Background technology
Along with the develop rapidly of sensor technology, wireless communication technology, embedded computing technique, distributed information processing and increasingly mature, microsensor and network with perception, computing capability and communication capacity begin to occur on a large scale.Wireless sensor network can detect collaboratively in real time, tracking and collection network are laid various environment in the zone or the information of monitoring target, and to these information work in coordination with, fusion treatment, obtain conclusion comparatively accurately, be sent to the user then.It is with a wide range of applications at aspects such as military affairs, environmental monitoring, health care, industrial automation, public safeties.
The microsensor node generally adopts powered battery to make that big characteristics of sensing net are energy constraints, in order to prolong network life overriding concern factor is to improve energy efficiency, makes node enter sleep state when need not to participate in network work or obtaining sensing data as far as possible.During network survivability, node can add and deviated from network dynamically, therefore requires network to be with good expansibility simultaneously.Another important feature of sensor network is to use specificity, upload data message entropy and incident accuracy rate in order to improve, often need polymorphic type transducer, multi-node collaborative information processing between network node, and requirement realizes the seamless covering of area to be monitored, therefore, sensor network has node layout density height, and node perceived information has characteristics such as redundancy.In addition, wireless sensor network does not rely on fixing infrastructure, and network node requires to have self organization ability.
At above characteristics, traditional flat topology controlling mechanism can not satisfy the requirement of the large-scale network-estabilishing of wireless sensor network, and can not support well that local data merges, the work of efficient managing power consumption and transducer, dormancy dispatching mechanism etc., therefore, sub-clustering mechanism becomes the topology control method of disposing and realizing the intensive laying wireless sensor network of efficient extensibility.It is compared with the flat topology structure, in convergence, fusion, energy efficiency, satisfy aspect such as network QoS requirement and can obtain preferably and improve.Existing cluster-dividing method has several classes, as low-yield self adaptation sub-clustering hierarchical algorithms (Low Energy AdaptiveClustering Hierarchy, LEACH), mix efficiency distribution cluster-dividing method (Hybrid Energy-EfficientDistributed Clustering, HEED) etc., how main concern forms cluster structured substantially, and do not consider communication mechanism between effective bunch, just simple hypothesis bunch head can one jumps to the base station, and this hypothesis is too harsh and difficult realization for the sensing node of energy constraint.Though comparatively famous S-MAC agreement has also proposed the thought of virtual cluster division, but from topology control, it still is a kind of planar structure, therefore can't effectively support the demand of large scale network, and S-MAC itself exist in design as: over-burden for boundary node, no network energy equilibrating mechanism, can't effectively support problem such as data fusion.Zigbee has proposed the topological controlling mechanism of bunch tree on the basis of 802.15.4, but this scheme is a kind of complete set Chinese style network-building method, the mode that network is laid is had relatively high expectations, and have serious restriction aspect network delay and scale extensibility.Though and some other sub-clustering algorithm considered bunch between communication mechanism, under the synchronous prerequisite of hypothesis length of a game, carry out often, and this hypothesis is difficult to realize for massive wireless sensor or network overhead is excessive.Also have communication mechanism between a class bunch be by bunch between gateway node realize, but for network life cycle is the wireless sensor network of main measurement index, to effectively keep with all neighbours on every side bunch bunch between communicate by letter, just must guarantee that gateway node is in running order between corresponding bunch, therefore, the whole sequencing control complexity of network is unfavorable for extension of network and energy-conservation.Problems such as communication between bunch is if realize based on competition access mechanism between bunch head fully, and serious interference between then can causing bunch, data transmission efficiency are low.Recently also relevant for adopt virtual multi-antenna (MIMO) technology realize bunch between communication, but this method realizes having relatively high expectations for a bunch physical layer, and to collaborative between node, synchronously and scheme is prefabricated requires highly, lacks realizability.
Summary of the invention
The objective of the invention is to, a kind of wireless sensor network distributed topology control method based on double cluster is provided.The present invention adopts following technical scheme:
A kind of wireless sensor network distributed topology control method based on double cluster may further comprise the steps:
Bunch head or bunch member's identity of each node are determined in A, physics sub-clustering, form physical cluster;
B, virtual cluster division, mode by distributed contention, bunch head of described physical cluster produces self or defers to the periodic scheduling table that adjacent cluster bunch head produces, make bunch head of adjacent physical cluster form periodic duty, the synchronous virtual cluster division of dormancy, and by the communication between described bunch of head of adaptive communications mechanism realization.The relevant position in this bunch super frame period is set according to initial relatively, the concluding time of communication period between the other side bunch the adaptive communications period between described virtual cluster.
As a kind of optimal way of the present invention, described steps A specifically may further comprise the steps:
Behind a1, the node electrifying startup, whether there is a bunch head near the scanning; If then enter step a3, otherwise change step a2;
A2, by the distributed contention mode, elect physical cluster bunch head, the identity marks of all the other nodes is a bunch member;
A3, add this bunch place physical cluster, and obey scheduling strategy cycle time of physical cluster with bunch member's identity application.
As another optimal way of the present invention, described step a2 is specially, node periodically sends the information frame that comprises self base attribute and weighted average value information, intercept, receive the information frame of neighbor node simultaneously, finishing the local message of neighbours' discovery and node collects, election algorithm according to weighted average elects physical cluster bunch head, and the identity marks of all the other nodes is a bunch member.
As an optimal way more of the present invention, after described step a2 determines bunch head or bunch member's identity of each node, described bunch of member's application behind the periodic scheduling strategy of described bunch of head establishment oneself adds this physical cluster, and bunch member of this physical cluster is one to this bunch head and jumps and can reach.
As an optimal way more of the present invention, described periodic scheduling table comprise the synchronous period, bunch in communicate by letter the period, bunch between communication period and the dormancy period, be a superframe dispatching cycle, and described bunch of head and bunch member all are provided with separately dormancy, working hour according to the periodic scheduling table of place physical cluster bunch head.
As a kind of improvement of the present invention, be specially at described step B,
After b1, described bunch of head are established identity, a period of time interscan neighbours bunch head comprise the synchronization frame of periodic scheduling epiphase to initial time information, if scan, then enter step b3, otherwise change step b2;
B2, described bunch of head judge whether to establish self periodic scheduling table, if then enter the network post stage at next cycle, otherwise produce the periodic scheduling table of self, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information;
B3, described bunch of head receive this synchronization frame, judge the periodic scheduling table that whether has self, if then enter step b4, otherwise change step b5;
Whether the periodic scheduling table that b4, described bunch of head are judged self with to receive the periodic scheduling table identical, if, then enter step b6, otherwise commentaries on classics step b7;
B5, described bunch of head are deferred to this and are received the periodic scheduling table, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information;
B6, described bunch of also more new neighbor bunch head tabulation of head record, the periodic scheduling information of these neighbours bunch list records self and neighbours' bunch head;
B7, described bunch of head judge whether there be the neighbours bunch head identical with self dispatching cycle according to neighbours' bunch head tabulation, if then enter step b8, otherwise described bunch of head abandoned self periodic scheduling table, changes step b5 then;
B8, described bunch of head be according to receiving cycle dispatch list correction self periodic scheduling table, set up and these neighbours bunch head bunch between communicate by letter the period more new neighbor bunch head tabulation, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information;
B9, described bunch of head enter the network post stage in next dispatching cycle.
As another improvement of the present invention, also comprise behind the described step B:
C, network are on duty, according to described periodic scheduling table with the superframe mode carry out periodic synchronously, bunch in communication, bunch between communication and dormancy dispatching.
One improve again as of the present invention, the periodic synchronous method of described step C is, described bunch of head is with the periodic scheduling table of a certain synchronizing cycle by synchronization frame broadcasting self, described neighbours bunch head weighed synchronously by this synchronization frame time of carrying out, newly-established physical cluster bunch head is obeyed the periodic scheduling table of existing virtual cluster division by scanning, receive this synchronization frame.
One improve as of the present invention again, comprise also among the described step C that neighbours' virtual cluster finds step, each bunch head is intercepted a synchronizing cycle in during a certain section, to find neighbours' virtual cluster.
One improve again as of the present invention, described intercepting to intercepting continuously, in this synchronizing cycle, bunch head all keeps the state of intercepting in the dormancy period of described superframe.
One improve again as of the present invention, also comprise newcomer's node invitation step among the described step C, described superframe bunch in communication the period be provided with the invitation time slot, described bunch of head inviting time slot to send the invitation frame with certain one-period, this invites frame to the newcomer by scanning, add existing physical cluster to described bunch of head application, and accept the periodic scheduling table of this physical cluster.
Distribution topology control method based on double cluster provided by the invention, it is cluster structured based on the physics with Network Load Balance characteristic and a bunch distributing homogeneity preferably, simultaneously by distributed selection with safeguard the periodic duty dispatch list of each physics sub-clustering, make bunch head of adjacent physical cluster form the autosynchronous virtual cluster division of dormancy awakening, realize communicating by letter between different bunch heads by adaptive communications mechanism, thereby need not under the synchronous prerequisite of global network, realize the local message transmission synchronously and bunch between multi-hop to the transfer of data of base station, and reduce the control expense and bunch between the frame collision probability.Change strategy based on the local bunch great wheel of virtual cluster division simultaneously and come further balance network load.The present invention has formed one and has made neighbours' bunch head backbone network topological structure of free communication each other on the basis of physics sub-clustering, form between synchronous bunch of head one do not have bunch in the virtual cluster of communication contention aware, be easy to adapt to the change of topological structure.The present invention can effectively realize Network Load Balance on the basis of satisfying the application-specific qos requirement, improve the network energy consumption efficiency, prolongs the network system life-span.According to the wireless sensor network that the present invention is set up, under single-channel condition, can realize, save the physical layer design cost greatly.
Further specify the present invention below in conjunction with drawings and Examples.
Embodiment
For making things convenient for the description of the embodiment of the invention, introduce earlier the superframe structure that uses in the embodiment of the invention and the property parameters of day part.Be a superframe dispatching cycle.Described periodic scheduling table has been stipulated a physical cluster bunch head and bunch member's work, dormancy dispatching mechanism, mainly comprise the synchronous period, bunch in communicate by letter the period, bunch between communication period and the dormancy period, be a superframe dispatching cycle, and described bunch of head and bunch member all are provided with separately work, dormancy period according to the periodic scheduling table of place physical cluster bunch head.As shown in Figure 1, complete superframe T
SfComprise the following period:
Be synchronous period T synchronously between bunch
Sync, for a bunch hair send the synchronization frame setting, the regular hour surplus was arranged in this period, send by CSMA-CA mechanism when making each transmission synchronization frame.This period can be adopted common signal channel, big level transmitting power.
Communication period T in bunch
Inter, invite time slot in mainly being divided into bunch
Tin vite, bunch inter-sync time slot T
Sync_pcWith bunch in uplink communication period T
IntraBunch member woke up in this period, all was in resting state in other periods of superframe, to save energy consumption.Wherein:
Invite time slot T in bunch
In viteDecide the transmission gap periods of inviting frame by bunch head according to the frequent degree that newcomer's node adds network, when newly added node listen to invite time slot after, promptly subsequently bunch in the application of communication period add this physical cluster, this time slot adopts common signal channel, more rudimentary transmitting power.
Bunch inter-sync time slot T
Sync_pc, between bunch head and bunch member synchronously and a bunch downlink frame be provided with, bunch head that synchronization frame mainly causes in order to correct clock drift and the time deviation of each bunch member node, downlink frame comprise mainly that number order controls and time slot reservation etc.This time slot adopts local channel, more rudimentary transmitting power.
Uplink communication period T in bunch
Intra, be mainly bunch member node upstream data to bunch head and being provided with, two main period of time are arranged: the non-competing up period, compete the period at random.Be divided into some time slots in the non-competing up period, preengage to a bunch head in the up period of competition at random of a last super frame period as required by bunch member node, in bunch inter-sync time slot of this superframe, broadcast the operating position of non-competing up period by bunch head then by the time slot reservation frame.Competing the up period at random adopts time slot CSMA-CA access mechanism to communicate with a bunch head by bunch member node.The uplink communication period is adopted local channel, more rudimentary transmitting power in bunch.
Communication period T between bunch
Inter, be mainly same virtual cluster intra-cluster head intercommunication and be provided with, guarantee in virtual cluster, to carry out data communication with less time delay.An if bunch CH
iHas only a kind of dispatching cycle, i.e. CH
iNeighbours' bunch head all in same virtual cluster division, then bunch between the communication period as long as communication period between safeguarding bunch.This period is adopted common signal channel, big level transmitting power.
Adaptive communications period T between virtual cluster
Inter-vc, be mainly and intercept communication between neighbours' virtual cluster bunch and be provided with.As a bunch CH
iWhen being positioned at two virtual cluster and having a common boundary, i.e. CH
iThere is neighbours' bunch head to be positioned at another virtual cluster, then CH
iNeed safeguard the adaptive communications period between virtual cluster, and for CH
iAll neighbours bunch head is all in same virtual cluster division the time, the adaptive communications period between the Maintenance free virtual cluster.This period is adopted common signal channel, big level transmitting power.
Dormancy period T
SleepIf bunch head does not have a common boundary two virtual cluster, then enters the dormancy power save mode, the communication module of node is all closed.
The length of above-mentioned day part exists as follows:
T
sleep>T
intra>T
inter>T
inter-vc>T
sync>T
sync_pc>T
in?vite。
A kind of wireless sensor network distributed topology control method based on double cluster may further comprise the steps:
Bunch head or bunch member's identity of each node are determined in A, physics sub-clustering, form physical cluster; As shown in Figure 1, specifically may further comprise the steps:
Behind a1, the node electrifying startup, whether there is a bunch head near the scanning; Promptly whether there is the physical cluster that has formed near the scanning; In order to add existing physical cluster in the network as far as possible, on common signal channel, scan earlier V
CThe individual cycle is T
SfDuration, by detecting the invitation frame (Invite-Frame) that bunch hair goes out, whether have physical cluster around differentiating.N
CAdd and factor decisions such as frequent degree of node motion and network lifetime by the newcomer.As set up initial stage N in physical cluster
CBe worth greatlyyer, bunch head can send with higher frequency and invite frame, and at physical cluster existence later stage N
CBe worth less.T
SfBe the duration of periodic scheduling table.Invite and to comprise in the frame as signal post in this bunch periodic scheduling table, this bunch bunch with information such as channel, this bunch member saturations.If there has been physical cluster on every side, then enters step a3, otherwise change step a2;
A2, by the distributed contention mode, elect physical cluster bunch head, the identity marks of all the other nodes is a bunch member; This step specifically can be, node periodically sends and (can adopt carrier sense conflict of the prior art to avoid access mode, Carrier Sense Multiple Access-Collision Avoid, CSMA-CA) comprise the information frame of self base attribute and weighted average value information, intercept, receive the information frame of neighbor node simultaneously, finishing the local message of neighbours' discovery and node collects, election algorithm according to weighted average elects physical cluster bunch head, and the identity marks of all the other nodes is a bunch member.And after bunch head or bunch member's identity of determining each node, described bunch of member's application behind the periodic scheduling strategy of described bunch of head establishment oneself adds this physical cluster, and bunch member of this physical cluster is one to this bunch head and jumps and can reach.Specifically can realize in the following way: do not receive the invitation frame in scan period, enter a bunch head competition election stage, each node is with T
hPeriodic broadcasting HELLO frame is collected information of neighbor nodes simultaneously.The weighted average that comprises node self in the HELLO frame.Node upgrades the weighted average of self network service cost by the HELLO frame that receives neighbor node, broadcasts by the HELLO frame after the renewal again.Node sends the HELLO frame at every turn and all adopts the CSMA-CA mode, and this process continues T
nAfter can reach the local nodes consistency on messaging, T
n=T
h* N
h, N
hBe HELLO frame number of broadcast times, N
h, T
hAll can dispose according to the node layout density.Wherein, node distributes according to collected neighbor node weighted average, and for example based on definite bunch head of minimum weight mean value algorithm election and bunch member, this strategy can adopt existing techniques in realizing to the strategy of use, in the detailed description of this omission to this part.
A3, add this bunch place physical cluster, and obey scheduling strategy cycle time of physical cluster with bunch member's identity application.Be that node is received the invitation frame, and subsequently bunch in communication the period rise and to join request to a bunch hair, obtain beginning after bunch head permission periodic scheduling table work by this bunch.
After finishing the physics sub-clustering, though bunch member has been known bunch head of place physical cluster, bunch member wouldn't formally add physical cluster to bunch head application, and the periodic scheduling table until physical cluster head establishment oneself begins to enter the network stage on duty.And physical cluster bunch member is one to a bunch head and jumps and can reach.In this stage, each node transmitting power is arranged on lower level, and this power level guarantees that statistically neighbours' number that node has a some gets final product.
B, virtual cluster division, after the physical cluster election of cluster head is finished, mode by distributed contention, bunch head of described physical cluster produces self or defers to the periodic scheduling table that adjacent cluster bunch head produces, make bunch head of adjacent physical cluster form periodic duty, the synchronous virtual cluster division of dormancy, and by the communication between described bunch of head of adaptive communications mechanism realization.Virtual cluster is an in esse bunch of head not, and is meant the virtual set of the physical cluster bunch capitiform one-tenth of safeguarding same periodic scheduling table.In this stage, each bunch hair is penetrated power setting in higher level, and this power level guarantees that statistically each bunch head has a suitable neighbours bunch number.Exchange cycle dispatch list between bunch head is promptly informed neighbours' initial relatively moment in a bunch local scheduling cycle in synchronization frame.
Wherein, described virtual cluster division step B can specifically adopt mode shown in Figure 3 to realize, specifically describes as follows:
After b1, described physical cluster bunch of head established identity, can adopt the CSMA-CA mode to intercept wireless channel equally, comprise the synchronization frame of periodic scheduling epiphase to initial time information, for example scan N superframe T dispatching cycle at a period of time interscan neighbours bunch head
Sf, had virtual cluster division with around finding.To guarantee that bunch head is at N superframe T dispatching cycle after network enters the stage on duty
SfIn successfully send synchronization frame one time with higher probability.If scan, then change step b3, otherwise enter step b2;
B2, described bunch of head are judged the periodic scheduling table of whether having established self, if then enter the network post stage in next dispatching cycle; Otherwise, produce the periodic scheduling table of self, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information; Promptly with current time t
iSet up the periodic scheduling table of oneself for benchmark.At this dispatching cycle of T
SfThe synchronous period in (t
i, t
i+ T
Sync) in the CSMA-CA mode to a neighbours bunch CHs
iCollect CMs with a bunch member
iThe synchronization frame of local cycle dispatch list is announced in broadcasting;
B3, described bunch of head receive this synchronization frame, judge the periodic scheduling table that whether has self, if then enter step b4, otherwise change step b5;
Whether the periodic scheduling table that b4, described bunch of head are judged self with to receive the periodic scheduling table identical, if, then enter step b6, otherwise commentaries on classics step b7;
B5, described bunch of head are deferred to this and are received the periodic scheduling table, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information; Promptly adopt and receive CH
kThe periodic scheduling table, establish CH
KThe initial moment of this dispatching cycle is t
k, current time is t, and remains the period (t, t in this dispatching cycle
k+ T
Sf) in CSMA-CA random back mode to a neighbours bunch CHs
iAnd bunch in member node CMs
iBroadcast the periodic scheduling table of oneself, and when begin next dispatching cycle t
k+ T
SfEnter the network post stage;
B6, described bunch of also more new neighbor bunch head tabulation of head record, the periodic scheduling information of these neighbours bunch list records self and neighbours' bunch head; Promptly write down CH
kThe period information of communicating by letter between dispatch list bunch is revised CH according to the periodic scheduling table that receives then
iThe periodic scheduling table, establish CH
iThe initial moment of this dispatching cycle is t
iCurrent time is t, and remains the period (t, t in this dispatching cycle
i+ T
Sf) in the CSMA-CA mode to a neighbours bunch CHs
iAnd bunch in member node CMs
iBroadcasting CH
iThe periodic scheduling table, and enter the network stage on duty by new dividing time-steps in next dispatching cycle, this kind situation shows a bunch CH
i, CH
kBe in the boundary of two virtual cluster, two bunches of heads all need to know the other side's dispatching cycle, and revise the local cycle dispatch list, superframe in dispatching cycle reserved time slot keep two proper communications between virtual cluster;
B7, described bunch of head judge whether there be the neighbours bunch head identical with self dispatching cycle according to neighbours' bunch head tabulation, if then enter step b8, otherwise described bunch of head abandoned self periodic scheduling table, changes step b5 then;
B8, described bunch of head be according to receiving cycle dispatch list correction self periodic scheduling table, set up and these neighbours bunch head bunch between communicate by letter the period more new neighbor bunch head tabulation, and broadcasting comprises self periodic scheduling epiphase synchronization frame to initial time information;
B9, described bunch of head enter the network post stage in next dispatching cycle.
C, network are on duty, according to described periodic scheduling table with the superframe mode carry out periodic synchronously, bunch in communication, bunch between communication and dormancy dispatching.The periodic synchronous method of described step C is, described bunch of head is with a certain synchronizing cycle of (each bunch CH for example
iWith N superframe T dispatching cycle
SfBe T synchronizing cycle
s=N*T
Sf, promptly at N
SyncSuccessfully send synchronization frame in the individual super frame period one time) periodic scheduling table by synchronization frame broadcasting self, described neighbours bunch head weighed synchronously by this synchronization frame time of carrying out, newly-established physical cluster bunch head is obeyed the periodic scheduling table of existing virtual cluster division by scanning, receive this synchronization frame.
Wherein, owing to may there be problem such as broadcast frame conflict in the virtual cluster division stage, thereby cause neighbours' bunch head can not successfully mutual periodic scheduling table, promptly can't proper communication between two virtual cluster; The virtual cluster of newly-established in addition virtual cluster and vicinity is owing to the periodic scheduling table is different, possibly can't normally receive the other side's synchronization frame, for guaranteeing during network is on duty, can find the existence of neighbours' virtual cluster, guarantee the connectedness of network, as a kind of improvement of the present invention, can comprise neighbours' virtual cluster discovery step among the described step C, because in the network stage on duty, bunch head can be to send synchronization frame certain synchronizing cycle at interval, therefore each bunch head is intercepted a synchronizing cycle in during a certain section, can realize the discovery of neighbours' virtual cluster.Described intercepting can be to intercept continuously, and in this synchronizing cycle, bunch head all keeps the state of intercepting in the dormancy period of described superframe.An i.e. bunch CH
iOne section long time T of continuous operation
Find, be T synchronizing cycle one section length that is right after then
sTime in, a bunch CH
iAll keep intercepting wake-up states in the dormancy period, the working mechanism of other periods remains unchanged.Because the synchronization mechanism setting of bunch head, in this time range, each bunch head all can successfully send synchronization frame one time with higher probability.Therefore, by neighbours' virtual cluster discovery mechanism, can find that neighbours' virtual cluster sets up the efficient communication between different virtual bunch.T
FindCan dynamically adjust according to application requirements and neighbours' bunch number etc.
Wherein, when network work on duty, may have new node and add network, perhaps the minority node moves to another physical cluster from a physical cluster, as a kind of improvement of the present invention, also comprise newcomer's node invitation step among the described step C, described superframe bunch in communication the period be provided with the invitation time slot, described bunch of head inviting time slot to send the invitation frame with certain one-period, this invites frame to the newcomer by scanning, add existing physical cluster to described bunch of head application, and accept the periodic scheduling table of this physical cluster.Described bunch of hair send the periodicity of inviting frame to determine according to the frequent degree of newcomer's adding and node motion.