CN101325556B - Multi-path routing method for wireless network based on network encoding - Google Patents

Abstract

A multi-route routing method based on network coding in wireless network includes the following steps: 1) routing discovering from originating nodes to object nodes; 2) after the routing discovering stage is finished, comparing route conversion incomes of each route, and originating nodes selecting the optimum W routes therein for data transmission. Compared with the prior technology, the invention proposes a multi-route routing method based on network coding in wireless network, dynamically transmits data packets on multiple routes according to reliability and coding opportunity of routes, dynamically creates rather than only waiting for coding opportunity through conversion transmission path, and multiple routes share network flow quantity loads, and maximize the route conversion, accordingly improving network output.

Description

The multi-path routing method of coding Network Based in a kind of wireless network

Technical field

The invention belongs to the radio network technique field, be specially the multi-path routing method of coding Network Based in a kind of wireless network, mainly be applicable to the discovery of multipath route in the multi-hop wireless network and the selection of data transfer path, also utilized network coding technique on the other hand to improve systematic function.

Background technology

Wireless network has notable features such as insecure transmission link, dynamic topological structure, limited channel width and node electric weight.In recent years, the multipath route has obtained using widely in wireless network.The multipath route can improve reliability of data transmission effectively, the frequent topologies change of balancing network flow load and electric quantity consumption, reduction end-to-end time delay and route discovery frequency, reply and the insecure communication service that brings thereof between node, and improves fail safe of network or the like.The network congestion problem that but the multipath route can cause sending packet simultaneously on the mulitpath again to be brought, during the transmits data packets of the particularly elected disjoint path of majority bar, congested meeting becomes more serious.

Existing multi-path routing mechanism has all pre-determined transmission path in the route discovery stage, does not consider the network coding opportunity that exists in the transmission of data packets.These multi-path routing mechanisms do not utilize existing code machine meeting well, so the throughput of network is not brought into play fully.

Summary of the invention

The problem to be solved in the present invention is: there is the network congestion problem in existing multi-path routing mechanism, does not utilize existing code machine meeting well, and the throughput of network is not brought into play fully.

Technical scheme of the present invention is: the multi-path routing method of coding Network Based in a kind of wireless network may further comprise the steps:

(1) to the relevant information of wherein adding link, finds route and write down expected transmission times ETX (the Expected Transmission Count) information of relevant link before source node is relayed away route request packet to the route discovery of destination node: extended dynamic source route algorithm EDSR (Extended DynamicSource Routing);

(2) after the route discovery stage finished, source node had obtained the ETX information of all links on many paths of leading to destination node and the every paths; By routing algorithm, relatively income is changed in the path of each paths again, and source node is chosen wherein optimum W bar and is used for transfer of data;

Routing algorithm with each bar path candidate with compare except the conversion income in disconnected path and all paths itself, select then and have maximum path and change the path candidate of income and come transfer data packets; If exchange income g _{I, j, k}For between node i and node k, successfully carry out once the number of transmissions that coded data packet switch Network Based is reduced through intermediate node j; By exchange income g _{I, j, k}Can obtain path code income g _{I, j, k} ^P, path code income g _{I, j, k} ^PBe by from path P _iBe transformed into path P _jSo that utilize path P _kOn the code machine meeting and the number of transmissions that reduces; Then income G is changed in the path _{I, j, k}For: when the path from P _iBe transformed into P _jSo that with path P _kDuring exchange data packets, path conversion income G _{I, j, k}Value be exactly path code income g _{I, j, k} ^PWith path ETX loss (ETX _j ^P-ETX _j ^P) between poor.

EDSR algorithm of the present invention can be improved to optimizes extended dynamic source routing discovery algorithm EDSR-OPT (ExtendedDynamic Transmission Count-Optimization), if each node is all understood own and source node and the residing geographical position of destination node, because the node in the wireless mesh network remains static substantially, each node can utilize GPS equipment on other nodes to carry out the trilateration operation at initial phase and obtain the positional information of self, utilizes geography information EDSR-OPT algorithm that the inundation scope of route requests is limited in the diamond-shaped area between source node and the destination node.

After the route discovery stage finished, by routing algorithm, relatively income was changed in the path of each paths, and source node is chosen wherein optimum W bar and is used for transfer of data, and W is specified by the user, generally is a smaller numerical value, as 3.According to path conversion income, on the intermediate node in path, choose a path candidate so that create the code machine meeting and maximization path conversion income for each paths.

The present invention compared with prior art, the multi-path routing method of coding Network Based in a kind of wireless network has been proposed, according to the reliability in path and code machine can be dynamically in the transmission of the enterprising line data bag of mulitpath, dynamically create but not only wait for the code machine meeting by the conversion transmission path, allow mulitpath share network traffic load, thus and the throughput of maximization path conversion income improvement network.

Description of drawings

Fig. 1 is the schematic diagram of algorithm EDSR.

Fig. 2 is the schematic diagram of algorithm EDSR-OPT.

Fig. 3 (a) is an exchange income exemplary plot.

Fig. 3 (b) is a path conversion income exemplary plot.

Embodiment

The present invention at first finds source node all routes to destination node, selects by the contrast of path conversion income again, is mainly realized by route discovery algorithm and routing algorithm.

(1) route discovery algorithm

Routing discovering method---extended dynamic source routing EDSR (Extended Dynamic Source Routing), develop by DSR DSR (Dynamic Source Routing), as Fig. 1, what it was different with DSR is to the relevant information of wherein adding link before route request packet is relayed away.After neighbor node received route requests, this node added expected transmission times ETX (the Expected Transmission Count) information of a last link in the route request packet to, again with its forwarding.If node is received the route requests of repetition, will abandon simply.

When destination node is received route requests, send a route replies that comprises every link ETX information on the route found and the route to source node.Route requests can be sent to destination node through mulitpath usually, so source node can obtain many routes of leading to destination node when route finding process finishes.

But destination node needn't be replied each route requests of receiving.We wish only to incite somebody to action preferably by destination node, and route returns to source node.Therefore destination node need filter out the route requests of having found relatively poor route.Generally, late route requests means that they have passed through the very low link of some transmission success rates, and it is defeated perhaps to have passed through too many jump set.Therefore, we set has only when route requests arrives in time restriction τ, and destination node just can provide and reply.So source node has added a timer τ in route request packet.When τ was expired, this route requests can not transmitted or reply to any node.

Similar with DSR, each node is noted the newfound route and the ETX of every link wherein among the EDSR in buffer memory.Caching mechanism not only can be accelerated route discovery processes and reduce the transmission number of times of route requests, and also the path conversion for the packet transfer phase provides the candidate route.

In order to find out the route of leading to destination node, EDSR carries out inundation and searches in network.But the inundation mode can increase the delay in route discovery stage and the load of emphasis network, especially under the bigger situation of network size.Though DSR can limit the number of transmissions by the life span TTL (Time To Live) that increases route requests gradually, but still can't avoid the increase of route discovery delay.

The present invention improves the EDSR algorithm, promptly optimize extended dynamic source routing EDSR-OPT (ExtendedDynamic Source Routing-Optimization), different with EDSR is, EDSR-OPT supposes that each node all understands own and source node and the residing geographical position of destination node.Because the node in the wireless mesh network remains static substantially, each node can utilize GPS equipment on other nodes to carry out the trilateration operation at initial phase and obtain the positional information of self.Utilize geography information EDSR-OPT to be limited in the inundation scope of route requests in the diamond-shaped area between source node and the destination node.

Fig. 2 has provided an EDSR-OPT examples of applications.According to the parameter θ that source node is set, we have defined the spread scope that a diamond-shaped area R limits route requests between source node and destination node.When a node was received route requests, it at first can check oneself whether in region R, and then whether decision relays this route requests to neighbor node.For example, in Fig. 2, Node B and node E abandon the route requests that transmits from node A and node D respectively, and node C and node F then are considered as suitable relay person to oneself.By introducing confined area R, EDSR-OPT does not just reduce the path at destination node, can carry out same operation in network yet.More important is that because the size of confined area and network size are irrelevant, the extensibility of route finding process will be significantly improved.

The physical coordinates of supposing present node, source node and destination node is respectively (x; Y), (x _sy _s) and (x _dy _d), the condition that present node is among the confined area R that parameter is θ is as follows:

$\left\{\begin{array}{c}x>(y-y_s)\&times;\tan (\arctan \left(\frac{y_d-y_s}{x_d-x_s}\right)+\mathrm{\&theta;})+x_s\\ x<(y-y_d)\&times;\tan (\arctan \left(\frac{y_d-y_s}{x_d-x_s}\right)+\mathrm{\&theta;})+x_d\\ y>(x-x_s)\&times;\tan (\arctan \left(\frac{y_d-y_s}{x_d-x_s}\right)-\mathrm{\&theta;})+y_s\\ y<(x-x_d)\&times;\tan (\arctan \left(\frac{y_d-y_s}{x_d-x_s}\right)-\mathrm{\&theta;})+y_d\end{array}\right.---\left(1\right)$

(2) routing algorithm

After the route discovery stage finished, source node had obtained the ETX information of all links on many paths of leading to destination node and the every paths.Source node compares and chooses wherein optimum W bar to the ETX of every paths.Here, W is one, and it has defined the granularity that flow is cut apart by user-specified parameters, and generally is a smaller numerical value, for example: 3.Suppose e _iBe the ETX value of path P uplink i, the ETX value of path P so, ETX ^PCan be calculated as follows:

${\mathrm{ETX}}^P=\mathrm{\&Sigma;}{e}_i(i\&Element;P)---\left(2\right)$

Suppose source node selection path collection { P _i} _I=1 ^KTransmit data flow F, and { P _i} _I=1 ^KPath ETX be { ETX _i ^P} _I=1 ^KBe assigned to path P among the data flow F so _iPart be:

$F_i=\frac{{\mathrm{ETX}}_i^P}{\underset{j=1}{\overset{K}{\mathrm{\&Sigma;}}}{\mathrm{ETX}}_j^P}\&times;F---\left(3\right)$

The path conversion of coding Network Based is not a shirtsleeve operation, must be transformed in the coding income that is obtained with from optimal path between the loss that brings in time good path and make a balance, have only when coding income during greater than the loss of path ETX value, the path conversion operations just can be performed.Otherwise, will still use the path of acquiescence.

Below by an example exchange income is provided the theoretical expectation value that defines and derive it, then based on exchange income definition path conversion income.By it is carried out quantitative analysis, whether determine the execution route conversion operations.

Exchange income g _{I, j, k}: exchange income g _{I, j, k}Be between node i and node k, successfully carry out once the number of transmissions that coded data packet switch Network Based is reduced through intermediate node j.

At first the simple application scene among Fig. 3 (a) is analyzed.Suppose that the unidirectional delivery success rate from node i to node j is r _{I → j}, use r _{I, j}Represent once to comprise the probability of success of the bi-directional of confirming operation, for example, r _{I, j}=r _{I → j}r _{J → i}According to the ETX standard, under the situation of not considering network coding opportunity, the expected transmission times of once successful swap operation is

Even considered network coding opportunity,, still need owing to have only node 1 to carry out encoding operation

Inferior transmission could be finished the data packet delivery of node 2 and node 3 to node 1 respectively.In order to finish whole exchange process, node 1 need successfully pass to node 2 and node 3. with the packet behind the coding

Suppose, before completing successfully data passes, need retransmission operation h time.Use p _1,2 ^hRepresent that from node 1 to node 2 data packet transmission task is only completing successfully for the h time and another is from node 1 to node that 3 data packet transmission task (comprised the h time) probability that just completes successfully before the h time.Therefore,

$p_{\mathrm{1,2}}^h=s_{\mathrm{1,2}}^{h-1}{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}\underset{i=1}{\overset{h}{\mathrm{\&Sigma;}}}{s}_{\mathrm{1,3}}^{i-1}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}$

3 data packet transmission task is only completing successfully for the h time and another is from node 1 to node that 2 data packet transmission task (comprised the h time) before the h time and is labeled as p with regard to the probability that completes successfully from node 1 to node _1,3 ^hTherefore,

$p_{\mathrm{1,3}}^h=s_{\mathrm{1,3}}^{h-1}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}\underset{i=1}{\overset{h}{\mathrm{\&Sigma;}}}{s}_{\mathrm{1,2}}^{i-1}{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}$

From node 1 to node 2 data packet transmission task and from node 1 to node 3 data packet transmission task all only is labeled as p at the probability that completes successfully for the h time _{1,2; 1,3} ^hTherefore,

${\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">p</figure-callout>}}_{\mathrm{1,2};\mathrm{1,3}}^h=s_{\mathrm{1,2}}^{h-1}{s}_{\mathrm{1,3}}^{h-1}{r}_{\mathrm{1,2}}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}$

So having under the situation of network code, the desired value of the number of transmissions is:

$T=\underset{h=1}{\overset{\&infin;}{\mathrm{\&Sigma;}}}(p_{\mathrm{1,2}}^h+p_{\mathrm{1,3}}^h-{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">p</figure-callout>}}_{\mathrm{1,2};\mathrm{1,3}}^h)h$

H is a number of retransmissions performed when completing successfully transmission.Here, s _{I, j}=1-r _{I, j}, can reduce $\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}}+\frac{1}{r_{\mathrm{1,3}}}-\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}+r_{\mathrm{1,3}}-r_{\mathrm{1,2}}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}}.$ Therefore, once the desired value of the number of transmissions of the exchange process of the coding Network Based of success needs is $\frac{2}{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}}+\frac{2}{r_{\mathrm{1,3}}}-\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}+r_{\mathrm{1,3}}-r_{\mathrm{1,2}}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}}.$ Simultaneously, the exchange income g on node 1 _1,2,3Value be $\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,2}}+r_{\mathrm{1,3}}-r_{\mathrm{1,2}}{\mathrm{<figure-callout\; id="3"\; label="r"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,3}}}.$ Suppose node i and node k successfully by node j exchange data packets, the exchange income at node j can be calculated as follows so:

$g_{i,j,k}=\frac{1}{r_{i,j}+r_{i,k}-r_{i,j}{r}_{j,k}}---\left(4\right)$

Based on the definition of exchange income, can provide the definition of path code income and path conversion income, and derive the computational methods of their theoretical value.

Path code income g _{I, j, k} ^P: path code income g _{I, j, k} ^PBe by from path P _iBe transformed into path P _jSo that utilize path P _kOn the code machine meeting and the number of transmissions that reduces.

Path conversion income G _{I, j, k}: when us the path from P _iBe transformed into P _jSo that with path P _kDuring exchange data packets, path conversion income G _{I, j, k}Value be exactly path code income g _{I, j, k} ^PWith path ETX loss (ETX _j ^P-ETX _i ^P) between poor.

Among Fig. 3 (b), node 1 is attempted from path P ₁Be transformed into P ₂Like this, it just can with path P ₃Carry out packet exchange and make full use of the code machine meeting.(G when but, having only the path ETX value of being brought greater than the path conversion when the path code income of node 5 and node 6 to increase _1,2,3＞0), the path conversion operations just can be performed.

${\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">g</figure-callout>}}_{\mathrm{1,2,3}}^P={\mathrm{<figure-callout\; id="5"\; label="g"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">g</figure-callout>}}_{\mathrm{1,5,6}}+{\mathrm{<figure-callout\; id="5"\; label="g"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">g</figure-callout>}}_{\mathrm{5,6,7}}---\left(5\right)$

${\mathrm{<figure-callout\; id="2"\; label="G"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">\; <figure-callout\; id="3"\; label="G"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">G</figure-callout>\; </figure-callout>}}_{\mathrm{1,2,3}}={\mathrm{ETX}}_1^P-{\mathrm{ETX}}_2^P+{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="S200810124277XE00013.png"\; state="\{\{state\}\}">g</figure-callout>}}_{\mathrm{1,2,3}}^P---\left(6\right)$

Suppose path P _iTo be switched to path P _jThereby and path P _kCarry out packet exchange.

Delegated path P _jAnd P _kTotal node set.So, path code income g _{I, j, k} ^PWith path conversion income G _{I, j, k}Can be calculated as follows:

$g_{i,j,k}^P=\underset{s=1}{\overset{m_{j,k}}{\mathrm{\&Sigma;}}}{g}_{{\mathrm{pre}}_j^s,s,{\mathrm{succ}}_j^s}---\left(7\right)$

$G_{i,j,k}={\mathrm{ETX}}_i^P-{\mathrm{ETX}}_j^P+g_{i,j,k}^P---\left(8\right)$

Here, pre ^JsAnd succ _j ^sIt is respectively path P _jLast node N _sForerunner's node and descendant node.Only work as G _{I, j, k}＞0 o'clock, path P _iAnd P _jBetween just can change.

Based on path conversion income calculation method, we are each paths P on intermediate node _kChoose a path candidate so that create the code machine meeting and maximization path conversion income.Routing algorithm with each bar path candidate with compare except the conversion income in disconnected path and all paths itself.Then, it selects the path candidate that has maximum return and comes transfer data packets.Suppose current m bar default path and the n bar path candidate of existing.The complexity of routing algorithm be O (m * n), routing algorithm and path conversion income algorithm details step are as follows:

Algorithm 1: routing algorithm

Input: set of paths $P={\left\{P_i\right\}}_{i=1}^m(m\&GreaterEqual;1),$ Default path P _kGather with path candidate

Output: the label l of the path candidate of maximization path conversion income.

1:{2-3 is capable: initialization }

2：G _MAX←0

3：l←0

4:{5-19 is capable: calculate maximum return }

5:for path candidate j=1to n do

6：G←0

7:for path i=1to n do

8:G _Old← G{ backs up currency }

9：if?i≠k?and?

then

10:{ calculates income }

11：G←CalcGain(P _k，?

P _i)

12：if?G≤G _old?then

13：G←G _old

14：else

15：continue

16：if?G _MAX＜G?then

17:G _MAX← G{ upgrades maximum return }

18：l←j

19:return l{ returns the label in selected path }

Algorithm 2:CalcGain (i, j, k) path conversion income algorithm

Input: path P _i, P _iWith

Output: G _{I, j, k}

1:{2 is capable: calculate and be transformed into the loss that bring in longer path }

2： $G_{i,j,k}\&LeftArrow;{\mathrm{ETX}}^{P_i}-{\mathrm{ETX}}^{P_j}$

3:{4-7 is capable: calculating path conversion income }

4:forall node v ∈ P _iDo

5:{ calculates the summation of exchange income on each node }

6： $G_{i,j,k}\&LeftArrow;G_{i,j,k}+g_{{\mathrm{pre}}_i^v,v,{\mathrm{succ}}_i^v}$

7：return?G _i，j，k。

Claims (3)

1. the multi-path routing method of coding Network Based in the wireless network is characterized in that may further comprise the steps:

(1) source node is to the route discovery of destination node: to the relevant information of wherein adding link, find route and write down the expected transmission times ETX information of relevant link before extended dynamic source route algorithm EDSR relays away route request packet;

(2) after the route discovery stage finished, source node had obtained the ETX information of all links on many paths of leading to destination node and the every paths; By routing algorithm, relatively income is changed in the path of each paths again, and source node is chosen wherein optimum W bar and is used for transfer of data, and W has defined the granularity that flow is cut apart, and is specified by the user;

Routing algorithm with each bar path candidate with compare except the conversion income in disconnected path and all paths itself, select then and have maximum path and change the path candidate of income and come transfer data packets; If exchange income g _{I, j, k}For between node i and node k, successfully carry out once the number of transmissions that coded data packet switch Network Based is reduced through intermediate node j; By exchange income g _{I, j, k}Can obtain the path code income

The path code income

Be by from path P _iBe transformed into path P _JSo that utilize path P _kOn the code machine meeting and the number of transmissions that reduces; Then income G is changed in the path _{I, j, k}For: when the path from P _iBe transformed into P _jSo that with path P _kDuring exchange data packets, path conversion income G _{I, j, k}Value be exactly the path code income

Lose with path ETX

Between poor.

2. the multi-path routing method of coding Network Based in a kind of wireless network according to claim 1, it is characterized in that finding that with optimizing the extended dynamic source routing algorithm EDSR-OPT substitutes the EDSR algorithm, EDSR-OPT establishes each node and understands own and source node and the residing geographical position of destination node, each node utilizes GPS equipment on other nodes to carry out the trilateration operation at initial phase and obtains the positional information of self, utilizes geography information EDSR-OPT algorithm that the inundation scope of route requests is limited in the diamond-shaped area between source node and the destination node.

3. the multi-path routing method of coding Network Based in a kind of wireless network according to claim 1 and 2, it is characterized in that:, on the intermediate node in path, choose a path candidate to create code machine meeting and maximization path conversion income for each paths according to path conversion income.

Images