of the acquisition of traditional DV-Hop location algorithm hop counts using broadcasting the message with flooding, just Initial value is set to 0, and forwarding increase by 1, can so obtain minimum hop count between node every time.Because distance is not between anchor node Together, this allow for hop count acquisition it is excessively simple, it is impossible to embody the how far of nodes.Method after improvement is node Between distance is big is jumped for 1, will be less than 1 apart from small and jump, hop count is no longer mechanical superposition.

(2) traditional average jump is away from suitable for intensive Node distribution, but in sparse network, euclidean distance between node pair If still easily causing node calculation error using the distance apart from the nearest anchor node of unknown point to add up.By to it is average jump away from Weighting, reduces anchor node to range error between unknown node, has a significant impact to improving positioning precision as far as possible.

(3) in the case of indoors, due to being disturbed by environment, larger range error is caused, using least-squares algorithm Positioning precision is not ideal, generally requires the positioning precision that substantial amounts of anchor node could be high, adds the hardware cost of positioning. In the case of not increasing anchor node number, the present invention uses genetic algorithm optimization result of calculation.

Brief description of the drawings

Fig. 1 is the schematic diagram of node intensity distribution.

Fig. 2 is the schematic diagram of node sparse distribution.

Fig. 3 is the excessive schematic diagram of nodal distance.

Fig. 4 is the schematic diagram of Experimental Network topological structure.

The schematic diagram of positioning performance contrast when Fig. 5 is R=30m.

The schematic diagram of positioning performance contrast when Fig. 6 is R=35m.

The schematic diagram of positioning performance contrast when Fig. 7 is R=40m.

Embodiment

With reference to specific drawings and examples, the invention will be further described.

Based on distance weighted and genetic optimization DV-Hop localization methods, mainly include hop count amendment, jump and lost away from weighted sum Propagation algorithm optimizes three parts.The present invention using 100m × 100m as the region of experiment simulation, in simulating area all nodes it Between communicate normal, communication radius is 30m, 35m and 40m；Node total number is 100, anchor node quantity is respectively 5,10,15,20, 25th, 30,35 and 40；Iterations is 60, and all l-G simulation tests are carried out 500 times, utilize positioning precision and average localization error Evaluation algorithm positioning precision：

Average positioning precision is

Wherein, (xⁱ,yⁱ) be unknown node physical location, (x'ⁱ,y'ⁱ) position is calculated for unknown node, R is communication half Footpath, m is network-in-dialing degree, and N is unknown node quantity.

In the present embodiment, anchor node refers to the wireless senser laid in advance, that is, the beaconing nodes in Fig. 4；Unknown section Point is sensor to be positioned；

It is proposed by the present invention based on distance weighted and genetic optimization DV-Hop localization methods, comprise the following steps：

(1) initialize network, each anchor node in the way of flooding to surrounding broadcast itself ID and receive unknown node RSSI；RSSI is signal intensity；

(2) according to the RSSI value of receiving, unknown node is corrected to the minimum of each anchor node with the average hop-by-hop of signal intensity Hop count；It is described in detail below：

Distance between the intensity estimation signal egress received according to node, its calculation is as follows

Wherein, P₀(d₀) it is reference mode d₀The signal intensity at place, P_r(d) it is strong for the signal in the unknown node at d Degree, η is path loss coefficient, and usual value is between 2~5；X_σIt is that standard deviation is that the Gaussian Profile that σ, average are zero becomes at random Amount；

It is reduced to

P_r(d)=A-10 η lgd (7)

In formula, A is the signal intensity received at 1m；Understand that distance is more between receiving the stronger node of signal intensity from formula It is short.Theoretical based on this, the signal intensity that each anchor node is received is designated as：RSSI₁,RSSI₂,…,RSSI_n, n is anchor node number Amount；All RSSI are averagedThen the signal intensity received is divided into two groups：

Average is less than or equal to according to signal intensity between above formula (8) nodeBe designated as 1 jump；It is on the contrary, it is believed that to jump Number, which is adjusted the distance, contributes small, is modified to using following formula (9) minimum hop count

In formula (9), RSSI_ijRepresent the signal intensity between two adjacent nodes i and j.Assuming that unknown node launches one group Signal, is not that machinery Jia 1 during forwarding, but by being compared to judge whether needs with signal intensity average Amendment, hop-by-hop carries out that a minimum node hop count minhop may finally be obtained_ij。

(3) using the minimum hop count after correction come weighted average jump away from；It is described in detail below：

Assuming that the minimum hop count between unknown node i and each anchor node j is minh_ij(it can be obtained by the method for step 2 ), k represents the anchor node nearest apart from unknown node, the average jump of Part I away from：

C represent the average jump of network anchor node away from；

Part II averagely jump away from for

Wherein, h_kjFor the anchor node nearest apart from unknown node to hop count, d between other anchor nodes_kjFor apart from unknown node Nearest anchor node is to distance between other anchor nodes；w_kjFor range error weights；Convolution (10) and (11) obtain unknown node To anchor node average jump away from

(4) the distance between anchor node is arrived away from unknown node is obtained according to the average jump of unknown node to anchor node；

The distance between unknown node and anchor node can be obtained according to formula (13)：

Wherein, d_xjFor the distance between unknown node to anchor node j,For the average jump of unknown node to anchor node Away from minhop_xjFor acquisition unknown node to minimum hop count between anchor node, obtained by method in step (2)；

(5) on the premise of position calculating is met, the target location of unknown node is calculated using least-squares algorithm；This portion Point content had made introduction in background technology again；On condition that could calculate unknown when effective anchor node number is more than or equal to 3 Node location；

(6) positioning result of unknown node is finally optimized using improved adaptive GA-IAGA.

The crossover probability and mutation probability of classical genetic algorithm are optimized, changes conventional fitness function and is allowed to applicable In wireless sensor network indoor positioning；Revised genetic algorithum key element and parameter setting are as follows：

1) chromosome and initial population：For the ease of genetic algorithm for solving, required unknown node coordinate (x, y) is regarded as The space of Solve problems, i.e. chromosome s；The coordinate transformation of unknown node coordinate (x, y) is

S=(v₁,v₂) (14)

(v₁,v₂) for conversion after unknown node coordinate；

2) fitness function：The object function of problem space is that unknown node coordinate is determined in GA-DV-Hop location algorithms Position error be

Revised genetic algorithum is using unknown node position error inverse as fitness function, and its value is bigger, chromosome Adaptability is stronger bigger by the hereditary possibility gone down, and fitness function is

(v₁',v'₂) unknown node after Coordinate Conversion calculates position；

3) crossover operator and mutation operator are respectively as shown in formula (17) and formula (18)：

Wherein, s_iAnd s_jIt is the parent chromosome of two random pairs in population, s_i' and s'_jIt is s_iAnd s_jBy intersecting behaviour The new chromosome generated after work, v_iRepresent new chromosome s two genes generated via previous step crossover operation, γ_iIt is dye Random number in the span of each gene of colour solid, v_i' represent the new chromosome s that is generated via previous step crossover operation The newest chromosome s' generated again after mutation operation two genes, y_iFor the random number between [0,1], P_cFor genetic algorithm Crossover probability, P_mFor the mutation probability of genetic algorithm；For the defect in genetic algorithm, crossover probability P_cWhen excessive, new individual The speed of generation is faster, but the destruction to original population is also bigger, otherwise search speed is slow；Mutation probability P_mIt is difficult to generation new Variation individual, otherwise be difficult to search optimal position.In this regard, the intersection respectively in mproved crossover operators, mutation operator is general Rate and mutation probability：

In formula, f_maxFor the maximum adaptation value in per generation population, f_avFor the average adaptive value of per generation population, f' is to intersect Two individuals in larger adaptive value, f is the larger adaptive value in two individuals to be made a variation, P_c1, P_c2, P_m1And P_m2For Constant less than 1；

4) selection opertor：Each chromosome in population is all made after intersecting, making a variation with formula (16) fitness function For foundation；Follow formula (19) and (20) in intersection and mutation process to intersect and mutation probability, selection fitness function value is big Chromosome be retained on chromosome position.

Algorithm Analysis：

1) anchor density

Wireless sensor network node positioning is to solve unknown node by the range information of anchor node around unknown node Position, this method to anchor node dispose quantitative requirement it is very high, when anchor node deployment density is high around unknown node, positioning calculate Method shows preferable performance, and when the quantity of anchor node is reduced, positioning precision significantly declines.As shown in Fig. 5,6,7, 10m × 10m localization regions, new localization method is in the case where anchor node ratio is relatively low, and higher positioning can so be kept by appointing Precision, with the increase of anchor node ratio, positioning precision improves obvious.

2) precision of positioning

For a kind of given location algorithm, positioning precision shows the matching journey for calculating position and physical location of node Degree.Specifically, positioning precision is decided to be the distance between location estimation value and its actual position of a unknown node.In nothing In line sensor network, the precision of indoor positioning is not universal high, it is impossible to meet the demand in market.As shown in Fig. 5,6,7, in 10m × 10m localization region, it can be seen that the present invention is maintained to higher positioning precision under different communication radius.

Claims (6)

1. it is a kind of based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that to comprise the following steps：

(1) initialize network, each anchor node in the way of flooding to surrounding broadcast itself ID and receive the RSSI of unknown node； RSSI is signal intensity；

(2) according to the RSSI value of receiving, unknown node is corrected to the minimum hop count of each anchor node with the average hop-by-hop of signal intensity；

(3) using the minimum hop count after correction come weighted average jump away from；

(4) the distance between anchor node is arrived away from unknown node is obtained according to the average jump of unknown node to anchor node；

(5) on the premise of position calculating is met, the target location of unknown node is calculated using least-squares algorithm；

(6) positioning result of unknown node is finally optimized using improved adaptive GA-IAGA.

2. it is as claimed in claim 1 based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that

Step (2) is specifically included：

The signal intensity that each anchor node is received is designated as：RSSI₁,RSSI₂,…,RSSI_n, n is anchor node quantity；To all RSSI averagesThen the signal intensity received is divided into two groups：

$\left\{\begin{array}{c}\stackrel{\&OverBar;}{RSSI}\&GreaterEqual;{\mathrm{RSSI}}_j\\ \stackrel{\&OverBar;}{RSSI}<{\mathrm{RSSI}}_j\end{array}\right.,j=1,2,...,n---\left(8\right)$

Average is less than or equal to according to signal intensity between above formula (8) nodeBe designated as 1 jump；Conversely, utilizing following formula (9) Minimum hop count is modified to

${\mathrm{hop}}_{ij}=\frac{{\mathrm{RSSI}}_{ij}}{\stackrel{\&OverBar;}{RSSI}}---\left(9\right)$

In formula (9), RSSI_ijRepresent the signal intensity between two adjacent nodes i and j；

Hop-by-hop carries out the final unknown node that obtains to the minimum hop count minhop of each anchor node_ij。

3. it is as claimed in claim 1 based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that

Step (3) is specifically included：

Assuming that the minimum hop count between unknown node i and each anchor node j is minh_ij, minh_ijObtained by the method for step (2) ；K represents the anchor node nearest apart from unknown node, the average jump of Part I away from：

$hop\_{\stackrel{\&OverBar;}{d}}_1=\frac{1}{\min h_{ij}}*C---\left(10\right)$

C represent the average jump of network anchor node away from；

Part II averagely jump away from for

$hop\_{\stackrel{\&OverBar;}{d}}_2=(1-\frac{1}{\min h_{ij}})*\frac{d_{kj}}{h_{kj}}*w_{kj}---\left(11\right)$

Wherein, h_kjFor the anchor node nearest apart from unknown node to hop count, d between other anchor nodes_kjTo be nearest apart from unknown node Anchor node to distance between other anchor nodes；w_kjFor range error weights；Convolution (10) and (11) obtain unknown node to anchor The average jump of node away from

$hop\_\stackrel{\&OverBar;}{d}=hop\_{\stackrel{\&OverBar;}{d}}_1+hop\_{\stackrel{\&OverBar;}{d}}_2---\left(12\right).$

4. it is as claimed in claim 1 based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that

Step (4) is specifically included：The distance between unknown node and anchor node are：

$d_{xj}=hop\_\stackrel{\&OverBar;}{d}*\min {\mathrm{hop}}_{xj}---\left(13\right)$

Wherein, d_xjFor the distance between unknown node to anchor node j,For unknown node to anchor node average jump away from, minhop_xjFor acquisition unknown node to minimum hop count between anchor node, obtained by method in step (2).

5. it is as claimed in claim 1 based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that

In step (5), on condition that unknown node position could be calculated when effective anchor node number is more than or equal to 3.

6. it is as claimed in claim 1 based on distance weighted and genetic optimization DV-Hop localization methods, it is characterised in that

It is specific to carry out in step (6)：

The crossover probability and mutation probability of classical genetic algorithm are optimized, changes conventional fitness function and is allowed to be applied to nothing Line sensor network indoor positioning；Revised genetic algorithum key element and parameter setting are as follows：

1) chromosome and initial population：For the ease of genetic algorithm for solving, required unknown node coordinate (x, y) is regarded as and is to solve for The space of problem, i.e. chromosome s；The coordinate transformation of unknown node coordinate (x, y) is

S=(v₁,v₂) (14)

2) fitness function：Revised genetic algorithum assign unknown node position error inverse as fitness function, fitness letter Number is

$f(v_1,v_2)=\frac{1}{\sqrt{{(v_1-v_1^{\&prime;})}^2+{(v_2-v_2^{\&prime;})}^2}}---\left(16\right)$

3) crossover operator and mutation operator are respectively as shown in formula (17) and formula (18)：

$\left\{\begin{array}{c}{s}_i^{\&prime;}=P_c{s}_i+(1+P_c)s_j\\ s_j^{\&prime;}=(1-P_c)s_i+P_c{s}_j\end{array}\right.---\left(17\right)$

$v_i^{\&prime;}=\left\{\begin{array}{cc}{\mathrm{\&gamma;}}_i& y_i\&le;P_m\\ v_i& y_i>P_m\end{array}\right.---\left(18\right)$

Wherein, s_iAnd s_jIt is the parent chromosome of two random pairs in population, s '_iAnd s'_jIt is s_iAnd s_jAfter crossover operation The new chromosome of generation, v_iRepresent new chromosome s two genes generated via previous step crossover operation, γ_iIt is chromosome Random number in the span of each gene, v '_iRepresent that the new chromosome s generated via previous step crossover operation is passed through again The newest chromosome s' generated after mutation operation two genes, y_iFor the random number between [0,1], P_cFor the friendship of genetic algorithm Pitch probability, P_mFor the mutation probability of genetic algorithm；

Difference mproved crossover operators, crossover probability and mutation probability in mutation operator：

$P_c=\left\{\begin{array}{cc}{P}_{c1}-\frac{(P_{c1}-P_{c2})(f^{\&prime;}-f_{av})}{f_{max}-f_{av}}& f^{\&prime;}\&GreaterEqual;f_{av}\\ P_{c1}& f^{\&prime;}<f_{av}\end{array}\right.---\left(19\right)$

$P_m=\left\{\begin{array}{cc}{P}_{m1}-\frac{(P_{m1}-P_{m2})(f-f_{av})}{f_{\max }-f_{av}}& f\&GreaterEqual;f_{av}\\ P_{m1}& f<f_{av}\end{array}\right.---\left(20\right)$

In formula, f_maxFor the maximum adaptation value in per generation population, f_avFor the average adaptive value of per generation population, f' is to be intersected two Larger adaptive value in individual, f is the larger adaptive value in two individuals to be made a variation, P_c1, P_c2, P_m1And P_m2For less than 1 Constant；

4) selection opertor：Each chromosome in population is by intersecting, after variation, all using formula (16) fitness function as according to According to；Formula (19) is followed in intersection and mutation process and (20) intersect and mutation probability, the dye of selection fitness function value greatly Colour solid is retained on chromosome position.