Background technology
Wireless sensor network be by be laid in monitored area have in a large number perception, communication, calculating and storage capacity node by the network of multi-hop wireless communication self-organizing.Due to its high flexibility, low cost, wireless sensor network is all widely used in fields such as environmental monitoring, national defence/safety, industrial automation, medical treatment & healths.In various application scenarios, wireless sensor network is faced with a common demand or task, and that collects the interested data of user exactly from physical environment.In wireless sensor network, the process that data flow converges to base station (or sink node) by a large amount of sensor node is called as Data Collection.Generally believe, Data Collection is the most important and the most basic a kind of business model of wireless sensor network.
In Data Collection, sensor node needs periodically the data of collection to be reported base station by multi-hop pass-through mode.Because number of sensors is comparatively large, and data flow all converges to direction, base station, thus there is a large amount of contiguous data transfer request in network, very easily causes the mutual interference between packet and conflict.In order to alleviate data packet collisions probability, improve network parallel transmittability, need reasonably to coordinate the access of each transmission node to wireless channel.At present, wireless channel cut-in method mainly comprises two large classes, namely based on the method for competition and the method based on scheduling.In the method based on competition, node is each other by the right to use of certain regular competitive channel.In the process of competitive channel, not only can produce a large amount of control overheads, and due to hidden station problem (Hidden Station Problem), the conflict controlling bag or packet still may occur in network.Therefore, although the method based on competition has its flexibility good and be easy to the advantages such as distributed implementation.But, as people such as Ganhdam at document [1.O.Durmaz Incel, A.Ghosh, B.Krishnamachari, and K.Chintalapudi, " Fast data collection in tree-based wireless sensor networks; " IEEE Trans.Mobi.Compu., vol.11, no.1, pp.86 – 99,2012] in pointed, method based on competition can cause extra energy ezpenditure and propagation delay time, and is not suitable for the larger data collection services of transmission request amount.Based on the method for scheduling, typical in TDMA (time division multiple access), by distributing different active slot to mutual afoul transmission node, effectively can evade the injury of interference to network performance, thus in wireless sensor network data is collected, obtaining investigation and application widely.
The dispatching method that existing data-oriented is collected concentrates on the link layer of network protocol stack.These dispatching methods depend on specifically linear or tree routing structure, there is following limitation.On the one hand, dispatching method needs a predetermined routing tree as input, and in the networking operational environment of reality, due to the factor such as barrier, node capacity, it is very difficult for many times constructing certain specific routing tree; On the other hand, from the angle of time slot allocation, predetermined transmission path may not be make to dispatch the less transmission path of total timeslot number, and total timeslot number, i.e. scheduling length, be the important indicator weighing scheduling performance.Owing to adopting fixing transmission path, existing dispatching method cannot make full use of the concurrent transmission ability between link, thus usually needs more timeslot number just can complete scheduler task, cannot meet the real-time of application and the throughput demand of network well.This dispatching method depending on specific routing tree is called as structuring dispatching method.In order to overcome the limitation of said structure dispatching method, the present invention provides a kind of semi-structured transmission dispatching method of Wireless Sensor Networks Data Collection targetedly.
Summary of the invention
The object of the present invention is to provide a kind of semi-structured transmission dispatching method of Wireless Sensor Networks Data Collection.Described method characteristic is the redundancy utilizing Data Collection path, constructs a lax semi-structured route topological figure.Then, based on cross layer design thought, propose the preferential dynamic link of minimal weight and select and Slot Allocation Algorithm.The traffic load of weight factor concentrated expression that the present invention defines link self and link mutually between interference relationships.The dispatching method that the present invention proposes avoids the dependence that conventional method is set particular data collect & route, and can significantly improve the utilance of time slot, thus reduces the time delay of Data Collection, improves network throughput.In addition, control and self-adapting data collection period modulation technology by introducing dynamic node sleep, the program also can meet the life cycle demand of application.
The technical scheme that technical solution problem of the present invention adopts comprises the following steps:
Step 1: according to geometric position and the transmission radius of node, set up the traffic diagram that wireless sensor network is corresponding.The sensor node be laid on two dimensional surface is become vertex set V with sink node modeling, and communication link all possible between node is modeled as limit set E, then wireless sensor network can describe with non-directed graph G (V, E) coming.Consider that node has the situation of isomorphism transmission radius r.Now, for arbitrary two node i and j, when the distance of and if only if i and j is less than or equal to r, just there is communication link between i and j, be expressed as (i, j) ∈ E.
Step 2: from sink node, network service figure performs breadth-first search, to obtain in network each sensor node to the jumping figure distance of sink node; According to the jumping figure distance of node, construct semi-structured route topological figure; Given traffic diagram G (V, E) take sink as starting point, performs a breadth-first search (Breadth-First Search, BFS).BFS is one of a kind of common graph search algorithm, why be referred to as breadth-first algorithm, be because algorithm from start to finish one straight through the border found and do not find between summit to external expansion, in other words, algorithm first search for sink distance be 1 all summits, and then removal search and S distance are other summits of 2, by that analogy, until all summits of search through.By having searched for all summits, the radius R of figure G can be obtained
g, and the jumping figure distance l of each summit i to sink in figure
i.L
ibe also called the number of plies of node i in figure G.Then, a semi-structured route topological figure T (V
t, E
t) can obtain by with under type: V
t=V, E
t=(i, j) | (i, j) ∈ E, l
i> l
j.T (V
t, E
t) feature be, in figure output (transmission) link of each summit i (except sink) be no longer specify with unique, a but alternative link set OL
i, be expressed as OL
i=(i, j) | (i, j) ∈ E
t, l
i> l
j.
Step 3: according to the dominant interference each other of the link in semi-structured route topological figure and time interference relationships, structure link interference figure; The link interference figure I that the route topological figure T constructed is corresponding; Link interference figure is used for describing the interference relationships in network between transmission link.Specifically, a summit in each link corresponding diagram I in T is schemed; Further, if two links in T conflict mutually, so scheme just there is a limit between corresponding two summits in I.At judgement link (i
1, j
1) and link (i
2, j
2) when whether conflicting, both needed to consider dominant interference, and needed again to consider secondary interference.In dominant interference, due to the restriction of Half-duplex transceiver, two links with coincidence node can not be dispatched simultaneously.In secondary interference, when the receiving node of a link is within the interference range of another link sending node, these two links can not be dispatched simultaneously.The same with other dispatching method in document, we also consider a kind of situation of simplification, namely disturb radius to equal to transmit the situation of radius.Now, link (i
1, j
1) and (i
2, j
2) sufficient and necessary condition that do not conflict mutually is i
1≠ i
2aMP.AMp.Amp & j
1≠ j
2aMP.AMp.Amp & | i
1j
2| > r & & | i
2j
1| > r.Wherein r is the transmission radius of node, is also the interference radius of node.
Step 4: perform based on the preferential link selection of minimal weight and Slot Allocation Algorithm.At the time slot scheduling t that each is concrete, to node waiting for transmission in network, the packet number sent in buffer memory according to it sorts from big to small, performs following steps:
Step 4.1: transmission node sorts.(q is expressed as according to bag quantity to be sent in the initial time node i buffer queue of time slot t
i(t)), to all q in network
it the node of () > 0 carries out descending sort.
Step 4.2: transmission node interference checks.To node sorted in step 4.1, carry out interference inspection successively, to judge whether this node can be dispatched in current time slots.Specifically, for node i to be sent, if its candidate sends link set OL
iin a certain bar link all dispatched with current time slots of each link conflict, then i can not transmit in time slot t.On the other hand, if OL
imiddle existence at least one link all links scheduled with current time slots do not conflict, then forward step 4.3 to.
Step 4.3: link selection and scheduling.For node i, use OL
i' represent OL
iin the link subset of not conflicting with all scheduled links.We define the weight interference factor w of certain link e to certain transmission node S set
s(e) be
wherein S
ebe a subset of S, comprise all transmission node set simultaneously can not dispatched with e because link e disturbs in S.Then, we are from OL
i' in choose institute's available for transmission and do not experience interference and check that the weight of node set (being expressed as D) of (step 4.2) disturbs minimum link as the transmission path of node i, and to dispatch in current time slots.Specifically, for node i selects link
And dispatch in current time slots.
Step 5: at current scheduling time slot, checks sorted node to be transmitted successively; According to link interference figure, whether every bar candidate transport link of decision node can be dispatched in current time slots; All alternative link all cannot must wait for next time slot scheduling at the node of current time slots scheduling.
Step 6: definition link weight interference factor.In each time slot scheduling, at the link of current time slots scheduling, its weight being calculated; For each node to be transmitted, the link selecting weight minimum from its alternative link as the transmission path of reality, and is dispatched in current time slots.
Step 7: repeat step 4-6, until all nodes in network complete transformation task all; Then, at each concrete transmission time slot, scheduling does not have the node of data transmit-receive task to enter resting state; Meanwhile, according to the life cycle demand of application, the cycle that setting data is collected.
As can be seen here, the invention is characterized in:
1. by sink node, perform breadth-first search, to obtain in network each sensor node to the jumping figure distance of sink, i.e. the number of plies of this node in traffic diagram.Then, from traffic diagram, delete the link that all end points have the identical number of plies, the node in remaining link and network together constitutes a semi-structured route topological figure.
2., in described semi-structured route topological figure, between each sensor node and sink node, there is the transmission path of one or more candidate; Further, each sensor node has the relay forwarding node of one or more candidate.
3. each link in described half structure route topological figure correspond to each summit in link interference figure.Meanwhile, if there is interference (dominant interference or secondary interference) between two links in semi-structured route topological figure, then there is a limit between two summits of the correspondence in link interference figure.
4. in each time slot scheduling, travel through all nodes waiting for transmission, produce with any scheduled link the node disturbed for inevitable, all do not allow it to dispatch in current time slots.Before time slot allocation, perform the correctness that such process can be guaranteed to dispatch.
5. the weight factor reflection link defined is to other node annoyance levels to be transmitted, described weight factor reflect the traffic load of link self and link mutually between interference relationships, consider the packet number in nodal cache queue, i.e. the traffic load of node.
6., in the alternative link set of each transmission node, the minimum link of prioritizing selection weight as the transmission path of reality, and is dispatched in current time slots.In each time slot scheduling, due to the impact of air interference, only have the link do not conflicted mutually to dispatch simultaneously.By the transmission path that prioritizing selection is less to other node annoyance levels, the number of links of scheduling in single time slot can be increased, improve the concurrent capability of network.Because total transmission data request amount is determined, therefore raising single time slot schedule link number is conducive to total timeslot number that minimizing dispatching office needs, thus improves the time delay of Data Collection, meets the real-time demand of application better.
7. at each time slot, allow the complete communication close equipment of node neither needing to send and do not need to receive packet, enter resting state, thus significantly reduce the energy ezpenditure of node.Meanwhile, collect the cycle by setting data, make in each data collection periods, after completing all transformation tasks, allow whole node enter dormancy, lower duty ratio can be realized, thus extend Network morals.
8. sleep scheduling and data collection periods are arranged.At each time slot, the possible state of node has transmission (TX), receives (RX), and dormancy (SLEEP).Current time slots is neither needed to send data, does not need again to receive the node of data, allow its communication close equipment, enter dormancy, thus save energy.Meanwhile, it is C that setting data collects the cycle, and each data collection periods is divided into two stages.First stage completes Data Collection scheduler task, and second stage the whole network node enters dormancy, thus realizes the mode of operation of duty ratio formula (duty cycle).Under the prerequisite meeting real-time demand, by increasing the value of data collection periods C, the duty ratio of network can be reduced, thus extend Network morals.
According to step described above, the advantage of semi-structured dispatching method proposed by the invention is:
(1) from step 2, the routing infrastructure that the method adopts is lax semi-structured route topological figure.In the figure, there is the transmission path of many candidates between each node to sink, meanwhile, the data of each node can be forwarded by the via node of multiple candidate.It is advantageous that, on the one hand, avoid the dependence of structuring dispatching method to particular path.When some path in network is unreachable, the performance of structuring dispatching method can be affected, and the scheduling scheme that the present invention proposes has certain fault-tolerant ability.On the other hand, owing to providing many optional transmission paths to dispatching method, when actual scheduling, can transmit in the prioritizing selection path less to other node annoyance levels to be sent.
(2) from step 3, the method both considered dominant interference, considered again time interference.In systems in practice, air interference is very complicated and is difficult to modeling.The dispatching method that the present invention proposes, by considering dominant interference and time interference, is more accurately described the impact of air interference, thus the practicality of strengthening dispatching method.
(3) from step 4.2, at each time slot, all first the method rejects the transmission node scheduled link being produced to interference.By filtering out potential conflicting nodes, ensure that the correctness of dispatching method.Specifically, ensure that in any one time slot scheduling, all concurrent links do not conflict between any two all mutually.
(4) from step 4.3, the method define reflection link to the weight factor of other node annoyance levels to be sent, and in the alternative link set of each transmission node, the minimum link of prioritizing selection weight is dispatched in current time slots as the transmission link of reality.Meanwhile, this weight factor also considers the packet number in nodal cache queue.By the transmission path that prioritizing selection is less to other node annoyance levels, the number of links of scheduling in single time slot can be increased, improve the concurrent capability of network.Because total transmission data request amount is determined, therefore raising single time slot schedule link number is conducive to the total time that minimizing dispatching office needs.
(5) from step 5, the method, while scheduling node transmission time slot, also scheduled the sleep state of node.At each time slot, allowing neither needs to carry out data and sends the node communication close equipment not needing again to carry out data receiver, enters resting state, significantly can reduce the energy ezpenditure of node.Meanwhile, this dispatching method dynamically can collect the cycle by setting data.In single dispatching cycle, after all communication tasks complete, allow whole nodes enter resting state, realize lower duty ratio, thus extend Network morals.
In a word, the invention discloses a kind of semi-structured transmission dispatching method of Wireless Sensor Networks Data Collection.It is characterized in that described method comprises: 1) by performing breadth-first search to network service figure, constructing a lax semi-structured data collect & route topological diagram; 2) according to the dominant interference in network between transmission link and time interference relationships, structure link interference figure; 3) link weight interference factor is defined, the interference strength other links of network caused with the traffic load of choosing comprehensively sending node and transmission link; 4) select and timeslot allocation scheme based on the dynamic link that minimal weight is preferential; 5) the node dormancy state of explicit control and the data collection periods of self-adaptative adjustment.The present invention is based on cross layer design thought and co-design is carried out to route and scheduling, strengthened the slot efficiency of scheduling by the redundancy making full use of path, thus reduce the time delay of Data Collection, improve the throughput of network.The present invention is applicable to the wireless sensor network application having requirement higher to real-time.
Embodiment
In order to methods and applications of the present invention are described better, below in conjunction with example and accompanying drawing, provide the specific embodiment of the present invention, to illustrate substantive distinguishing features of the present invention and significant progress further, but the present invention is only confined to embodiment by no means.
Fig. 1 is the traffic diagram that the wireless sensor network derived from certain actual environment monitoring system is corresponding.Fig. 2 is the flow chart that dispatching method of the present invention realizes.As shown in Figure 3, the traffic diagram of dispatching method structure wireless sensor network, and construct semi-structured route topological figure and link interference figure, then carry out the distribution of arrangement and the time slot dispatched, the resting state of node is finally set.In order to more intuitively, more generically set forth the implementation detail of the flow chart shown in Fig. 2, now for the data gathering problem on the wireless sensor network shown in Fig. 1, illustrate in greater detail.Specifically, the process of scheduling is as follows:
1, tectonic network traffic diagram and semi-structured route topological figure.Network service figure is determined by internodal geometric position and communication radius, when the distance between two nodes is less than or equal to communication radius, means to there is a communication link between these two nodes.Fig. 1 has provided the traffic diagram of this wireless sensor network, is input now, constructs semi-structured route topological figure with Fig. 1.Specifically, from sink node s, the upper execution breadth-first search (BFS) of the network service figure (representing with symbol G) shown in Fig. 1, obtains the network hops radius R centered by s
g=3, and each node is to the jumping figure distance of s, the i.e. number of plies of node.Such as, the number of plies l of No. 1 node can be obtained
1the number of plies l of=3, No. 8 nodes
8=1.Then, delete the link between the number of plies is identical in G node, and to set each node to the transmission path of s be shortest path (shortest path), namely obtain semi-structured route topological figure as shown in Figure 3.Can observe, in figure 3, each node may have the via node of multiple candidate.Such as, No. 1 node has No. 2, No. 3 these two possible down hop via nodes, and No. 7 nodes have No. 4, No. 6 these two possible down hop forward node.
2, the link interference figure of structural map 3 correspondence.In order to carry out modeling to the air interference in actual environment better, the present invention both considered dominant interference, considered again time interference.In dominant interference, due to the restriction of Half-duplex transceiver, two links with coincidence node can not be dispatched simultaneously.Such as, link (1,2) and (2,5) can not be dispatched simultaneously.In secondary interference, when the receiving node of a link is within the interference range of the sending node of another link, these two links can not be dispatched simultaneously.Such as, link (1,2) and (3,5) can not be dispatched simultaneously.In the present embodiment, the situation when disturbing radius to equal to transmit radius is considered.Now, the link interference figure that Fig. 3 is corresponding as shown in Figure 4.Every bar link in Fig. 3 correspond to each summit in Fig. 4; If there is limit between two summits in Fig. 4, then two corresponding in representative graph 3 links conflict mutually, can not at same timeslot scheduling.Specifically, in Fig. 4, between summit (1,2) and summit (2,5), there is limit, mean that the link between node 1 and node 2 conflicts mutually with the link between node 2 and node 5.And the reason of conflict is, node 2 can not communicate with node 5 with node 1 simultaneously, otherwise corresponding transmission can produce collision.Therefore, these two links must be dispatched at different time-gap.
3, based on the preferential link selection of minimal weight and Slot Allocation Algorithm.At initial time, i.e. first time slot scheduling (t=1), all 1-8 nodes all have 1 packets need to send, and scheduled link set Sche (t) is empty.Suppose that from No. 1 node, carry out node interference checks.Because Sche (t) is empty, therefore No. 1 node must be dispatched in current time slots.There are two possible transmission paths in No. 1 node, i.e. link (1,2) and link (1,3).According to the definition of given link weight interference factor, the weight factor that can calculate link (1,2) and link (1,3) is respectively 4 and 5.Obviously, comparatively link (1,3) is little for the weight of link (1,2), and therefore we dispatch by prioritizing selection link (1,2), and are added current time slots.Next, travel through all the other nodes successively, and judge whether it can dispatch in current time slots.Such as, because link (1,2) is in current time slots scheduling, and the transmission path of No. 2 nodes (2,5) and link (1,2) conflict, and therefore No. 2 nodes can not be dispatched in current time slots.If node have passed interference inspection, then according to the method adopted during above-mentioned scheduling node 1, link selection and time slot allocation are carried out to it.After having traveled through all nodes, time slot is added 1, then repeated said process, until all nodes have completed transmission task all, now dispatched.Fig. 5 gives the detailed dispatching method of the wireless sensor network example shown in Fig. 3.
4, node sleep scheduling and data collection periods adjustment.After completing link transmission scheduling, in order to save energy, need the sleep state of further scheduling node.The rule of node sleep scheduling is as follows: at each time slot, neither needs to send packet, does not need again the node communication close equipment accepting packet, enter resting state.Such as at the 1st time slot, No. 3, No. 5 and No. 6 nodes enter resting state (as shown in Figure 5).Data collection periods C is divided into two benches, i.e. transmitting and scheduling stage and net dormancy stage.In the present embodiment, transmitting and scheduling stage length is 9 time slots, and thus network duty ratio is
obviously, by increasing data collection periods C, the duty ratio of network can be reduced, thus extend Network morals.
In order to further show the advantage of the scheduling scheme that the present invention proposes, the present inventor has also carried out a large amount of MATLAB emulation experiments.In simulated experiment, be set in certain regional extent and lay sensor node at random, and change some network parameter, observe the performance comparison of scheduling scheme and other the known typical methods proposed.In MATLAB experimental situation, fixed network area size is 100 meters × 100 meters.In this region, lay the sensor node of some at random equably, and the heart fixedly puts a sink node in the zone.The dispatching method carrying out performance comparison with the present invention has: LTSA method (document [2.H.Choi, J.Wang, and E.A.Hughes, " Scheduling for information gathering on sensor network, " Springer Wire.Netw., vol.15, no.1, pp.127 – 140, 2009.]), GANHDAM method (document [3.S.Gandham, Y.Zhang, and Q.Huang, " Distributed time-optimal scheduling for convergecast in wireless sensor networks, " Elsevier Comp.Netw., vol.52, no.3, pp.610 – 629, 2008]) and LPIPE method (document [4.W.Wang, Y.Wang, X.-Y.Li, W.-Z.Song, and O.Frieder, " Efficient interference-aware tdma link scheduling for static wireless networks, " in Proc.2006 ACM Mobicom, pp.262 – 273.]).Three kinds of methods described in document (2-4) are all representational Data Collection scheduling schemes in the middle of document.First, stationary nodes transmission radius is 30 meters, and progressively increases the sensor node quantity in network, the mean value of the time slot scheduling quantity required for more various dispatching algorithm and variance.Fig. 6 gives experimental result.When number of nodes increases, the equal monotone increasing of total number of timeslots required for various dispatching method.But can find, comparing other three kinds of methods the total time wanted required for the present invention has obvious minimizing.Then, if number of nodes is fixed as 300, and progressively increase node-node transmission radius, Fig. 7 gives experimental result now.Can observe, performance of the present invention is still better than other three kinds of methods.Simulation result shown in Fig. 6 and Fig. 7 demonstrates the superiority of scheduling scheme proposed by the invention further.