CN106162790A - The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network - Google Patents

Abstract

The invention discloses the multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network.The method uses hierarchical logical design, respectively neighbor table to set up, routing table sets up and routing table sets up three phases.For avoiding the most mobile frequent updating causing neighbor table because of sink node, use the movement of the restriction on the parameters sink such as dump energy, time and object function, it is achieved that node optionally carries out the renewal of neighbor table.Neighbor table and routing table use separate type to set up process, brought renewal amount is moved with minimizing sink, at search routing phase, the tabu search algorithm improved is used to search for global optimum's path collection, route is concentrated at optimal path, for avoiding invalid route, " short plank law " is utilized to weigh the health index of path link, when it is less than when calculating threshold value, if link is critical path, then sink node moves towards " short slab " path, it is achieved that the code-on-demand migration of sink, efficient balance network energy consumption, extends network lifecycle.

Description

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network

Technical field

The multipath Geographic routing that the present invention relates to code-on-demand migration sink in a kind of wireless sense network is excellent Change method, belongs to mobile communication routing algorithm field in wireless sense network.

Background technology

Along with developing rapidly of information age, the application of Internet of Things is also come thick and fast, and sensor It is ingredient indispensable in Internet of Things, currently the research temperature of wireless sense network is also healed Drilling the strongest, wireless sense network (WSN) is by the sensor node group of many distributed deployments The wireless network become, these sensors are responsible for the data of perception by single-hop list footpath or multi-hop Multipath is transmitted to aggregation node (sink node), owing to sensor is by the battery of finite energy Power supply, therefore, node deployment and Design of Routing Protocol are required for from energy-conservation angle, The energy balance of limits, extends network lifecycle.This is general to radio sensing network application And there is important theoretical and practical values.

In wireless sense network, if sink node is fixed, then the most static wireless sense network, Its advantage is that network topology is the most stable, and shortcoming is that network " the most dead " phenomenon easily occurs.This The neighbor node being because sink node not only sends the data of self perception to sink node, And undertake the relay task of other sensing node data, cause node energy consumption uneven.

Hidenhisa et al. proposes SPAT (Sink Mobility Based on Clustering and Set Packing Techniques) algorithm, use executor to control the movement of sink, it will divide Bunch and mass-rent technology combine control sink movement.Improve data collection efficiency, but increase Network cost, and, the holding that sink frequently movement cannot continue is normal with backbone network Connect.

For improving network throughput, equalising network resource maximized prolongation network life week Phase, introducing multipath mechanism, Chen Min et al. proposes efficiency and the multipath of load balancing Geographic routing algorithm (DGR), carries out Path selection according to the key energy calculated, avoids remaining The less node of complementary energy is with maximization network life span.But do not account for sink node Mobile to solve " energy hole " problem that neighbor node occurs.

Summary of the invention

Technical problem: the present invention is directed to deficiency present in such scheme, it is proposed that a kind of wireless The multipath Geographic routing optimization method of code-on-demand migration sink, i.e. DMRA algorithm in Sensor Network. The method has not only equalized network energy consumption, improves network performance, and, maximized prolongation Network lifecycle.

Technical scheme: the present invention proposes the multipath of code-on-demand migration sink in a kind of wireless sense network Geographic routing optimization method, the method utilizes the multiple ginsengs such as dump energy, time and object function The frequent movement of number constraint sink node, neighbor table and routing table use the mistake that separate type is set up Journey, to reduce the renewal amount of neighbor node neighbor table, when search route, uses the taboo improved Avoid searching algorithm and effectively prevent the loss of optimal path.For avoiding invalid route, use " short slab " The health index of link is detected by principle.It is effectively improved network performance, extends network life Cycle.

Assume that N number of wireless sensor node is uniformly random to be distributed in sink node for the center of circle partly Footpath is the border circular areas of l, and the communication radius of node is r (r ＞ 0, r ＜ ＜ l).Assuming that all nodes Data genaration rate is identical, and time per unit produces a bag.Node data transfer rate is identical, joint The through-put power of point is constant.The total time of node-node transmission data is designated as ∑ t；The renewal week of node Phase is designated as T；Mobile update status indicator is designated as MFlag；Dump energy is E_rem, successfully pass The minimum energy of defeated k Bit data is designated as E_min(k).The primary power of node is E₀, node Perception, registration, wireless transmission and the total energy consumption E of reception k bit bag_dlK (), by formula (1) Calculate.

E_all(k)=E_sen(k)+E_mem(k) ten E_tra(k, d)+E_rec(k) (1)

Wherein, E_senK () represents perception energy consumption, E_memK perception data is read and write internal memory by () expression Energy consumption, E_tra(k, d) represents the transmission energy consumption of transmission k bit bag, and d represents sensor and event There is the distance of point, E_recK () represents the reception energy consumption receiving k bit bag.

Residue energy of node E_remCalculated by formula (2).

E_rem=E₀-E_all(k)-E_upd(k) (2)

Wherein, E_updK () represents the energy consumption that neighbor table updates.When neighbor table does not updates, E_upd(k)=0.

It specifically comprises the following steps that

Step one, neighbor table establishment stage:

S1: initialize network configuration, the MFlag=false of sink node；

S2: if the ∑ t ＞ T of node i, then set up neighbor table according to following steps；

A, iteration: calculated distance D of node i (i=1) and node j (j=i+1) by formula (3)；

$D=\sqrt{{(S_i^x\left(t\right)-S_j^x\left(t\right))}^2+{(S_i^y\left(t\right)-S_j^y\left(t\right))}^2}---\left(3\right)$

Wherein,Represent node i, the position coordinates when t.

If b D is ＜=r, then node i broadcast REQ message；

If it is sink node that c receives the node of REQ message, its MFlag of sink nodal test is No is true, if very, illustrates that sink is moved, then re-establishes neighbor table, turn S5, Otherwise, neighbor table is not updated；

If it is not sink node that d receives the node of REQ message, then whether judge REQ interior joint i Meet condition: E_rem＞ E_min, the most then decision node i receives two or two the most simultaneously Above REQ, if so, by the hop count information in REQ, according to jumping figure from small to large suitable Sequence, adds neighbor table by node, turns S6；Otherwise, e is turned；

E, according to the REQ received time sequencing add neighbor table, turn S6；

If f REQ interior joint i is unsatisfactory for condition: E_rem＞ E_min, then node i is added without neighbor table, i++；

If g is D ＞ r, i++, turn h；

If h is i ＜ N-1, then turn a；Otherwise, S4 is turned；

S3: if the ∑ t ＜ T of node i, then turn S4；

S4: judge whether MFlag is true, if very, then turns S5；Otherwise node enters dormancy or prison Survey state；

S5:sink node broadcasts current location information and previous neighbours and current neighbor nodes；Neighbours Node updates neighbor table；

S6: output neighbor table NQue；

Step 2, routing table establishment stage:

T1: by S6, calculate distance D between node i and sink according to geometric formula (4)；

$D=\sqrt{{(S_i^x\left(t\right)-{\mathrm{DS}}^x\left(t\right))}^2+{(S_i^y\left(t\right)-{\mathrm{DS}}^y\left(t\right))}^2}---\left(4\right)$

T2: if D is ＜=r, then and all nodes in communication range all send ACK message, if sink joint Point receives the ACK message of node i, then node adds approach according to the reception time sequencing of ACK Footpath table queue PQue, turns T8；Otherwise, node failure is described, then node i adds taboo list team Row TQue, turns T8；

T3: if D is ＞ r, then the node j outside sink communication range sends ReACK bag by multi-hop；

T4: if sink receives Re ACK bag, then by node j according to the time sequencing receiving Re ACK Add PQue；Otherwise turn T5；

Whether T5: decision node j meets following two condition:

Condition (1): the E of node j_rem＜=E_min

Condition (2): the path at node j place is " short slab " path

T6: if meeting, then sink node motion, MFlag=true, go to step one；

T7: if being unsatisfactory for, then sink node does not moves, and obtains j's by the neighbor node of node j Information；

T8: output PQue and TQue.

Step 3, routing table establishment stage:

R1: obtained PQue by T8, if PQue is not empty, iteration compares, and target function value is optimum Path add PQue and TQue；Turn R2；

R2: output routing table RQue；

R3: when PQue is empty, terminates iteration, turn R4；

R4: invocation step two, sets up routing table, if MFlag=true, then invocation step one；No Then, R5 is turned；

The link health index η of R5: calculating, if η is more than the healthy threshold value of link, then defeated according to R2 The RQue queue route gone out；Otherwise, R6 is turned；

R6: if path p_nBe " short slab " path, then MFlag=true, sink node is towards " short Plate " path is moved, and p_nLift a ban from TQue；

R7: if path p_nNot " short slab " path, then MFlag=false；RQue according to R2 output Route.

Object function described in step 3 constitutes by formula (5), (6) definition:

$\min f_2\left(p_n\right)=\frac{\mathrm{\Σ}\left|(p_m\backslash \{S,\mathrm{DS}\}\∩p_n\backslash \{S,\mathrm{DS}\left\}\right)\right|}{\mathrm{\Σ}\left|(p_m\backslash \{S,\mathrm{DS}\}\∪p_n\backslash \{S,\mathrm{DS}\left\}\right)\right|}---\left(6\right)$

And meet: $E_{\mathrm{all}}^i\left(b\right)*\mathrm{\Σ}{t}_i+E_{\mathrm{up}}^i\left(b\right)*y_i\≤E_0$

Wherein,WithRepresentation unit time path p respectively_nOn node i transmission k ratio The energy consumption of special data and renewal energy consumption, y_iAlong with MFlag value changes, if MFlag=true, then y_i=1, otherwise, y_i=0.

Multiple object functions are changed into single-object problem by the method, i.e.

$\max (F\left(p_n\right)=\frac{f_1\left(p_n\right)}{f_2\left(p_n\right)+1})---\left(7\right)$

Formula (7) expression obtains a paths p from meeting condition path concentration_nSo that single goal is excellent Change functional value maximum.Due to f_a(p_n) may be 0, it is nonintersecting paths, therefore formula (7) Middle denominator is made to add 1 process.

Described " short slab " path is the definition of the health index to link:

$\mathrm{\η}=\frac{\underset{i\∈p_n}{\min \left(E_{\mathrm{rem}}^i\right)}}{\underset{i=1}{\overset{i=p_n}{\mathrm{\Σ}}}{E}_{\mathrm{rem}}^i}---\left(8\right)$

Beneficial effect: the multipath Geographic routing of mobile sink in wireless sense network is calculated by the present invention Method is studied, and it is many to propose code-on-demand migration sink node in a kind of wireless sense network Footpath Geographic routing optimization method.According to the demand of reality application, the method utilizes multiple parameter about The movement of bundle sink node, thus reduce the renewal amount that sink node motion is brought, and carry Go out the tabu search algorithm improved and realize the search to multipath parallelization, improve route effect Rate, greatly reduces network energy consumption, for avoiding invalid route, utilizes the detection of " short slab " principle The health index of link, when it is less than when calculating threshold value, if this link is critical path, then sink Node moves towards " short slab " path, it is achieved that the code-on-demand migration of sink node.Theoretical and Experimental results demonstrate, the method effectively reduces network energy consumption, improves network performance, its net Network life cycle extends nearly 4 times than the network lifecycle of static wireless sense network.

Accompanying drawing explanation

The architectural schematic of Fig. 1 wireless sense network；

Fig. 2 is that neighbor table sets up algorithm flow chart；

Fig. 3 is that routing table sets up algorithm flow chart；

Fig. 4 is the TABU search routing algorithm flow chart improved；

Fig. 5 is the comparison schematic diagram of the average residual energy of various routing algorithm；

Fig. 6 is the comparison schematic diagram of the network lifecycle of various routing algorithm；

Detailed description of the invention

Below in conjunction with the accompanying drawings and detailed description of the invention, the invention will be further described.

Basic thought: the present invention provides code-on-demand migration sink node in a kind of wireless sense network Multipath Geographic routing optimization method, the method uses hierarchical logical design, is adjacent respectively Occupy table foundation, routing table sets up and routing table sets up three phases.For avoiding because of sink node The frequently mobile frequent updating causing neighbor table, use dump energy, time and object function Movement etc. multiple restriction on the parameters sink node, it is achieved that node optionally carries out neighbor table Renewal.Neighbor table and routing table use separate type to set up process, move institute reducing sink The renewal amount brought, at search routing phase, uses the tabu search algorithm parallel search improved Global optimum's path collection, concentrates at optimal path and route, and for avoiding invalid route, utilizes " short-board effect " weighs the health index of path link, when it is less than calculating threshold value, and sink Node is towards " short slab " node motion, it is achieved that the code-on-demand migration of sink node, effectively Equalize network energy consumption, extend network lifecycle.

Definition: outer neighborhood

If geometry A, B are the regions closed, outer neighborhood Ω is designated as:

Ω=A (A ∩ B)=A-B (9)

Data Structure Design is as follows:

The fixed field of node (including sink node):

SID

DID

SeqNum

SPOS/DPOS

Variable field:

NQue	HopC	NexHop	PreHop
				t	Erem	Data	Eupd

Sink node parameter field:

SID and DID represents source node and the ID of sink node respectively, and SeqNum represents sequence Column number, SPOS and DPOS is with two tuple (S^x(t), S^y(t)) represent source node and sink respectively The position both horizontally and vertically gone up in node rectangular coordinate system；HopC is used for representing jumping figure；t For packet being transmitted timing, and compare with update cycle T.In the field of sink node, MFlag is Boolean variable, represents whether sink node moves；F(p_n) it is compound function, it is used for The measurement condition of optimal route selection.NQue, PQue, RQue and TQue are to record shape The queue of formula storage, represents neighbor table, routing table, routing table and taboo list respectively, and TQue deposits Exoergic is less than energy threshold node, and this node is the most accessed, and deposits this repeatedly Path optimum in Dai, when, after node motion, being lifted a ban by the optimal path in TQue table, represents Can be accessed again；For reducing message packet length, REQ message represents a record, comprises The information of node fixed field and HopC parameter.MES message comprise node fixed field, MFlag and NQue parameter.ACK message comprises SID, E_rem, HopC, t and in Ω region In neighbor node NQue.ReACK message comprises SID, E_rem, t and all of NQue.

The present invention uses OPNET emulation platform, and simulated environment parameter configuration is as shown in table 1: Node total number is 500, and in network size is the region of 100 square metres, network topology is joined at random Putting, report interval is 0.2s, and data transmission rate is 1Mbps, and node communication radius is 50m, Node primary power is 36J, and the total energy consumption of node Successful transmissions k Bit data is 50nJ/bit, Update cycle is 0.8s, perception data size acquiescence 1KB, message packet size acquiescence 128B.

Table 1 simulated environment parameter configuration

Fig. 2 provides neighbor table and sets up the flow chart of algorithm, as MFlag=true, i.e. and sink Node motion, then sink node broadcasts current location information to the neighbor node of previous position and The neighbor node of current location, these nodes are updated neighbor table, it is achieved that the most more Newly, decrease renewal amount, save the energy consumption of node and network.

Fig. 3 provides routing table and sets up algorithm flow chart, the basis that this algorithm is set up in neighbor table On, use ACK and Re ACK message to determine whether path adds routing table or taboo Table.

Fig. 4 provides the TABU search routing algorithm flow chart of improvement, is in place of this algorithm improvement Taboo list not only deposits the energy node less than threshold value, and these nodes will be the most accessed, And deposit optimal path, as MFlag=true, optimal path just can be lifted a ban, Ke Yizai Secondary accessed, effectively prevent the loss of optimal solution.

Average residual energy with the relation of link health index as it is shown in figure 5, DGR's is average Dump energy is below SPAT and DMRA, this is because in the case of network size is less, DGR is setting up a plurality of nonintersecting paths, and energy expenditure is bigger.And optimization side in this paper The dump energy averaged behavior of method (DMRA) is the highest, and reason is that the movement of sink is subject to Distance, dump energy and time isoparametric constraint, so optionally updating neighbor table and road Footpath table makes that energy consumption is maximized is saved.Along with the expansion of network size, DGR remains Energy significantly reduces.The dump energy of SPAT also has significantly reduction, but remains than DGR Complementary energy is high, illustrates that the route of mobile sink has equalized energy consumption to a certain extent.And this The dump energy of the DMRA of bright proposition all high than other strategies, illustrate along with network size Increase, the multipath Geographic routing strategy of code-on-demand migration sink effectively reduce bottleneck joint Point energy consumption, the sensor network of big data quantity is risen by the parallel data transmission of the most optimum multipath To energy-conservation, act on efficiently.

The network lifecycle of various routing algorithms the most as shown in Figure 6: DGR Life Cycle Phase is consistently lower than DMRA and SPAT strategy, because DGR is not set up mobile mechanism, with The increase of network size, the relay task of the neighbor node of sink is more and more heavier, and energy disappears Consumption is accelerated, and therefore life cycle is gradually lowered.And the life cycle of DMRA than SPAT and DGR's is equal high, and reason is that the code-on-demand migration of sink node has effectively equalized bottleneck node Energy consumption.And parallel from optimal path collection routing data, maximized to extend network raw The life cycle.Generally, the average life cycle of this strategy is the life of static DGR strategy 3.96 times of cycle, 2.13 times of SPAT.

Foregoing description is only used as the enforceable technical scheme of the present invention and proposes, not as to its technology The single restrictive condition of scheme itself.

Claims (7)

1. the multipath Geographic routing optimization method of code-on-demand migration sink in a wireless sense network, it is characterised in that use hierarchical logical design, is that neighbor table is set up, routing table sets up and routing table sets up three phases respectively.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterized in that, network model is: assumes that N number of wireless sensor node is uniformly random and is distributed in sink node for the center of circle radius border circular areas as l, the communication radius of node is r (r ＞ 0, r ＜ ＜ l).Assuming that the data genaration rate of all nodes is identical, time per unit produces a bag.Node data transfer rate is identical, and the through-put power of node is constant.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterised in that energy model: assume that the primary power of node is E₀, node perceived, registration, wireless transmission and the total energy consumption E of reception k bit bag_dllK (), is calculated by formula (1).

E_all(k)=E_sen(l)+E_mem(k)+E_tra(k, d)+E_rec(k) (1)

Wherein, E_senK () represents perception energy consumption, E_memK () represents the energy consumption that perception data reads and writes internal memory, E_tra(k, d) represents the transmission energy consumption of transmission k bit bag, and d represents that sensor and event occur the distance of point, E_recK () represents the reception energy consumption receiving k bit bag.

Residue energy of node E_remCalculated by formula (2).

E_rem=E₀-E_all(k)-E_upd(k) (2)

Wherein, E_updK () represents the energy consumption that neighbor table updates.When neighbor table does not updates, E_upd(k)=0.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterised in that the total time of node-node transmission data is designated as ∑ t；The update cycle of node is designated as T；Mobile update status indicator is designated as MFlag；Dump energy is E_rem, the minimum energy of Successful transmissions k Bit data is designated as E_min(k).Specifically comprising the following steps that of the method

Step one, neighbor table establishment stage:

S1: initialize network configuration, the MFlag=false of sink node；

S2: if the ∑ t ＞ T of node i, then set up neighbor table according to following steps；

A, iteration: calculated distance D of node i (i=1) and node j (j=i+1) by formula (3)；

Wherein,Represent node i, the position coordinates when t.

If b D is ＜=r, then node i broadcast REQ message；

If it is sink node that c receives the node of REQ message, whether its MFlag of sink nodal test is true, if very, illustrates that sink is moved, then re-establishes neighbor table, turn S5, otherwise, does not update neighbor table；

If it is not sink node that d receives the node of REQ message, then judge whether REQ interior joint i meets condition: E_rem＞ E_min, the most then decision node i receives two or more REQ the most simultaneously, if so, by the hop count information in REQ, according to jumping figure order from small to large, node is added neighbor table, turns S6；Otherwise, e is turned；

E, according to the REQ received time sequencing add neighbor table, turn S6；

If f REQ interior joint i is unsatisfactory for condition: E_rem＞ E_min, then node i is added without neighbor table, i++；

If g is D ＞ r, i++, turn h；

If h is i ＜ N-1, then turn a；Otherwise, S4 is turned；

S3: if the ∑ t ＜ T of node i, then turn S4；

S4: judge whether MFlag is true, if very, then turns S5；Otherwise node enters dormancy or monitoring state；

S5:sink node broadcasts current location information and previous neighbours and current neighbor nodes；Neighbor node updates neighbor table；

S6: output neighbor table NQue；

Step 2, routing table establishment stage:

T1: by S6, calculate distance D between node i and sink according to geometric formula (4)；

T2: if D is ＜=r, then and all nodes in communication range all send ACK message, if sink node receives the ACK message of node i, then node adds routing table queue PQue according to the reception time sequencing of ACK, turns T8；Otherwise, node failure is described, then node i adds taboo list queue TQue, turns T8；

T3: if D is ＞ r, then the node j outside sink communication range sends Re ACK bag by multi-hop；

T4: if sink receives Re ACK bag, then node j is added PQue according to the time sequencing receiving Re ACK；Otherwise turn T5；

Whether T5: decision node j meets following two condition:

Condition (1): the E of node j_rem＜=E_min

Condition (2): the path at node j place is " short slab " path

T6: if meeting, then sink node motion, MFlag=true, go to step one；

T7: if being unsatisfactory for, then sink node does not moves, and is obtained the information of j by the neighbor node of node j；

T8: output PQue and TQue.

Step 3, routing table establishment stage:

R1: obtained PQue by T8, if PQue is not empty, iteration compares, path addition PQue and TQue that target function value is optimum；Turn R2；

R2: output routing table RQue；

R3: when PQue is empty, terminates iteration, turn R4；

R4: invocation step two, sets up routing table, if MFlag=true, then invocation step one；Otherwise, R5 is turned；

The link health index η of R5: calculating, if η is more than the healthy threshold value of link, then according to the RQue queue route of R2 output；Otherwise, R6 is turned；

R6: if path p_nBe " short slab " path, then MFlag=true, sink node moves towards " short slab " path, and p_nLift a ban from TQue；

R7: if path p_nNot " short slab " path, then MFlag=false；RQue route according to R2 output.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterised in that the object function described in step 3 constitutes by formula (5), (6) definition:

And meet:

Wherein,WithRepresentation unit time path p respectively_nOn node i transmission k Bit data energy consumption and update energy consumption, y_iAlong with MFlag value changes, if MFlag=true, then y_i=1, otherwise, y_i=0.f₂(p_n) represent path multiplicity, i.e. path p_mAnd p_nThe number of same node point, f₂(p_n) the least expression path is the most irreplaceable, i.e. path is the most crucial.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterised in that multiple object functions are changed into single-object problem by the method, i.e.

Formula (7) expression obtains a paths p from meeting condition path concentration_nSo that single object optimization functional value is maximum.Due to f₂(p_n) may be 0, it is nonintersecting paths, therefore the denominator of formula (7) is made to add 1 process.

The multipath Geographic routing optimization method of code-on-demand migration sink in a kind of wireless sense network the most according to claim 1, it is characterised in that described " short slab " path is the definition to link health index:

。