CN103916927A - Wireless sensor network routing method based on improved harmony search algorithm - Google Patents

Abstract

The invention provides a wireless sensor network routing method based on the improved harmony search algorithm. The method comprises the following steps: 1, initializing relevant parameters HMS, HMCR and PAR of the algorithm and the evaluation time eval-Nomax; 2, utilizing a roulette to initialize a harmony memory bank HM; 3, evaluating the fitness of harmony paths in the harmony bank; 4, setting eval-No to be equal to 0; 5, setting i to be equal to 0; 6, generating candidate harmony; 7, setting eval-No to be ++, and executing the step 8 if eval-No is smaller than eval-Nomax, or else executing the step 11; 8, carrying out neighborhood search on the ith harmony Xi equal to (s, x2, ...xj, ..., d); 9, setting eval-No to be ++, and executing the step 10 if eval-No is smaller than eval-Nomax, or else executing the step 11; 10, setting i to be ++, executing the step 6 if i is smaller than HMS, or else executing the step 5; 11, recording the optimum harmony path in the harmony memory bank. The routing method has high energy efficiency and can effectively prolong the life cycle of the whole network.

Description

A kind of wireless sensor network routing method based on improving harmony searching algorithm

Technical field

The invention belongs to wireless sensor network technology field, be specifically related to a kind of method for routing that improves harmony searching algorithm.

Background technology

Wireless sensor network (Wireless Sensor Network, WSN) and traditional internet and develop swift and violent wireless self-networking (MobileAd Hoc Network) recent years and have to a great extent different.In traditional internet and Ad Hoc network, each network node is a PC or mobile device under normal circumstances, if using internet and Ad Hoc host to be used for obtaining network, user shares information, so no matter be internet or AdHoc network, the object of Design of Routing Protocol is mainly in order to improve service quality, and needn't consider user consumes how many energy.Radio sensing network is different, radio sensing network is to disseminate a large amount of microsensor nodes in surveyed area (the very severe or people of environment cannot arrive under normal circumstances), these sensing nodes do not have fixing ID, but form a sensor network by the mode of self-organizing.Wireless sensor network, once appearance, has just obtained the favor of various circles of society, is applied in military affairs, medical treatment, the every field such as environment.

Routing algorithm plays an important role in wireless sensor network, and its energy consumption to each node, life cycle and the communication quality of whole network play critical effect.Therefore, the research of routing algorithm receives increasing concern.Because sensor network has the features such as energy constraint, resource-constrained, topologies change is frequent, therefore, traditional routing mechanism can not adapt to radio sensing network, must design routing algorithm correspondingly.

So far mainly comprise for the Routing Protocol of wireless sensor network: Routing Protocol (GAF agreement, GEAR agreement) and the Routing Protocol based on data flow and QoS of plane Routing Protocol (Flooding agreement, SPIN agreement, MTE agreement, Directed Diffusion agreement), hierarchical routing (LEACH agreement), position-based information.

At present, method for routing based on intelligent optimization algorithm has also had some application in wireless sensor network technology field, mainly based on ant group algorithm (Ant Colony Optimization, ACO) method for routing and the method for routing based on ant colony algorithm (Bee Colony Optimization, BCO).

The research of radio sensing network Routing Protocol has become focus gradually, the research of Routing Protocol is from simple to complexity, from data-centered to high-quality requirement, and towards intelligent direction development, how setting up efficient, adaptive sensing routing algorithm becomes a focus of current research.

Harmony searching algorithm is the intelligent optimization algorithm of a kind of novelty of people's propositions such as calendar year 2001 Korea S scholar Geem.Algorithm simulation in musical performance musicians rely on oneself memory, by repeatedly adjusting the tone of each musical instrument in band, finally reach the process of a beautiful harmony state.Harmony searching algorithm has very strong ability of searching optimum, is easy to converge to globally optimal solution, and its flow chart as shown in Figure 1, performs step as follows:

Step1: initialization algorithm relevant parameter.

The size of initialization HMS(harmony data base, i.e. population amount of capacity), HMCR(selects probability), PAR(adjusts probability), BW(adjusts bandwidth), evaluate number of times eval_No _max.

Step2: harmony data base (Harmony Memory, HM) initialization.

$\mathrm{HM}=\left[\begin{array}{c}{X}_1\\ X_2\\ \·\\ \·\\ \·\\ X_{\mathrm{HMS}}\end{array}\right]=\left[\begin{array}{cccc}{x}_{\mathrm{1,1}}& x_{\mathrm{1,2}}& \·\·\·& x_{1,n}\\ x_{\mathrm{2,1}}& x_{\mathrm{2,2}}& \·\·\·& x_{2,n}\\ \·& \·& & \·\\ \·& \·& \·\·\·& \·\\ \·& \·& & \·\\ x_{\mathrm{HMS},1}& x_{\mathrm{HMS},2}& \·\·\·& x_{\mathrm{HMS},n}\end{array}\right]---\left(1\right)$

In this step, the HM shown in formula (1) passes through one group of random number of interior generation is carried out initialization, wherein 1≤i≤n.Thereby obtain i variate-value of j solution vector according to formula (2):

$x_i^j=x_i^L+\mathrm{Random}\left(\mathrm{0,1}\right)\×(x_i^U-x_i^L)---\left(2\right)$

In above formula, j=1,2 ..., HMS, Random (0,1) is the random number between 0 to 1.

Step3: the fitness that calculates each harmony in harmony storehouse.

Step4: eval_No=0 is set.

Step5: extemporize candidate harmony.

Step5.1: in this step, a new harmony vector produced by three rules:

1. memory is selected;

2. select at random;

3. pitch adjustment.

1. and 2. produce candidate's harmony and be referred to as extemporize, first determine it is that memory is selected or selected at random by rule, specific as follows shown in:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_{\mathrm{ran}\left(d\right)j}& i_{j}f(P_1<\mathrm{HMCR})\\ x_j\&Element;{\mathrm{\&Omega;}}_j& \mathrm{otherwise}.\end{array}---\left(3\right)\right.$

In formula, P ₁be the random number between 0 to 1, x _{rand (i), j}be illustrated in random one-component, the Ω of selecting in the j row component of HM _jrepresent the domain of definition of j component.

Step5.2: each tone obtaining through memory selection will further be checked and determine whether to need pitch adjustment, this operation use PAR parameter, it is as follows that pitch is adjusted decision-making:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_j^{\&prime;}\&PlusMinus;\mathrm{Random}\left(\mathrm{0,1}\right)\&times;\mathrm{BW}& \mathrm{if}(P_2<\mathrm{PAR}),\\ x_j^{\&prime;}& \mathrm{otherwise}.\end{array},---\left(4\right)\right.$

In formula, P ₂it is the random number between 0 to 1.

Step5.3: the fitness value f (π) of calculated candidate harmony;

Step5.4: upgrade HM.

According to target function value, if candidate's harmony vector is better than vector the poorest in HM, new vector replaces the poorest harmony vector in HM, otherwise inoperation.

Step6: check and whether stop iteration.

Eval_No++, if eval_No < is eval_No _max, carry out Step5; Otherwise carry out Step7.

Step7: record the optimal solution in harmony storehouse.

But traditional harmony searching algorithm can not be directly used in wireless sensor network route.

Summary of the invention

Given this, the object of the invention is to propose a kind of wireless sensor network routing method based on improving harmony searching algorithm, by path fitness model is improved, make this agreement in efficiency and extend aspect whole wireless sensor network life to there is great superiority.

The present invention is by the following technical solutions to realize foregoing invention object:

Based on a wireless sensor network routing method that improves harmony searching algorithm, comprise the following steps:

Step1, initialization algorithm relevant parameter HMS, HMCR, PAR and evaluation number of times eval_No _max;

Step2, utilize roulette initialization harmony data base HM;

The fitness of each harmony (path) in Step3, calculating harmony storehouse;

Step4, eval_No=0 is set;

Step5, i=0 is set;

Step6, generation candidate's harmony also upgrade harmony data base;

Step7, eval_No++, if eval_No < is eval_No _max, carry out Step8; Otherwise carry out Step11;

Step8, to i article of harmony Xi={s in harmony storehouse, x ₂..., x _j..., d} carries out neighborhood search;

Step9, eval_No++, if eval_No < is eval_No _max, carry out Step10; Otherwise carry out Step11;

Step10, i++, if i < is HMS, carry out Step6; Otherwise carry out Step5;

Step11, record the optimum harmony (path) in harmony data base.

Compared with prior art, the present invention has following beneficial effect: routing algorithm has higher efficiency, and can extend the life cycle of whole network.Technique effect of the present invention also will further be embodied in the elaboration of embodiment.

Accompanying drawing explanation

Fig. 1 is traditional HS algorithm flow chart;

Fig. 2 is the flow chart that in the embodiment of the present invention, improved harmony searching algorithm solves WSN optimal path;

Fig. 3 is the path candidate product process figure that improves harmony searching algorithm in the embodiment of the present invention;

Fig. 4 is the Path neighborhood search routine figure that improves harmony searching algorithm in the embodiment of the present invention;

Fig. 5 is harmony library initialization schematic diagram in the embodiment of the present invention;

Fig. 6 is that in the embodiment of the present invention, candidate's harmony generates schematic diagram;

Fig. 7 is neighborhood search schematic diagram in path in the embodiment of the present invention;

Fig. 8 is the random WSN scene that generates 100 nodes in the embodiment of the present invention;

Fig. 9 is the improvement harmony searching algorithm (IHS) of the embodiment of the present invention and the contrast of the convergence situation of ant group algorithm average fitness.

Embodiment

For making the object, technical solutions and advantages of the present invention clearer, describe the present invention in detail below with reference to the accompanying drawings and in conjunction with example.

The present invention is directed to the feature of wireless sensor network route, traditional harmony searching algorithm improved, improve flow chart that harmony searching algorithm carries out WSN route as shown in Figure 2, specifically comprise following content:

(1) initialization of harmony data base

In traditional harmony searching algorithm, every harmony in harmony data base necessarily requires dimension identical, as the formula (1).In embodiments of the present invention, every harmony in harmony data base produces by roulette, and their dimension can be not identical, and the head and the tail element of every harmony is respectively source node and aggregation node, and HM as the formula (5).

$\mathrm{HM}=\left[\begin{array}{c}{X}_1\\ X_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_i\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_{\mathrm{HMS}}\end{array}\right]=\left[\begin{array}{cccccc}s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{1,j}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& d\\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{2,j}& \&CenterDot;\&CenterDot;\&CenterDot;& d& \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{i,j}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& d\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{\mathrm{HMS},j}& \&CenterDot;\&CenterDot;\&CenterDot;& d& \end{array}\right]---\left(5\right)$

In formula, s represents source node numbering, and d represents aggregation node numbering, x _i,jrepresent other sensor node numbering.

As can be seen from the above equation, the length of the each harmony in harmony storehouse (being dimension) may be not identical.

(2) generation of candidate's harmony

Candidate's harmony generation as shown in Figure 3, wherein:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_{\mathrm{rand}\left(i\right),j}& \mathrm{if}(P_1<\mathrm{HMCR})\&\&\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)\&cap;\{x_{1,j},x_{2,j},...,x_{\mathrm{HMS},j}\}\&NotEqual;\&empty;,\\ x_j^{\&prime;}\&Element;\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)& \mathrm{otherwise}.\end{array}\right.---\left(6\right)$

In formula, be illustrated in node the set of communication range interior nodes, { x _{1, j}, x _{2, j}..., x _{hMS, j}represent that the j in harmony storehouse is listed as, P ₁be the random number between 0 to 1, x _{rand (i), j}be illustrated in the random one-component of selecting in the j row component of HM, be illustrated in node communication range in random select a node.

As random number P ₁while being less than the selection probability HMCR of harmony searching algorithm, the down hop of candidate's harmony is selected from harmony data base; Otherwise, in the communication range of present node, select the node not arriving as down hop at random.

If next-hop node is taken from harmony data base, need to judge whether it needs to adjust: as random number P ₂be less than while adjusting probability P AR, the tone of taking from harmony storehouse is adjusted, in the communication range of present node, select the node replacement not arriving to fall selecteed node as down hop at random, otherwise, keep selecteed node as down hop.If next-hop node is only taken from the communication context of present node, irrelevant with harmony storehouse, without adjustment.By that analogy, until arrive aggregation node.

Wherein, to candidate's harmony in to take from the adjustment that the tone in harmony storehouse carries out as follows:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_j^{\&prime;}\&Element;\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)& \mathrm{if}(P_2<\mathrm{PAR}),\\ x_j^{\&prime;}& \mathrm{otheerwise}.\end{array}\right..---\left(7\right)$

In formula, P ₂it is the random number between 0 to 1.

(3) neighborhood search

As shown in Figure 4, to i article of harmony X in harmony storehouse _i={ s, x ₂..., x _j..., d} carries out neighborhood search, and Mathematical Modeling is as follows:

$x_j\&LeftArrow;\left\{\begin{array}{cc}{x}_j\&Element;(\mathrm{Neib}\left(x_{j-1}\right)\&cap;\mathrm{Neib}\left(x_{j+1}\right))& \mathrm{if}(\mathrm{Neib}\left(x_{j-1}\right)\&cap;\mathrm{Neib}\left(x_{j+1}\right)\&NotEqual;\&empty;),\\ x_j& \mathrm{otherwise}.\end{array},-,-,-,\left(8\right)\right.$

This field searching method can significantly improve convergence of algorithm speed.

(4) the fitness function model in path

In the embodiment of the present invention, the path fitness function model of wireless sensor network route as the formula (9), use this model can make routing algorithm have higher efficiency, and can extend the life cycle of whole network, particularly for the not identical situation of the initial gross energy of each node.

$f\left(\mathrm{\&pi;}\right)=\frac{2^{*}{(L-1)}^{*}{E_{\mathrm{elec}}}^{*}k+{E_{\mathrm{amp}}}^{*}{k}^{*}\underset{i=1}{\overset{L-1}{\mathrm{\&Sigma;}}}{d}_{i,i+1}^2}{{E_{\mathrm{Min}}}^{*}{E}_{\mathrm{Avg}}}---\left(9\right)$

In formula, molecule is radio energy consumption model, and denominator is the dump energy index of node in path.L is the length in path, E _elecfor the unit energy consumption of transmission and acceptance, value is 50nJ/bit, E _ampfor the unit energy consumption that transmission is amplified, value is 100pJ/ (bit ^*m ²), k is the data package size that source node sends, d _{i, i+1}represent the distance between i node and i+1 node, E _minrepresent the dump energy of the minimum node of dump energy in path, E _avgrepresent the average residual energy of all nodes in path.

Particularly, the method for the embodiment of the present invention comprises the following steps:

Step1, initialization algorithm relevant parameter HMS, HMCR, PAR and evaluation number of times eval_No _max.

Step2, utilize roulette initialization harmony data base HM.

While utilizing roulette initialization harmony data base, the selection probability P (i, j) of the node j in node i communication range is as follows:

$P(i,j)=\left\{\begin{array}{cc}\frac{\underset{k\&Element;{\mathrm{allowed}}_i}{\mathrm{\&Sigma;}}({\mathrm{hop}}_k/{\mathrm{hop}}_{\max }+1/E_k)-({\mathrm{hop}}_j/{\mathrm{hop}}_{\max }+1/E_j)}{{(\mathrm{No}\left({\mathrm{allowed}}_i\right)-1)}^{*}\underset{k\&Element;{\mathrm{allowed}}_i}{\mathrm{\&Sigma;}}({\mathrm{hop}}_k/{\mathrm{hop}}_{\max }+1/E_k)}& \mathrm{if}(j\&Element;{\mathrm{allowed}}_i),\\ 0& \mathrm{otherwise}.\end{array},-,-,-,\left(10\right)\right.$

In formula, hop _jrepresent the jumping figure of node j, hop _krepresent the jumping figure of node k, hop _maxrepresent the jumping figure of the node of the jumping figure maximum in all nodes, E _jrepresent the dump energy of node j, E _krepresent the dump energy of node k, allowed _iexpression can become the node set of the down hop of node i, No (allowed _i) expression set allowed _inumber of elements.

By this selection probability, can make initial path with larger probability selection near aggregation node and the more node of dump energy, the initialization schematic diagram of harmony data base is as shown in Figure 5.

The fitness f (π) of each harmony (path) in Step3, calculating harmony storehouse.

Step4, eval_No=0 is set.

Step5, i=0 is set.

Step6, generation candidate's harmony also upgrade harmony data base.

As shown in Figure 6, as can be seen from Figure 6, first node of candidate's harmony is source node to the generation schematic diagram of candidate's harmony, while selecting down hop, and r ₁< HMCR, and r ₂> PAR, therefore, algorithm selects node (the example selection node 31 here) in source node communication range as down hop from the secondary series in harmony storehouse at random; While selecting the 3rd node, r ₁> HMCR, therefore, algorithm selects a node not arriving (the example selection 33 here) as down hop at random in the communication range of node 31; The rest may be inferred, can obtain candidate's harmony as shown in Figure 6.

The fitness value f (π) of Step6.1, calculated candidate harmony;

Step6.2, harmony the poorest in candidate's harmony and HM is compared, if be better than this poorest harmony, this poorest harmony is replaced out to harmony data base.

Step7, eval_No++, if eval_No < is eval_No _max, carry out Step8; Otherwise carry out Step11.

Step8, to i article of harmony X in harmony storehouse _i={ s, x ₂..., x _j..., d} carries out neighborhood search.

Neighborhood search schematic diagram as shown in Figure 7, by the node in random selecting paths, is selected at random a node not arriving by selected node replacement, thereby is completed neighborhood search in the upper hop of selected node and the communication range of down hop common factor.For the path in figure, select the 71st node to carry out neighborhood search, as can be seen from the figure, the upper hop node 33 of node 71 and the communication range of next-hop node 80 occur simultaneously and comprise node 63,64,65,71,82, therefore, algorithm therefrom selects a node (the example selection node 63 here) substitute node 71 to complete neighborhood search at random.

The fitness value f (π) of the harmony that Step8.1, calculating neighborhood search obtain.

Step8.2, the harmony that neighborhood search is obtained compare with carrying out neighborhood search harmony before, if the harmony before being better than is replaced out harmony data base by harmony before.

Step9, eval_No++, if eval_No < is eval_No _max, carry out Step10; Otherwise carry out Step11.

Step10, i++, if i < is HMS, carry out Step6; Otherwise carry out Step5.

Step11, record the optimum harmony (path) in harmony data base.

Effect of the present invention can be verified by following emulation experiment with more further:

Choose the WSN scene shown in Fig. 8, the improvement harmony searching algorithm (IHS) in the present invention and ant group algorithm (ACO) are compared.Programming language is C++, and allocation of computer is: intel I7-3610QM processor, 8GB internal memory, 2GB solely show, the notebook computer of windows764 bit manipulation system.In scene, respectively as shown in Table 1 and Table 2, the initial gross energy of each node is as shown in table 3 for the abscissa of each node and ordinate.

The abscissa of the each node of table 1WSN

No	X-axis coordinate	No	X-axis coordinate	No	X-axis coordinate	No	X-axis coordinate	No	X-axis coordinate
										1	1038.063699	21	288.486155	41	93.880562	61	108.570707	81	332.141475
2	470.015997	22	1070.806875	42	1000.758772	62	350.735654	82	434.161440
										3	796.022940	23	663.764978	43	684.108914	63	260.472487	83	877.499586
4	760.394988	24	141.228770	44	130.028695	64	581.335254	84	927.195665
										5	451.998943	25	950.243855	45	1032.180685	65	918.536073	85	757.365042
6	73.157485	26	700.444053	46	80.002362	66	1014.104825	86	510.760722
										7	268.720797	27	551.871345	47	70.394198	67	807.038443	87	697.026192
8	18.310381	28	200.715905	48	810.720847	68	360.427584	88	1050.648601
										9	506.471469	29	40.233365	49	370.548313	69	322.964728	89	181.477631
10	830.160846	30	232.451389	50	500.487349	70	772.977461	90	657.995492
										11	225.383031	31	402.776618	51	847.995887	71	958.541697	91	746.170122
12	586.077114	32	674.172879	52	54.363618	72	1060.103491	92	676.354322
										13	180.533438	33	1083.709733	53	900.195729	73	100.013495	93	335.769779
14	316.659651	34	178.979018	54	785.705198	74	543.693628	94	407.495452
										15	615.102143	35	760.686848	55	499.784261	75	119.200772	95	164.825008
16	825.980727	36	1004.310202	56	617.938018	76	276.033431	96	880.467342
										17	936.457298	37	518.066244	57	450.612913	77	978.149229	97	560.717385
18	191.847072	38	150.599881	58	200.698715	78	172.276119	98	374.317783
										19	467.889095	39	850.394587	59	610.976658	79	103.613207	99	815.726448
20	966.266690	40	500.090378	60	1025.537819	80	599.772413	100	236.224196

The ordinate of the each node of table 2WSN

No	Y-axis coordinate	No	Y-axis coordinate	No	Y-axis coordinate	No	Y-axis coordinate	No	Y-axis coordinate
										1	881.514269	21	199.332398	41	321.105262	61	426.254537	81	1010.474719
2	900.139546	22	1069.791995	42	700.612919	62	424.629072	82	650.299109
										3	899.920884	23	801.707617	43	358.620706	63	412.729792	83	946.840883
4	410.732085	24	200.400755	44	125.729568	64	433.350737	84	450.859689
										5	385.646460	25	911.142851	45	978.116910	65	682.606005	85	677.851248
6	66.067060	26	900.857773	46	740.034856	66	1063.605136	86	185.154071
										7	913.267222	27	1000.791881	47	801.882039	67	226.182588	87	995.057549
8	482.812712	28	680.135671	48	805.298556	68	343.500789	88	788.639293

9	614.032108	29	595.278889	49	910.798429	69	288.815268	89	871.881713
										10	721.787934	30	100.733508	50	460.712389	70	303.128055	90	115.568055
11	1000.455846	31	501.791736	51	101.356461	71	186.416999	91	770.734480
										12	560.215866	32	580.014018	52	374.851807	72	20.033960	92	496.312489
13	775.619619	33	111.453285	53	863.584556	73	525.222154	93	111.029792
										14	650.651168	34	486.843473	54	133.825872	74	267.788594	94	212.913741
15	952.032544	35	545.073400	55	339.555109	75	945.644309	95	350.252580
										16	465.951681	36	150.193973	56	1047.642382	76	521.872265	96	250.712968
17	126.031318	37	693.612356	57	750.816902	77	982.082183	97	791.564358
										18	166.525170	38	50.084703	58	272.840217	78	1065.785511	98	830.837048
19	253.999471	39	350.309441	59	710.146587	79	646.616879	99	615.348936
										20	780.701144	40	531.949780	60	90.656761	80	179.078778	100	583.095624

The primary power (unit: J) of the each node of table 3WSN

No	Primary power	No	Primary power	No	Primary power	No	Primary power	No	Primary power
										1	189.099	21	100.381	41	151.735	61	119.724	81	155.074
2	129.258	22	153.197	42	102.072	62	199.359	82	140.593
										3	164.165	23	187.918	43	179.553	63	159.133	83	146.913
4	156.905	24	192.096	44	161.199	64	182.574	84	124.812
										5	145.439	25	141.026	45	125.343	65	157.619	85	194.046
6	152.181	26	100.421	46	101.685	66	185.379	86	147.032
										7	147.108	27	181.948	47	128.184	67	126.374	87	147.581
8	170.327	28	173.424	48	146.522	68	131.690	88	192.737
										9	166.277	29	195.279	49	154.796	69	177.456	89	190.847
10	162.810	30	177.532	50	119.761	70	165.133	90	123.875
										11	126.447	31	127.323	51	166.588	71	146.883	91	108.466
12	106.641	32	118.311	52	117.698	72	158.251	92	156.359
										13	110.291	33	184.161	53	189.349	73	131.391	93	199.634
14	143.953	34	140.440	54	150.172	74	108.939	94	183.139
										15	127.235	35	149.321	55	138.179	75	179.272	95	120.127
16	120.252	36	121.894	56	119.306	76	127.662	96	114.154
										17	175.655	37	159.423	57	120.710	77	149.406	97	164.483
18	166.973	38	172.442	58	191.479	78	198.212	98	124.055
										19	187.793	39	132.496	59	176.263	79	115.293	99	140.373
20	189.001	40	177.221	60	114.594	80	106.696	100	176.168

The node of ant group algorithm is selected Probability p _k(r, s) is as follows:

$p_k(r,s)=\left\{\begin{array}{cc}\frac{{\left[T(r,s)\right]}^{\mathrm{\&alpha;}}{\left[E\left(s\right)\right]}^{\mathrm{\&beta;}}}{\underset{u\&NotElement;M_k}{\mathrm{\&Sigma;}}{\left[T(r,u)\right]}^{\mathrm{\&alpha;}}{\left[E\left(u\right)\right]}^{\mathrm{\&beta;}}}& \mathrm{if}(s\&NotElement;M_k),\\ 0& \mathrm{otherwise}.\end{array}\right.---\left(11\right)$

In formula, T (r, s) is the pheromones on path (r, s), and E (s) is the dump energy of node s, M _kit is the taboo list of ant k.

Pheromones Renewal model is as follows:

${\mathrm{\&Delta;T}}_k=\frac{1}{{(2-{\mathrm{EMin}}_k/{\mathrm{EAvg}}_k)}^{*}{\mathrm{Fd}}_k}---\left(12\right)$

In formula, EMin _kbe the energy of the minimum node of dump energy in k paths, EAvg _kbe the average residual energy of node in k paths, Fd _kit is the number of nodes of k paths.

This pheromones Renewal model had both considered that efficiency also considered the life cycle of whole network, and for ant group algorithm, the effect of gained is better.

For the WSN scene shown in Fig. 8, the relative parameters setting of improving harmony searching algorithm and ant group algorithm is as shown in table 4, and in table, the algorithm parameter of two kinds of algorithms is the nearly figure of merit; The fitness function model in path as the formula (9).

The relative parameters setting of two kinds of algorithms of table 4

In Fig. 8, the node that five-pointed star represents is source node, and the node that inverted triangle represents is aggregation node.More as shown in Figure 9, transverse axis represents iterations to the operation result of two kinds of algorithms, and the longitudinal axis represents the average fitness average of the population of per generation after 50 independent operatings.

As can be seen from Figure 9, the effect of IHS algorithm will be significantly better than the effect of ACO algorithm, and the path that IHS algorithm obtains is more energy-conservation, and in path, the dump energy of node is more.

The optimal path fitness of IHS algorithm gained is 8.88593e-006, the energy that path consumes is 0.136J, the jumping figure in path is 15, and corresponding optimal path is: 6 → 44 → 18 → 21 → 69 → 68 → 5 → 50 → 64 → 92 → 35 → 99 → 10 → 65 → 20 → 88.

The optimal path fitness of ACO algorithm gained is 9.27144e-006, the energy that path consumes is 0.143J, the jumping figure in path is 17, and corresponding optimal path is: 6 → 38 → 30 → 93 → 21 → 69 → 68 → 5 → 50 → 64 → 92 → 35 → 99 → 10 → 65 → 20 → 42 → 88.

Therefore the optimal path of IHS algorithm gained can be saved 4.9% energy compared with the optimal path of ACO algorithm gained, and can reduce by 11.8% jumping figure, propagation delay time is less.

More than prove that IHS algorithm, the in the situation that of assigned source node, can find better path than ACO algorithm.In order to compare the performance situation of two kinds of algorithms at whole network lifecycle, utilize two kinds of algorithms respectively the WSN scene shown in Fig. 8 to be carried out to route circulation, the initial gross energy of each node is as shown in table 3, relative parameters setting is as shown in table 5, record respectively the periodicity that first node death is experienced, result is as shown in table 6.

The relative parameters setting of two kinds of algorithms of table 5

Two kinds of algorithms of table 6 are in the comparison of WSN life cycle

As can be seen from Table 6, compared with utilizing ACO algorithm route, utilize IHS algorithm to carry out route and can make the life cycle of whole WSN extend 40.8%.Therefore, IHS algorithm is better than ACO algorithm.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any modifications of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.

Claims (10)

1. the wireless sensor network routing method based on improving harmony searching algorithm, comprises the following steps:

Step1, initialization relevant parameter HMS and evaluation number of times eval_No _max;

Step2, utilize roulette initialization harmony data base HM;

The fitness f (π) of each harmony (path) in Step3, calculating harmony storehouse;

Step4, eval_No=0 is set;

Step5, i=0 is set;

Step6, generation candidate's harmony also upgrade harmony data base;

Step7, eval_No++, if eval_No < is eval_No _max, carry out Step8; Otherwise carry out Step11;

Step8, to i article of harmony X in harmony storehouse _i={ s, x ₂..., x _j..., d} carries out neighborhood search;

Step9, eval_No++, if eval_No < is eval_No _max, carry out Step10; Otherwise carry out Step11;

Step10, i++, if i < is HMS, carry out Step6; Otherwise carry out Step5;

Step11, record optimum and acoustic path in harmony data base.

2. method according to claim 1, in described Step2, the selection probability P (i, j) of the node j in node i communication range is as follows:

$P(i,j)=\left\{\begin{array}{cc}\frac{\underset{k\&Element;{\mathrm{allowed}}_i}{\mathrm{\&Sigma;}}({\mathrm{hop}}_k/{\mathrm{hop}}_{\max }+1/E_k)-({\mathrm{hop}}_j/{\mathrm{hop}}_{\max }+1/E_j)}{{(\mathrm{No}\left({\mathrm{allowed}}_i\right)-1)}^{*}\underset{k\&Element;{\mathrm{allowed}}_i}{\mathrm{\&Sigma;}}({\mathrm{hop}}_k/{\mathrm{hop}}_{\max }+1/E_k)}& \mathrm{if}(j\&Element;{\mathrm{allowed}}_i),\\ 0& \mathrm{otherwise}.\end{array}\right.$

In formula, hop _jrepresent the jumping figure of node j, hop _krepresent the jumping figure of node k, hop _maxrepresent the jumping figure of the node of the jumping figure maximum in all nodes, E _jrepresent the dump energy of node j, E _krepresent the dump energy of node k, allowed _iexpression can become the node set of the down hop of node i, No (allowed _i) expression set allowed _inumber of elements.

3. method according to claim 1, in described Step6, described candidate's harmony , wherein:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_{\mathrm{rand}\left(i\right),j}& \mathrm{if}(P_1<\mathrm{HMCR})\&\&\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)\&cap;\{x_{1,j},x_{2,j},...,x_{\mathrm{HMS},j}\}\&NotEqual;\&empty;,\\ x_j^{\&prime;}\&Element;\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)& \mathrm{otherwise}.\end{array}\right.$

In formula, s represents source node numbering, and d represents aggregation node numbering, be illustrated in node the set of communication range interior nodes, { x _{1, j}, x _{2, j}..., x _{hMS, j}represent that the j in harmony storehouse is listed as, and P1 is the random number between 0 to 1, HMCR is for selecting probability, x _{rand (i), j}be illustrated in the random one-component of selecting in the j row component of HM, be illustrated in node communication range in random select a node.

4. method according to claim 3, in described Step6:

As random number P ₁while being less than the selection probability HMCR of harmony searching algorithm, the down hop of candidate's harmony is selected from harmony data base; Otherwise, in the communication range of present node, select the node not arriving as down hop at random;

If next-hop node is taken from harmony data base, whether judgement is got needs to adjust: as random number P ₂be less than while adjusting probability P AR, the tone of taking from harmony storehouse is adjusted, in the communication range of present node, select the node replacement not arriving to fall selecteed node as down hop at random, otherwise, keep selecteed node as down hop; Until arrival aggregation node, wherein, P ₂it is the random number between 0 to 1.

5. method according to claim 4, is describedly specially the tone adjustment of taking from harmony storehouse:

$x_j^{\&prime;}\&LeftArrow;\left\{\begin{array}{cc}{x}_j^{\&prime;}\&Element;\mathrm{Neib}\left(x_{j-1}^{\&prime;}\right)& \mathrm{if}(P_2<\mathrm{PAR}),\\ x_j^{\&prime;}& \mathrm{otheerwise}.\end{array}\right..$

6. method according to claim 1, described Step6 comprises:

The fitness value f (π) of Step6.1, calculated candidate harmony;

Step6.2, harmony the poorest in candidate's harmony and HM is compared, if be better than this poorest harmony, this poorest harmony is replaced out to harmony data base.

7. method according to claim 1, in Step8, by the node in random selecting paths, in the upper hop of selected node and the communication range of down hop common factor, select at random a node not arriving by selected node replacement, thereby complete neighborhood search.

8. method according to claim 1, Step8 specifically comprises:

The fitness value f (π) of the harmony that Step8.1, calculating neighborhood search obtain;

Step8.2, the harmony that neighborhood search is obtained compare with carrying out neighborhood search harmony before, if the harmony before being better than is replaced out harmony data base by harmony before.

9. method according to claim 1, wherein, the head and the tail element of every harmony in harmony data base is respectively source node and aggregation node, HM is as follows for harmony data base:

$\mathrm{HM}=\left[\begin{array}{c}{X}_1\\ X_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_i\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ X_{\mathrm{HMS}}\end{array}\right]=\left[\begin{array}{cccccc}s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{1,j}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& d\\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{2,j}& \&CenterDot;\&CenterDot;\&CenterDot;& d& \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{i,j}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& d\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& & \\ s& \&CenterDot;\&CenterDot;\&CenterDot;& x_{\mathrm{HMS},j}& \&CenterDot;\&CenterDot;\&CenterDot;& d& \end{array}\right]$

In formula, s represents source node numbering, and d represents aggregation node numbering, x _i,jrepresent other sensor node numbering.

10. method according to claim 1, described fitness function model is:

$f\left(\mathrm{\&pi;}\right)=\frac{2^{*}{(L-1)}^{*}{E_{\mathrm{elec}}}^{*}k+{E_{\mathrm{amp}}}^{*}{k}^{*}\underset{i=1}{\overset{L-1}{\mathrm{\&Sigma;}}}{d}_{i,i+1}^2}{{E_{\mathrm{Min}}}^{*}{E}_{\mathrm{Avg}}},$

In formula, the length that L is path, E _elecfor the unit energy consumption of transmission and acceptance, E _ampfor the unit energy consumption that transmission is amplified, k is the data package size that source node sends, d _{i, i+1}represent the distance between i node and i+1 node, E _minrepresent the dump energy of the minimum node of dump energy in path, E _avgrepresent the average residual energy of all nodes in path.