CN102612080B - Minimum-energy-consumption self-adaption convergent routing method on basis of second-generation wavelet zero tree encoding - Google Patents

Abstract

The invention discloses a minimum-energy-consumption self-adaption convergent routing method on the basis of second-generation wavelet zero tree encoding, which belongs to the technical field of energy-efficient routing of the wireless sensor network. The minimum-energy-consumption self-adaption convergent routing method includes: firstly building an alpha-balance generation tree with performance between the minimum generation tree and the shortest-path tree as an initial transmission path of data; then transmitting node data according to the initial path, and self-adaptively judging whether transmission paths are converged and adjusted correspondingly or not according to convergent gains during transmission; and finally, compressing data subjected to convergent process by second-generation wavelet zero tree so as to reduce data transmission. By the method, aiming to the wireless sensor network of monitoring data which are complicated data such as voice, images or videos and the like, energy-efficient routing can be built and network survival time can be prolonged on the condition of necessary data convergence and convergence expense which cannot ignore.

Description

Least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image

Technical field

The present invention relates to the convergent routing method of wireless sensor network, refer to particularly a kind of least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image.

Background technology

Wireless sensor network node is all very limited in the resource of the each side such as energy, computing capability, memory capacity and communication bandwidth, and the life span of how to save energy, raising sensor network is the matter of utmost importance of Design of Wireless Sensor Network.Be in the sensor network of the complex datas such as sound, image or video in Monitoring Data, the Monitoring Data that each node is collected sends to Sink node (aggregation node) independently, then carry out data processing and can cause information transfer efficiency significantly to reduce, waste a large amount of node energies and communication bandwidth.For fear of the problems referred to above, must in sending the process of data, network node adopt convergence technology to process the Monitoring Data collecting.Converging of Monitoring Data is the computational process of a more complicated, especially network need to gather the multi-medium data of more complicated time, and converging energy consumption can not be left in the basket.Existing experimental result shows converging more than energy consumption often reaches 10nJ/bit of voice signal, and view data converges to consume energy and reaches 75nJ/bit.In sensor network, the energy consumption of node in the time receiving data is about 50nJ/bit.Therefore in this class network, converge expense suitable with communication overhead, even if carry out the simplest convergence processing, converging energy consumption can not be left in the basket.But the classical routing algorithm that converges all does not consider to converge energy consumption, as GIT, PEGASIS and TEEN etc.

Converging in the network that expense is larger, if the degrees of fusion between joint place data is less, the minimizing of the transport overhead that convergence brings may can't make up the energy that convergence processing itself consumes, convergence not only there is no need so, but also increase network energy consumption, so need to consider the transport overhead of network and converge expense, select optimum adaptive convergence route, make network total energy consumption minimum, at present also less to the research of this class routing algorithm.Bhong etc. have proposed a kind of DAGP and have converged routing algorithm, studied and converged expense and transport overhead to converging the impact of route, but this method is for the sensor network of periodicity reported data inapplicable.Luohong etc. have proposed a kind of adaptive convergence routing algorithm AFST that considers transmission and converge expense.But the transmission topological structure of this algorithm take binary tree as Foundation network node, has seriously increased hop count and the transmission range of network data, has affected the further optimization of network performance.

Summary of the invention

The object of the invention is to overcome above-mentioned the deficiencies in the prior art and a kind of least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image is provided, the method considers and converges expense and communication overhead, reduce volume of transmitted data, saved network energy consumption.

The technical scheme that realizes the object of the invention employing is: a kind of least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image, comprises the following steps:

(1) establish network G=(V, E) limit e=(u in, v) network energy consumption when weights omega (e) is unit of transfer's data, wherein V represents all source nodes and Sink node, E represents the transmission link between institute's active node and Sink node;

(2) utilize respectively Prime algorithm and dijkstra's algorithm to calculate minimum spanning tree and the shortest path tree take Sink node as root node;

(3) α-balance spanning tree of performance of structure between described minimum spanning tree and shortest path tree is as the initial transmission path of network data;

(4), in described α-balance spanning tree, if node v is short transmission path to the path of Sink node, make the judgement mark θ (v)=0 of this node v; Otherwise, make the judgement of this node v identify θ (v)=1;

(5) in the process of transfer of data, for the node of θ (v)=0, calculate it and converge benefit, and carry out adaptive convergence, then the data after utilizing zerotree encoding algorithm to adaptive convergence are compressed, data after compression are directly sent to Sink node along short transmission path, carry out decompress(ion) at Sink Nodes;

(6) for the node of θ (v)=1, if data are benefited and are greater than 0 converging of v place, carry out convergence processing, utilize zerotree encoding algorithm to compress the data after converging, and press transmission path and send to Sink node; Otherwise, node v sends to next-hop node by after himself data compression by transmission path, and each child node of node v is obtained shortest path limit separately from shortest path tree, its data is utilized zerotree encoding algorithm compression by each child node, directly be sent to Sink node by shortest path limit separately, carry out decompress(ion) at Sink Nodes.

In technique scheme, in step (3), structure α-balance spanning tree comprises the following steps:

(3-1) value of given α, and α > 1, each node v in minimum spanning tree distributes a variable d (v), for record v to the path weight value of root node with, when initial, to the each non-root node v in minimum spanning tree, make d (v)=∞; To root node r=Sink, make d (r)=0;

(3-2) traversal minimum spanning tree, to each node v, judge this node in minimum spanning tree to the path weight value of root node and whether exceed v in shortest path tree to the path weight value of root node and α doubly: if exceeded, v in shortest path tree is added in current network to the limit, path of root node, the value of adjusting d (v) be in shortest path tree v to the path weight value of root node with; Otherwise, keep the path of v constant;

(3-3) in follow-up traversal, if the path weight value of certain node v and value d (v) diminish, while traversing so the node u being directly connected with v, node u is done to following adjustment to the path weight value of Sink node and the value of d (u) and transmission path: if d (u) > ω is (u, v)+d (v), d (u) is adjusted into d (u)=ω (u, v)+d (v), node u is u → v → path (v to the route adjust of r, r), wherein path (v, r) be illustrated in current minimum spanning tree v to the path of root node r, otherwise node u remains unchanged to the path weight value of Sink node and d (u) and transmission path, after all nodes of minimum spanning tree have traveled through, the spanning tree structure after adjusting is α-balance spanning tree.

Further, in α-balance spanning tree, the value of α is

In technique scheme, converging benefits draws by following steps:

(4-1) in the time that node v only has a child node u, converge benefit through type (1) and calculate,

$\mathrm{\Δ}(u,v)=\mathrm{\Δ}{E}_T-E_A=[(m_0\left(u\right)+m_0\left(v\right))-\tilde{m}\left(v\right)]T(v,S)-\left[(m_0\left(u\right)+m_0\left(v\right))\right]q\left(e\right)---\left(1\right)$

Wherein, Δ E _tbe illustrated in and on this limit, carry out the transmission energy consumption of saving after convergence, E _arepresent to converge energy consumption, m ₀and m (u) ₀(v) represent respectively to converge the data volume that front nodal point u and v self gather, T (v, S) representation unit data are from node v along the transmission energy consumption that converges tree path and be sent to Sink, q (e) representation unit data volume converge energy consumption,

for the data volume after convergence,

can calculate by through type (2),

$\tilde{m}\left(v\right)=\max (m_0\left(u\right),m_0\left(v\right))+\min (m_0\left(u\right),m_0\left(v\right))(1-{\mathrm{\ρ}}_{\mathrm{uv}})---\left(2\right)$

Wherein, ρ _uvrepresent the coefficient correlation between node u and v;

(4-2) when node v has the individual child node u of K (K>=2) ₁, u ₂..., u _ktime, adopt the method progressively converging to obtain the converge benefit of each child node at node v place, first by the data volume after formula (2) convergence

obtain and converge benefit Δ (u according to formula (1) ₁, v); Then order

can obtain respectively child node u according to formula (2) and formula (1) ₂arrive after v the data volume after converging

with converge benefit Δ (u ₁, u ₂, v), repeat above-mentioned steps, obtain all child nodes and converge benefit Δ (u after v converges ₁, u ₂..., u _k, v).

In technique scheme, described adaptive convergence judgement is as follows:

(1) if converging to benefit is greater than 0, converge at node v place;

(2) if converging to benefit is not more than 0, directly data are sent to Sink node by transmission path.

The inventive method has considered and has converged expense and transport overhead, is only greater than zero node and converges converging to benefit, and has avoided the unnecessary expense that converges, and has solved tradition and has converged and in routing algorithm, converge the too much problem of number of times.Secondly, the data acquisition of this method after to adaptive convergence use the zerotree encoding algorithm (EZC-SGW) that decomposes based on Second Generation Wavelets by processings afterwards data compress, further reduced volume of transmitted data, saved network energy consumption.The present invention is adapted to dissimilar sensor network, is particularly suitable in sensor network that Monitoring Data is sound, image or video.

Accompanying drawing explanation

Fig. 1 is the operational flowchart that the present invention is based on the least energy consumption adaptive convergence method for routing of Second Generation Wavelets zerotree image;

Fig. 2 is the energy consumption testing figure that the present invention adopts EZC-SGW encryption algorithm;

Fig. 3 is the simulation result performance comparison diagram of the present invention and other related algorithms under Omnet++ environment.

Embodiment

Below in conjunction with the drawings and specific embodiments, the present invention is further illustrated.

As shown in Figure 1, the least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image comprises the following steps:

Step S101: netinit setting.

If network G=(V, E) limit e=(u in, v) network energy consumption when weights omega (e) is unit of transfer's data, wherein V represents all source nodes and Sink node, E represents the transmission link between institute's active node and Sink node; The present invention is considering under the prerequisite of convergence energy consumption, find a transmission paths, energy consumption minimum while making all nodes, along this path, data are sent to Sink node, also find a subgraph G who comprises all information source nodes and Sink node of wireless sensor network G=(V, E) ^*=(V ^*, E ^*), make

reach minimum, E _a(e) for converging energy consumption, E _t(e) be transmission energy consumption.

Step S102: utilize respectively Prime algorithm and dijkstra's algorithm to calculate minimum spanning tree MST and the shortest path tree SPT take Sink node as root node.

Step S103: structure α-balance spanning tree (α-BST).

In convergence route, if the degrees of fusion between joint data is all 100%, to converge tree be minimum spanning tree MST take Sink as root node to least energy consumption; If the degrees of fusion between joint data is all 0, to converge tree be shortest path tree SPT take Sink as root node to least energy consumption.In general wireless sensor network, the degrees of fusion between node Monitoring Data is normally between 0 and 1.Therefore, the present invention constructs the α-BST of a performance between minimum spanning tree SPT and shortest path tree MST as initial transmission path, to adapt to dissimilar sensor network.

The structure of α-BST comprises the following steps:

(1) value of given α > 1, in the present embodiment,

time α-BST combination property optimum.

(2) be that each node v in minimum spanning tree MST distributes variable d (v), be used for recording v to the path weight value of root node with.When initial, to the each non-root node v in MST, make d (v)=∞; To root node r=Sink, make d (r)=0.

(3) traversal minimum spanning tree MST, in ergodic process to each node v, judge this node in minimum spanning tree MST to the path weight value of root node and whether exceed v in shortest path tree SPT to the path weight value of root node and α doubly: if exceeded, v in shortest path tree SPT is added in current network to the limit, path of root node, the value of adjusting d (v) be in shortest path tree v to the path weight value of root node with; Otherwise, keep the path of v constant.

(4) in follow-up traversal, if the path weight value of certain node v and value d (v) diminish, while traversing so the node u being directly connected with v, u is arrived to the path weight value of Sink and the value of d (u) and transmission path and do following adjustment: if d (u) > ω is (u, v)+d (v), d (u) is adjusted into d (u)=ω (u, v)+d (v), u is u → v → path (v to the route adjust of r, r), wherein path (v, r) be illustrated in current minimum spanning tree MST v to the path of root node r, otherwise u remains unchanged to the path weight value of Sink and d (u) and transmission path.After all nodes of minimum spanning tree MST have traveled through, the spanning tree structure after note is adjusted is α-BST.

Below illustrate in two steps that the performance of α-BST of above-mentioned structure is between shortest path tree SPT and minimum spanning tree MST, and work as

time α-BST combination property optimum.

(a) in α-BST the weight on all limits and be no more than minimum spanning tree MST weight and $(1+\frac{2}{\mathrm{\α}-1})$ Doubly.

If v ₀=Sink.In the process of traversal minimum spanning tree MST, all nodes can be divided into two classes: a class is the node that has been added shortest path limit, and another kind of is all the other nodes.Suppose that total K node has been added shortest path limit, is designated as respectively v ₁, v ₂..., v _k, as node v _iafter (1≤i≤K) shortest path limit in shortest path tree SPT is added, it in current balanced tree structure to the path weight value of root node with equal SP (v _i, S).If v _ito the path weight value of Sink be d (v _i), investigate and work as v _i-1shortest path limit be added rear to d (v _i) impact: due to v _i-1be added shortest path limit, so when having access to node v _itime d (v _i) may be adjusted, according to the building method of α-BST, if d (v now _i)≤SP (v _i-1, S) and+MS (v _i-1, v _i), d (v _i) should remain unchanged, wherein MS (v _i-1, v _i) be illustrated in v in MST _ito v _i-1path weight value and; If d is (v _i) > SP (v _i-1, S) and+MS (v _i-1, v _i), d (v _i) should be adjusted into d (v _i)=SP (v _i-1, S) and+MS (v _i-1, v _i), therefore, work as v _i-1shortest path limit be added after, d (v _i) necessarily meet

d(v _i)≤SP(v _i-1，S)+MS(v _i-1，v _i)， (3)

Because node v _iafter accessed, its shortest path limit is finally added in current network, so the d (v after known adjustment _i) also necessarily meet

d(v _i)＞αSP(v _i，S) (4)

By formula (3) and the known α SP of formula (4) (v _i, S) and < SP (v _i-1, S) and+MS (v _i-1, v _i).

Above formula is added up and transplanted from 1 to K

$(\mathrm{\&alpha;}-1)\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\mathrm{SP}(v_i,S)<\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\mathrm{MS}(v_{i-1},v_i)---\left(5\right)$

Because in the time of depth-first traversal every limit of minimum spanning tree all accessed twice, therefore

$\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\mathrm{MS}(v_{i-1},v_i)\&le;2\mathrm{\&omega;}\left(\mathrm{MST}\right)---\left(6\right)$

Association type (5) and formula (6) can obtain

Therefore, the weight of last obtained α-BST and in α-BST the weight on all limits and be no more than minimum spanning tree MST weight and

doubly.

(b) from the construction process of α-BST, each source node v to the path weight value of Sink node and be no more than this node in shortest path tree SPT to Sink node path weight and α doubly.Order α and β represent respectively α-BST and SPT and the MST similarity degree in performance, comprehensively weigh the performance of α-BST with γ=alpha+beta.In the structure that converges route, in the time that γ reaches minimum value, it is optimum that the combination property of α-BST reaches, known by calculating, when

time, the value minimum of γ.So the present invention with

time balance spanning tree as the initial transmission path of adaptive convergence routing algorithm.

Step S104: in described α-balance spanning tree, if node v is short transmission path to the path of Sink node, make the judgement mark θ (v)=0 of this node v; Otherwise, make the judgement of this node v identify θ (v)=1;

Step S105: in the process of transfer of data, for the node of θ (v)=0, calculate it and converge benefit, and carry out adaptive convergence, then the data after utilizing zerotree encoding algorithm to adaptive convergence are compressed, data after compression are directly sent to Sink node along short transmission path, carry out decompress(ion) at Sink Nodes; For the node of θ (v)=1, if data are benefited and are greater than 0 converging of v place, carry out convergence processing, utilize zerotree encoding algorithm to compress the data after converging, and press transmission path and send to Sink node; Otherwise, node v sends to next-hop node by after himself data compression by transmission path, and each child node of node v is obtained shortest path limit separately from shortest path tree SPT, its data is utilized zerotree encoding algorithm compression by each child node, directly be sent to Sink node by shortest path limit separately, carry out decompress(ion) at Sink Nodes.

Above-mentioned convergence is benefited and is calculated by following steps:

In wireless sensor network, on limit e=(u, v), converge benefit Δ (u, v)=Δ E _t-E _a, wherein Δ E _tbe illustrated in and on this limit, carry out the transmission energy consumption of saving after convergence, E _arepresent to converge energy consumption, the data of node u and v converge at v.The minimizing of converging rear data volume determined by two internodal data dependences, known according to spatial data correlation models, node u, the correlation coefficient ρ between v _uvfor: ρ _uv=0, as d > r _ctime;

as d≤r _ctime.Wherein r _srepresent the sensing range of node, r _c=2r _srepresent internodal relevant range, d represents two internodal distances.

Converge the calculating of benefit in two kinds of situation:

(1) in the time that v only has a child node u, converge benefit through type (1) and calculate,

$\mathrm{\&Delta;}(u,v)=\mathrm{\&Delta;}{E}_T-E_A=[(m_0\left(u\right)+m_0\left(v\right))-\tilde{m}\left(v\right)]T(v,S)-\left[(m_0\left(u\right)+m_0\left(v\right))\right]q\left(e\right)---\left(1\right)$

Wherein T (v, S) representation unit data are sent to the transmission energy consumption of Sink from node v along converging tree path, q (e) representation unit data volume converge energy consumption, m ₀and m (u) ₀(v) and respectively represent to converge the data volume of front nodal point u and v self collection, represent the data volume after converging,

can calculate by through type (2),

$\tilde{m}\left(v\right)=\max (m_0\left(u\right),m_0\left(v\right))+\min (m_0\left(u\right),m_0\left(v\right))(1-{\mathrm{\&rho;}}_{\mathrm{uv}})---\left(2\right)$

Wherein, ρ _uvrepresent the coefficient correlation between node u and v;

(2) when v has the individual child node u of K (K>=2) ₁, u ₂..., u _ktime, adopt the method progressively converging to ask the converge benefit of each child node at node v place: can obtain the data volume after convergence according to formula (2)

can obtain and converge benefit Δ (u according to formula (1) ₁, v); Order

can obtain respectively child node u according to formula (2) and formula (1) ₂arrive after v the data volume after converging

with converge benefit Δ (u ₁, u ₂, v); By that analogy, can obtain all child nodes and converge benefit Δ (u after v converges ₁, u ₂..., u _k, v).

According to the benefit that converges of above-mentioned gained, as follows to the judgement of adaptive convergence:

(1) if converging to benefit is greater than 0, converges and can reduce network energy consumption at node v place;

(2) if converging to benefit is not more than 0, directly data are sent to Sink by transmission path.

If the data of the child node u of node v do not need to converge at v, in transmission path so afterwards, the data of u do not need to converge yet again.

In α-BST, establish u and send the path of data to Sink and be: u → v → w → ... → Sink.Because node u does not need to converge through judgement at v place, so converge benefit Δ (u, v) < 0.From the character of minimum spanning tree, u is less to the distance of other nodes on minimum spanning tree than u to the distance of v, so necessarily there is d (u, w) >=d (u, v).Because coefficient correlation is along with the increase of euclidean distance between node pair reduces, so ρ _uw≤ ρ _uv.From the computational methods that converge benefit, converging benefits reduces along with reducing of coefficient correlation, so Δ (u, w)≤Δ (u, v) < 0, be u does not need to converge in w place and later transmission path, can directly send data to Sink yet.

Calculate the energy consumption of EZC-SGW algorithm self below, and illustrate that the present invention adopts EZC-SGW encryption algorithm to carry out data compression and can save network energy.

The present invention selects the test platform of Sim-panalyzer system as EZC-SGW compression algorithm energy consumption.Energy consumption when CPU work comprises static energy consumption and dynamic energy consumption two parts, and common static energy consumption is certain value, and dynamic energy consumption is relevant with complexity and the deal with data of handling procedure.CPU dynamic energy consumption E _totaladopt following universal measurement model:

$E_{\mathrm{total}}=C_{\mathrm{total}}{V}_{\mathrm{dd}}^2+V_{\mathrm{dd}}\left(I_0{e}^{\frac{V_{\mathrm{dd}}}{{\mathrm{nV}}_T}}\right)\left(\frac{N}{f}\right)$

Wherein, N is illustrated in the periodicity that while carrying out given program task, CPU need to move, C _totalrepresent switching capacity amount total in performed algorithm, general and N is proportional.F represents the clock frequency of CPU, V _ddrepresent the kernel supply power voltage of CPU, f and V _ddcan configure voluntarily as required, the present invention gets f=100MHz, V _dd=1.5V.I ₀with n be the kernel parameter of CPU, be set to respectively I ₀=1.196mA, n=21.26.In test, select respectively the gray scale test pattern of 5 width standards as node perceived data, the size of image is risen to 512 × 512 pixels from 128 × 128 pixels, rate of rise is 16 × 16 pixels, the variation of analog node perception data amount, the experimental data of test pattern gained is got to average, as final experimental result.Experimental result shows, the corresponding increase along with the increase of input data volume of the total energy consumption of EZC_SGW encryption algorithm, and the compression/decompression energy consumption E of unit data quantity _com=E _decom≈ 40nJ/bit, as shown in Figure 2.

The following describes node data transmits one again and saves surely network energy after EZC-SGW compression.

In the present invention, node can be divided into two classes according to the subsequent transmission mode of node data: a class is that data are continued to send by the transmission path that converges route, in follow-up transmitting procedure, need to proceed to converge to data; Another kind of is that data are directly sent to Sink node by shortest path, in follow-up transmitting procedure, all no longer converges.If the compression ratio of EZC-SGW algorithm is τ, in the time of Lossless Compression, 6≤τ≤8; In the time of lossy compression method, 20≤τ≤40.The data volume of node v is k, and the distance between node v and u is d.

If node v belongs to above-mentioned first kind node.After data are compressed in node v, in the next-hop node u of v, need just can converge after decompress(ion).Transmission energy consumption E while sending to u from v before now needing comparing data to compress ₀whether (v, u), be greater than transmission energy consumption and compression/decompression energy consumption sum E after compression ₁(v, u).If E ₀(v, u) > E ₁(v, u), shows that compression is effective; Otherwise, show that compression is invalid.Adopt free-space communication model, if the data of node v are not compressed, directly send to node u, the network energy consuming is:

E ₀(v，u)＝E _Tx-elec(k)+E _Tx-amp(k，d)+E _Rx(k)＝k·(2E _elec+ε _fsd ²)

By the data acquisition of node v with being sent to the network energy that u consumes after EZC-SGW compression be:

$E_1(v,u)=E_{\mathrm{Tx}-\mathrm{elec}}(k/\mathrm{\&tau;})+E_{\mathrm{Tx}-\mathrm{amp}}(k/\mathrm{\&tau;},d)+E_{\mathrm{Rx}}(k/\mathrm{\&tau;})+k(E_{\mathrm{comp}}+E_{\mathrm{decom}})$

$=\frac{k}{\mathrm{\&tau;}}\&CenterDot;({2E}_{\mathrm{elec}}+{\mathrm{\&epsiv;}}_{\mathrm{fs}}{d}^2)+{2\mathrm{kE}}_{\mathrm{comp}}$

If data compression is effective, should there is E ₀> E ₁,

$k\&CenterDot;(2E_{\mathrm{elec}}+{\mathrm{\&epsiv;}}_{\mathrm{fs}}{d}^2)>\frac{k}{\mathrm{\&tau;}}\&CenterDot;({2E}_{\mathrm{elec}}+{\mathrm{\&epsiv;}}_{\mathrm{fs}}{d}^2)+2k\&CenterDot;E_{\mathrm{comp}}$

Abbreviation obtains $\mathrm{\&tau;}>\frac{{2E}_{\mathrm{elec}}+{\mathrm{\&epsiv;}}_{\mathrm{fs}}{d}^2}{{2E}_{\mathrm{elec}}+{\mathrm{\&epsiv;}}_{\mathrm{fs}}{d}^2-{2E}_{\mathrm{comp}}}$

While asking extreme value to obtain τ > 5, compression is effectively, wherein E _comp=40nJ/bit, E _elec=50nJ/bit, ε _fs=10pJ/bit/m ².

If node v belongs to above-mentioned Equations of The Second Kind node, suppose that node v need to jump data are sent to Sink node through n, when i jumps, the distance of point-to-point transmission is d _i.If its data are not compressed, be directly sent to Sink node, the network energy that consumed is:

wherein ${\stackrel{\&OverBar;}{d}}^2=\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i^2.$

After adopting EZC_SGW encryption algorithm to compress data, then be forwarded to the gross energy that Sink node consumes and be:

$E_1(v,\mathrm{Sink})=\frac{k}{\mathrm{\&tau;}}\&CenterDot;[(2n-1)E_{\mathrm{elec}}+{\mathrm{n\&epsiv;}}_{\mathrm{fs}}{\stackrel{\&OverBar;}{d}}^2]+k\&CenterDot;E_{\mathrm{comp}}$

If data compression is effective to saving network energy, should there is E ₀> E ₁,

$k\&CenterDot;[(2n-1)E_{\mathrm{elec}}+{\mathrm{n\&epsiv;}}_{\mathrm{fs}}{\stackrel{\&OverBar;}{d}}^2]>\frac{k}{\mathrm{\&tau;}}[(2n-1)E_{\mathrm{elec}}+{\mathrm{n\&epsiv;}}_{\mathrm{fs}}{\stackrel{\&OverBar;}{d}}^2]+k\&CenterDot;E_{\mathrm{comp}}$

The known τ of abbreviation should meet

$\mathrm{\&tau;}>\frac{(2n-1)E_{\mathrm{elec}}+{\mathrm{n\&epsiv;}}_{\mathrm{fs}}{\stackrel{\&OverBar;}{d}}^2}{(2n-1)E_{\mathrm{elec}}+{\mathrm{n\&epsiv;}}_{\mathrm{fs}}{\stackrel{\&OverBar;}{d}}^2-E_{\mathrm{comp}}}$

Ask extreme value to obtain: if τ > 5, compression effectively.

Because the compression ratio of EZC_SGW algorithm is at least 6, so, for arbitrary node data acquisition in network with after EZC-SGW compression algorithm again transmission necessarily can save network energy.

On Omnet++ platform, the least energy consumption adaptive convergence method for routing (CMEAAT) that the present invention is based on Second Generation Wavelets zerotree encoding algorithm is converged to routing algorithm with GIT, DAGP, AFST below and compare explanation.Suppose that 100 nodes are randomly dispersed in the square area of 100m × 100m, in network, only have a Sink node, the maximum communication radius R=30m of node, the data that each node collects in a monitoring periods are packets of a 2000bit.The present embodiment is by correlation radius r _sspan be chosen as [0.05m, 500m] so that degrees of fusion from 0 to 1 variation between data, thereby simulate the sensor network of different degrees of fusion.For the performance of test different types network, choose 3 kinds of different units and converge expense and test, be respectively and slightly converge expense q ₀=10nJ/bit; Moderate converges expense q ₀=60nJ/bit; Severe converges expense q ₀=120nJ/bit, carries out 20 emulation to every kind of experiment, gets the average of each the simulation experiment result as final result, respectively as shown in Fig. 3-1, Fig. 3-2 and Fig. 3-3.Result shows, slightly converging in the network of expense, the total energy consumption of the inventive method is compared with DAGP and reduced respectively approximately 27%, 25% and 24% with GIT, AFST; Converge in the network of expense in moderate, the total energy consumption of the inventive method is compared with DAGP with GIT, AFST, has reduced respectively approximately 47%, 42% and 39%; Converge in the network of expense in severe, the total energy consumption of the inventive method is compared with DAGP with GIT, AFST, has reduced respectively approximately 56%, 43% and 48%.In sum, the inventive method can be saved energy effectively, improves the life cycle of network, particularly converges in the network of expense for severes such as some monitoring sound, image or videos, and the effect of the inventive method is very obvious.

Claims (3)

1. the least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image, is characterized in that comprising the following steps:

(1) with network G=(V, E) limit e=(u in, v) network energy consumption when weights omega (e) is unit of transfer's data, wherein V represents all source nodes and Sink node, E represents the transmission link between institute's active node and Sink node;

(2) utilize respectively Prime algorithm and dijkstra's algorithm to calculate minimum spanning tree and the shortest path tree take Sink node as root node;

(3) α-balance spanning tree of performance of structure between described minimum spanning tree and shortest path tree is as the initial transmission path of network data; Structure α-balance spanning tree comprises the following steps:

(3-1) value of given α, and α >1, each the node v in minimum spanning tree distributes a variable d (v), for record v to the path weight value of root node with, when initial, to the each non-root node v in minimum spanning tree, make d (v)=∞; To root node r=Sink, make d (r)=0;

(3-2) traversal minimum spanning tree, to each node v, judge this node in minimum spanning tree to the path weight value of root node and whether exceed v in shortest path tree to the path weight value of root node and α doubly: if exceeded, v in shortest path tree is added in current network to the limit, path of root node, the value of adjusting d (v) be in shortest path tree v to the path weight value of root node with; Otherwise, keep the path of v constant;

(3-3) in follow-up traversal, if the path weight value of certain node v and value d (v) diminish, while traversing so the node u being directly connected with v, node u is done to following adjustment to the path weight value of Sink node and the value of d (u) and transmission path: if d (u) > ω is (u, v)+d (v), d (u) is adjusted into d (u)=ω (u, v)+d (v), node u is u → v → path (v to the route adjust of r, r), wherein path (v, r) be illustrated in current minimum spanning tree v to the path of root node r, otherwise node u remains unchanged to the path weight value of Sink node and d (u) and transmission path, after all nodes of minimum spanning tree have traveled through, the spanning tree structure after adjusting is α-balance spanning tree,

(4), in described α-balance spanning tree, if node v is short transmission path to the path of Sink node, make the judgement mark θ (v)=0 of this node v; Otherwise, make the judgement of this node v identify θ (v)=1;

(5) in the process of transfer of data, for the node of θ (v)=0, calculate it and converge benefit, and carry out adaptive convergence, then the data after utilizing zerotree encoding algorithm to adaptive convergence are compressed, data after compression are directly sent to Sink node along short transmission path, carry out decompress(ion) at Sink Nodes; The described benefit that converges draws by following steps:

(4-1), in the time that node v only has a child node u, the benefit through type (1) that converges on limit e=(u, v) calculates,

Wherein, Δ E _tbe illustrated in and on this limit, carry out the transmission energy consumption of saving after convergence, E _arepresent to converge energy consumption, m ₀and m (u) ₀(v) represent respectively to converge the data volume that front nodal point u and v self gather, T (v, S) representation unit data are from node v along the transmission energy consumption that converges tree path and be sent to Sink, q (e) representation unit data volume converge energy consumption,

for the data volume after convergence,

through type (2) calculates,

Wherein, ρ _uvrepresent the coefficient correlation between node u and v;

(4-2) when node v has the individual child node u of K (K>=2) ₁, u ₂..., u _ktime, adopt the method progressively converging to obtain the converge benefit of each child node at node v place, first obtain the data volume after convergence by formula (2)

obtain and converge benefit Δ (u according to formula (1) ₁, v); Then order obtain respectively child node u according to formula (2) and formula (1) ₂arrive after v the data volume after converging

with converge benefit Δ (u ₁, u ₂, v), by above-mentioned steps, obtain all child nodes and converge benefit Δ (u after v converge ₁, u ₂..., u _k, v);

(6) for the node of θ (v)=1, if data are benefited and are greater than 0 converging of v place, carry out convergence processing, utilize zerotree encoding algorithm to compress the data after converging, and press transmission path and send to Sink node; Otherwise, node v sends to next-hop node by after himself data compression by transmission path, and each child node of node v is obtained shortest path limit separately from shortest path tree, its data is utilized zerotree encoding algorithm compression by each child node, directly be sent to Sink node by shortest path limit separately, carry out decompress(ion) at Sink Nodes.

2. the least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image according to claim 1, is characterized in that: in described α-balance spanning tree, the value of α is

3. the least energy consumption adaptive convergence method for routing based on Second Generation Wavelets zerotree image according to claim 1, is characterized in that described adaptive convergence judgement is as follows:

(1) if converging to benefit is greater than 0, converge at node v place;

(2) if converging to benefit is not more than 0, directly data are sent to Sink node by transmission path.

Images