CN108683468A - AUV mobile data collection algorithms in underwater sensing network based on data prediction - Google Patents
AUV mobile data collection algorithms in underwater sensing network based on data prediction Download PDFInfo
- Publication number
- CN108683468A CN108683468A CN201810390515.5A CN201810390515A CN108683468A CN 108683468 A CN108683468 A CN 108683468A CN 201810390515 A CN201810390515 A CN 201810390515A CN 108683468 A CN108683468 A CN 108683468A
- Authority
- CN
- China
- Prior art keywords
- auv
- cluster
- data
- node
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/373—Predicting channel quality or other radio frequency [RF] parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/364—Delay profiles
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
- H04B17/3913—Predictive models, e.g. based on neural network models
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
AUV mobile data collection methods in the invention discloses a kind of underwater sensing network based on data prediction, including：AUV initializes maximum neighbor node density sub-clustering with centralized algorithm, prediction is fitted to gathered data with SVR, establish prediction model, sub-clustering update is carried out according to the similarity of prediction model trend, AUV is synchronous with corresponding cluster to preserve identical prediction model, the access that these clusters are skipped in data-gathering process is collected then is directly predicted with prediction model, when corresponding cluster prediction model is more than preset maximum tolerance error threshold or latency sensitive threshold value, update request is then sent immediately, the current locations AUV are found using bidirectional research and inform variation, AUV plans residual paths to obtain new prediction model again.By data prediction, AUV traverse paths length is reduced to reduce AUV energy consumptions and collect time delay, while reducing whole network data volume, reduce part of nodes energy consumption, improve overall network performance.
Description
Technical field
The invention belongs to fields, and in particular to AUV mobile datas are received in a kind of underwater sensing network based on data prediction
Set algorithm.
Background technology
With the continuous hair of underwater sensor network (Underwater Wireless Sensor Networks, UWSNs)
Exhibition, by diversified submerged applications, we can obtain more and more specifying informations about ocean or river.For example, logical
Temperature and the sulfur dioxide concentration variation for crossing underwater sensing net monitoring submarine volcano region, predict volcano state, to possible with this
The eruptivity of appearance carries out early warning；And by monitoring military area, prevent enemy's submarine, the invasions such as battleship.However it is underwater
Sensor Network is different from land Sensor Network, and under water, radio signal decaying increases with frequency increase and much larger than land
Decaying, this is a mortal wound for underwater telecommunication.Therefore, underwater wireless communication generally use sound leads into line number
According to transmission, while sound communication is similarly limited to transmission range, studies have shown that when sound leads to transmission range more than threshold value, energy consumption meeting
Exponentially rise again, this underwater sensor just supplemented for node energy inconvenience is a stern challenge.
Due to the above problems, carrying out data collection using AUVs short distance single-hop transmissions in Sensor Network under water is
One feasible scheme.In the case where AUVs energy is enough, AUVs traverses whole network.Due to AUVs speed relative to
Acoustic speed is partially slow, and a caused prominent question is exactly that collect time delay long, therefore considers how that it is one to reduce time delay
The solution of focus, mainstream has Sub-region and hierarchical, covers redundant area prototype selection and path optimization etc..And this calculation
Method is mainly set about from shortening AUVs collecting paths, and passage path reduction reduces delay and whole network data traffic.
Shorten AUV data collections path, most simple most common method is the thought of sub-clustering layering.Yuh-Shyan Chen
Deng 2013《IEEE SENSORS JOURNAL》On deliver " Mobicast Routing Protocol for
Underwater Sensor Networks " carry out 3D-ZOR region divisions, are a spheric regions centered on AUV.AUV
Obtain the position (x of oneselfA,yA,zA), and create 3-DRegion Zt(Ni), wherein：Zt(Ni)=(xi-xA)2+(yi-yA)2+
(zi-zA)2-R2, R is the integral multiple of communication radius, and AUV can use the mode of multicast to send out the control for including area information
Packet, and advance along predefined paths, constantly collect data in sensor node of the different time out of a series of 3-D ZOR.
Shanshan Li et al. 17 years exists《Sensor》Deliver " Probabilistic Neighborhood-Based Data
Collection Algorithms for 3D Underwater Acoustic Sensor Networks " are successfully passed with maximum
Defeated probability is radius, and carrying out probability neighborhood covering collection based on neighbor node density establishes, and AUV is heuristic using improved arest neighbors
Strategy traverses each probability neighborhood covering collection node, reduces the length of traversal whole network.But this kind of method, it improves
Arest neighbors heuristic strategies and the path planning algorithms such as user's predefined paths be not optimal, if secondly node mistake
More, the quantity of cluster also can be excessive, the collecting path limited length of shortening.That similar is exactly layering thought, Pir Masoom Shah
Deng 2016《IEEE Computer Society》Deliver " MobiSink:Cooperative routing protocol for
Underwater sensor networks with sink mobility " partition the network into multiple level courses, in each area
The node data of single-hop or multi-hop in the mobile sink linear movement receiving area disposed in domain, due to multiple mobile sinks and often
A is responsible for one layer, and displacement distance substantially reduces.But mobile sink cooperation be a difficult point, there is no described in article so
Simply.Arshad Sher etc. were in 17 years《International Journal of Distributed Sensor
Networks》Publish an article " Monitoring square and circular fields with sensors using
energy-efficient cluster-based routing for underwater wireless sensor
Networks ", it is proposed that a kind of energy saving Clustering protocol based on sparse perception, it is sparse for identification that network is divided into multiple regions
Region and dense Region, if the node number in region is more than threshold value, region is dense Region, is otherwise sparse region.
Dump energy is high in dense Region, and the node of depth as shallow becomes cluster head.Node broadcasts hello packets after becoming cluster head, collects cluster
Interior data.Mobile sink each round, which can be moved to another sparse region while be received from a sparse region, comes from dense Region
The data that cluster head is sent.Subregion thought can be with, but the distribution of sparse and dense Region is not necessarily uniform, and mobile sink is collected
When do not ensure that collection all areas, it is excessive to carry out sending control packet in addition when network partition, wastes time and energy, unfavorable
In network performance.
It is to obtain coverage collection that another kind, which shortens Path Method then, selects representative point therein as AUV data collections
Park point, also referred to as travel a little, a park point is collected simultaneously multiple node datas, avoids the unnecessary movements of AUV.Ke Li et al.
In 10 years《IEEE Mobile Ad hoc and Sensor Systems》Deliver " Energy-constrained Bi-
Objective Data Muling in Underwater Wireless Sensor Networks " propose UDMP methods,
It maximizes each collector node number and reduces cruising time to the greatest extent, the shortcomings that this method is once access institute
There is node.In order to solve this problem, Shih-Hao Chang etc. were in 15 years《International Conference on
Network-Based Information Systems》On publish an article " Tour Planning for AUV Data
Gathering in Underwater Wireless ", it is proposed that find the method for representing point, calculate node spherical shape communication range
External cube overlapping region, it be representative point to select overlapping cuboid center, while use information is worth, distance and when transmitting
Between calculate corresponding weight, according to weight order accessed node to avoid repeated accesses during data collection.
Also one is region divisions and deployment travelling point to combine, and traversal collection is carried out by disposing multiple AUV.Jawaad
Ullah Khan etc. 16 years exist《Sensor》On deliver《Data-Gathering Scheme Using AUVs in Large-
Scale Underwater Sensor Networks:A Multihop Approach》By the way that node deployment is planarized,
Thiessen polygon is divided in plane, and whole network is divided into multiple regions, one AUV of each regional deployment, while each area
AUV chooses coverage redundancy and carries out path planning as travelling point in domain, this kind of method whole concept is thought of dividing and ruling, but
It is that AUV specifically disposes quantity and do not have specific theories integration, secondly cooperates highly difficult, how to collect each AUV between AUVs
Collect fast-forwarding to sink be a prodigious problem.
Consider algorithm above, proposes that the delay based on data prediction optimizes AUV mobile data collection algorithms, by dividing
Cluster thought shortens a part of paths AUV, then advanced optimizes reduction traverse path length with prediction model, simultaneously drops
Low whole network data total flow promotes network performance.
Invention content
In order to solve to cause to collect using AUV mobile data collections in underwater sensor network to postpone excessive, information value
Constantly devalue over time, considered existing reduction delayed data collection scheme, the present invention proposes pre- based on data
AUV mobile data collection algorithms in the underwater sensing network of survey are fitted prediction, according to prediction using SVR algorithms to data
Trend carries out sub-clustering update, AUV and corresponding cluster prediction model having the same, is skipped in data-gathering process to these clusters
Access transfer directly to be predicted with prediction model, in this way shorten AUV traverse paths length to accelerate collection process.Simultaneously
When corresponding cluster prediction model update, find the positions AUV using bidirectional research and inform the fact that model changes, AUV according to
The urgency of request plans residual paths to obtain new prediction model again.
It realizes above-mentioned technical purpose, reaches above-mentioned technique effect, the invention is realized by the following technical scheme：
A kind of AUV mobile data collection methods in the underwater sensing network based on data prediction, including：
(1) AUV obtains whole network node deployment situation from water surface sink nodes, and node is initialized using centralized algorithm
Sub-clustering.
AUV centralized algorithms carry out initialization sub-clustering-and are based on neighbors density cluster algorithm.In view of being carried out just by node
Beginningization sub-clustering, node need to send a large amount of control packet switch information, this process not only takes considerable time, while can be lost
Node energy, therefore, it is a rational selection that centralized approach, which carries out initialization,.Detailed process is：
1.1：AUV obtains network deployment information, calculates neighbor node number in each node its communication radius；
1.2：It is arranged according to neighbor node quantity descending, selects the node with most neighbor nodes as cluster head, it is adjacent
Node is occupied as cluster member, while these vertex ticks are to have selected；
1.3：To be deleted from other nodes neighbors nodes labeled as the node selected, repeat before the step of until all
Node is labeled.
(2) after AUV traversals initialization cluster, acquisition completes current pass data and uses supporting vector according to the data of acquisition
Machine algorithm establishes prediction model, while updating initialization cluster, and the node with same or similar anticipation trend is formed new cluster.
SVR, that is, Support vector regression is a kind of theoretical based on Statistical Learning Theory and VC, estimates letter using perception data
Several methods.The reason of selecting the algorithm mainly has following：
First, underwater sensor node is not after all computer, and computing capability is limited, so needing to select simple SVR
Carry out data fitting and prediction；
Secondly, interfere under water bigger, data error relatively also can be larger, and data are largely non-linear, corresponding
SVR finds an error tolerance highest straight line and data is fitted and is predicted under linear conditions, under nonlinear situation
Nonlinear data is mapped to after high dimensional plane makes it linearize using kernel function and carries out data fitting and prediction；
Finally, SVR has complete theoretical system and proves to derive, and feasibility is high.
Forming new cluster is specially：
2.1：Node perceived data, a node once send multiple data packets AUV to collection of coming, AUV by data
According to 6:2:2 ratio is divided into training set, verification collection and test set；
2.2：According to SVR formula, suitable kernel function is selected to calculate corresponding w vector sums b vectors, according to y=wK
(x)+b fits corresponding curve；Wherein, y indicates that predicted value, K (x) indicate that kernel function time series, w expression parameters, b indicate
Deviation, b i.e. W=[w, b] included in the matrix of w；
2.3：AUV reconfigures the node with identical anticipation trend to form new cluster, to avoid scale is excessive from causing
Inter-node communication, the problems such as energy consumption, for arrange parameter Hop=n to control cluster scale, n indicates node hop count.
(3) AUV traverses whole network again according to new sub-clustering situation, and prediction model is transmitted to corresponding cluster knot
Point, follow-up AUV no longer traverses newly-established cluster, and is directly predicted using prediction model.
Again whole network is traversed, specially：
3.1：AUV broadcasts initial position message, and leader cluster node is calculated at a distance from AUV after receiving this information, calculated
I-th of cluster and AUV initial distances：
Wherein (a, b, c) initial AUV position coordinates, (xi,yi,zi) it is i-th of cluster position coordinates；
Data volume total in cluster and the dump energy of leader cluster node are obtained simultaneously, these information are stored in control message packet
Carry out broadcast request；
3.2：AUV receives these control message packets, and the coefficient of competition of request cluster is calculated according to formula,
Calculate coefficient of competition：
Wherein num indicates that num cluster, M are network boundary length, and F (i) indicates i-th of cluster data flow, andIt represents
All cluster total data flows of whole network；E0Node primary power, E (i) are then i-th of cluster dump energies；α and χ coefficients；Meet
α ＞ χ ＞ (1- α-χ)；
Select the maximum cluster of coefficient of competition as path starting point, using shortest path first, that is, Dijkstra's algorithm into
Row traversal, while prediction model is transmitted to corresponding cluster node, this path is as original path；
3.3：Second leg, AUV traverse all clusters for not receiving prediction model and carry out data collection again, form new road
Diameter.
(4) bunch member node for obtaining prediction model voluntarily perceives prediction data, if the error of perception and prediction is more than
The worst error of setting is tolerated, updates prediction model, and send request and inform AUV.
Updating prediction model is specially：
4.1：Maximum tolerance error threshold δ and adjustable delay sensitive threshold gamma are set；
4.2：Node calculates the square error and J of the time series data perceived, if J ∈ [0, δ), acquiescence does not need
Update, whereas if J ∈ [δ ,+∞), necessarily update；
4.3：Calculate whole section of time series data square error and while, calculate individual data predicted value square mistake
Abnormal data is inserted into abnormal data queue (EQ), if size (EQ) ＞ γ, update prediction model immediately by difference if more than δ
And inform AUV by sending solicited message.
(5) AUV, which receives forecast updating request, will suspend current running orbit, be planned again using request cluster as park point
Data collection path, while obtaining data and updated prediction model.
5.1：Cluster is asked to search for the positions AUV from both direction clockwise and anticlockwise along original path；
5.2：The request that AUV once receives a direction currently cruises pause, considers request cluster to plan road again
Diameter, while one agent node is set in current location, once the request message in another direction reaches, agent node will be obtained
Message does not continue to search, request terminates to packet loss.
Beneficial effects of the present invention：
The present invention proposes AUV mobile data collection algorithms in a kind of underwater sensing network based on data prediction, for AUV
Mobile data collection postpones excessive problem, and algorithm is reduced in conjunction with previous delay, cluster can be predicted by being established with SVR, simultaneously
AUV obtains corresponding prediction model and transfers directly to predict without repeated accesses data.In addition, when prediction data and sense
Primary data error, which is more than worst error tolerance and can send request message, informs AUV, AUV will again access request cluster it is new to obtain
Prediction model, by this sequence of operations, AUV access path shortens, and reduces and collects time delay, while cluster being asked only to need to send
The simple signal that whether updates reduces node-node transmission energy consumption without sending complete prediction model, realizes that network performance is promoted.
Description of the drawings
Fig. 1 is the assumed condition schematic diagram of an embodiment of the present invention；
Fig. 2 is the network model schematic diagram of an embodiment of the present invention；
Fig. 3 is that request carries out bidirectional research schematic diagram；
Fig. 4 is AUV work flow diagrams；
Fig. 5 is node work flow diagram.
Specific implementation mode
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that specific implementation described herein is only used to explain the present invention, it is not used to limit
The fixed present invention.
The present invention postpones excessive problem for AUV mobile data collections, updates clustering architecture with SVR, is formed a kind of special
Cluster can be predicted in different cluster-, and cluster can be predicted by the corresponding prediction model of acquisition in AUV to avoid repeated accesses, therefore shortens and collect
Path length.Specifically：The AUV starting stages receive the coefficient of competition of all clusters, select maximum contention coefficient as starting point,
Start data collection with shortest path；After AUV accessed predictable cluster, corresponding prediction model is obtained, AUV is in second leg
Repeated accesses are no longer carried out, are directly predicted, update solicited message until predictable cluster sends out prediction model, AUV is accessed and obtained again
Take new prediction model.
The present invention is further described below in conjunction with the accompanying drawings.
It is underwater sensing network assumed condition schematic diagram as shown in Figure 1, the underwater sensing network is a M × M × M
3D region, underwater sensing network include a water surface sink node and several sensor nodes, the sink nodes
Positioned at water surface center；Under water in the 3D region of sensing network, underwater movement is a Mass disturbance, a section
The variation of point data not can be shown that movable generation, and underwater movement source node and movable ordinary node are an entirety, therefore,
The generation of underwater movement and change are with uniformity, while variation or constant simultaneously.Corresponding to prediction model with cluster to be whole,
Change simultaneously.
It is illustrated in figure 2 underwater sensing network model schematic diagram, the underwater sensing network is the three-dimensional of a M × M × M
Region, underwater sensing network include a water surface sink node, several sensor nodes and an AUV, the sink
Node is located at water surface center；Underwater sensor node location is opposing stationary, and the communication radius of node is adjustable in a certain range, section
Point knows self-position, and AUV energy is unrestricted to carry out data collection with constant speed.
AUV mobile data collection algorithms in a kind of underwater sensing network based on data prediction, specifically include following steps：
(1) AUV carries out initialization sub-clustering using centralized algorithm, and reduction is largely controlled as caused by Node distribution formula sub-clustering
Message redundancy.Using based on neighbors density cluster algorithm, it ensure that node maximum transmission distance is a jump, while density is got over
Greatly, the interstitial content of single collection is more, and collection frequence is smaller.Specific mainly point three steps：
(1.1) AUV obtains nodes deployment information from water surface sink, calculates in each node its communication range
Node number, that is, neighbor node number；
(1.2) AUV establishes neighbor node table, is arranged according to neighbor node quantity descending, and selection has most neighbor nodes
Node as cluster head, neighbor node is collectively labeled as having selected as cluster member, while by these nodes；
(1.3) node of label is deleted from the neighbor node of other nodes, repeats neighbor node number and calculates, sequence
Process and selection course constantly repeat, until all nodes are all labeled.
(2) current pass data are completed in AUV traversals initialization sub-clustering, acquisition, according to gathered data, are carried out using SVR pre-
Model foundation is surveyed, and updates sub-clustering result；Specifically include following steps：
(2.1) target：
Wherein λ is parameter, and N is data bulk, wTIt is parameter, w with wTznIndicate the prediction result of data on the domains z, ynIt indicates
Actual value,Indicate mean square deviation.
Final result is necessarilyw*Indicate optimized parameter, βnParameter, znIt is z numeric field datas.
So target formula can be rewritten into
Introduce kernel function concept：The operation of the two steps of merging Feature Conversion and calculating inner product is called
Kernel Function, i.e. K (xn,xm)=znzm, target formula is further rewritten asIt can be expressed as after vectorizationIt minimizes i.e. derivation,
β=(λ I+K)-1Y, then can be obtained optimal parameter w*, finally obtain prediction model y=w*·K(x)+b。
(2.2) AUV is by the new cluster that re-forms with same or similar prediction model, dump energy highest in cluster
Node be chosen as leader cluster node, while limiting new cluster scale using parameter Hop=n, communicated caused by avoiding scale excessive with
And energy consumption problem.
(3) AUV traverses whole network again according to new sub-clustering situation, and prediction model is transmitted to corresponding cluster knot
Point, follow-up AUV no longer traverses newly-established cluster, and is directly predicted using prediction model.
New sub-clustering information and initial position message are broadcasted whole network node by AUV, and node is according to this information updating
Sub-clustering, while sending solicited message and AUV is promoted to access collection.The specific steps are：
(3.1) each leader cluster node obtains AUV initial position messages, calculates i-th of cluster and AUV initial distancesWherein (a, b, c) initial AUV position coordinates, (xi,yi,zi) it is i-th of cluster position
Set coordinate；In addition, cluster head obtains data traffic F in clusteriAnd dump energy Ei, calculate coefficient of competitionWherein num indicates num cluster, and M is network boundary length, F (i)
Indicate that i-th of cluster data flow, E (i) are then i-th of cluster dump energies；E0Node primary power, α and χ coefficients；Consider distance
In contrast it is most important factor, meets α ＞ χ ＞ (1- α-χ).Coefficient of competition broadcast is transmitted to AUV by leader cluster node.
(3.2) AUV receives all clusters to coefficient of competition, and shortest path rule are carried out by starting point of maximum contention coefficient cluster
It draws, AUV, which is accessed, can be predicted cluster while prediction model can be transmitted to corresponding cluster, and be marked, and second leg, AUV is not revisited
It asks the cluster for collecting label, directly prediction model is used to carry out data prediction, so as to shorten AUV collecting paths.
(4) bunch member node for obtaining prediction model voluntarily perceives prediction data, if the error of perception and prediction is more than
The worst error of setting is tolerated, updates prediction model, and send request and inform AUV, specifically：
(4.1) the default exception queue EQ of node, a worst error tolerate δ, a delay sensitive threshold gamma, can be with
It requiring to be adjusted according to delay, if delay requires height, needs to update rapidly, then γ settings are relatively small, otherwise similarly.
(4.2) node calculates the data mean square deviation newly perceived,Wherein N indicates data volume, yiIt indicates
Perception data value, yi' indicate prediction data value.If J ∈ [0, δ), without update, if J ∈ (δ ,+∞), update immediately；
(4.3) it while calculating mean square deviation, calculates individual data and abnormal data is entered EQ if more than δ with predicted value error
Queue sends update request message at once as size (EQ) ＞ γ.
(5) as shown in Figure 3：Cluster to be updated sends a request message to AUV by bidirectional research along initial path, and AUV connects
An agent node is arranged in current location in the request message for receiving a direction, and AUV obtains cluster location information to be updated, will
Remaining cluster re-starts shortest path planning to obtain updated prediction model.Meanwhile when the request message in another direction
Agent node is reached, Forwards Forwarding will not continued to and transfer packet loss, request terminates.
To sum up：As shown in figure 4, specific work process is as follows in a network by AUV：
401) AUV obtains global deployment information, and initialization sub-clustering is carried out using the cluster algorithm based on neighbor node density；
402) AUV receives the coefficient of competition that all clusters are sent；
403) AUV is according to the coefficient of competition of reception, selects maximum as starting point, and shortest path first is used to carry out just
Beginningization path planning；
404) AUV is calculated using SVR and is obtained prediction model；
405) AUV is run along initial path, prediction model is transmitted to corresponding cluster, while updating new route, is skipped visit
Ask predictable cluster；
406) judge whether to receive prediction model update request；
407) prediction model is updated.
As shown in figure 5, the node course of work is as follows：
501) node, which receives AUV and is sent to sub-clustering information, carries out initialization cluster；
502) leader cluster node obtains distance, flow and dump energy, calculates and accordingly arrives coefficient of competition, and is forwarded to AUV；
503) leader cluster node forwards data to the AUV for accessing this cluster；
504) node receives new sub-clustering information update sub-clustering；
505) node obtains prediction model；
506) whether node updates prediction model according to predicted value and perception value error judgment；
507) bidirectional research sends prediction model update request.
In summary：
AUV mobile data collection algorithms in the invention discloses a kind of underwater sensing network based on data prediction, including：
Prediction model is established to perception data with SVR, to carrying out sub-clustering update with same or similar prediction model；AUV
The coefficient of competition sent according to request cluster determines path starting point, to carry out path planning with shortest path first；AUV is excellent
Change and access collecting path, skip predictable cluster, shortens and access collecting path；Node is sentenced by the mean square deviation of predicted value and perception value
It is disconnected whether to carry out error update, while threshold value is set and considers update delay sensitive.By this serial procedures, shorten entire number
According to AUV access path length during collection, simultaneously, also effectively reduces data total flow in network and reduce node
Energy consumption improves whole network performance.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (6)
1. a kind of AUV mobile data collection methods in underwater sensing network based on data prediction, which is characterized in that including：
(1) AUV obtains whole network node deployment situation from sink nodes, and node clustering is initialized using centralized algorithm；
(2) after AUV traversals initialization cluster, acquisition is completed current pass data and is calculated using support vector machines according to the data of acquisition
Method establishes prediction model, while updating initialization cluster, and the node with same or similar anticipation trend is formed new cluster；
(3) AUV traverses whole network again according to new sub-clustering situation, and prediction model is transmitted to corresponding cluster node, after
Continuous AUV no longer traverses newly-established cluster, and is directly predicted using prediction model；
(4) bunch member node for obtaining prediction model voluntarily perceives prediction data, if the error of perception and prediction is more than setting
Worst error tolerance, update prediction model, and send request and inform AUV；
(5) AUV, which receives forecast updating request, will suspend current running orbit, using request cluster as park point again layout data
Collecting path, while obtaining data and updated prediction model.
2. AUV mobile data collection methods in the underwater sensing network according to claim 1 based on data prediction, special
Sign is：Centralization initialization node clustering in the step (1) is based on neighbor node density sub-clustering, specially：
2.1：AUV obtains network deployment information, calculates neighbor node number in each node its communication radius；
2.2：It is arranged according to neighbor node quantity descending, selects the node with most neighbor nodes as cluster head, neighbours' section
Point is used as cluster member, while these vertex ticks are to have selected；
2.3：To be deleted from other nodes neighbors nodes labeled as the node selected, repeat before the step of until all nodes
It is labeled.
3. AUV mobile data collection methods in the underwater sensing network according to claim 1 based on data prediction, special
Sign is：New cluster is formed in the step (2) is specially：
3.1：Node perceived data, a node once send multiple data packets to come collect AUV, AUV by data according to
6:2:2 ratio is divided into training set, verification collection and test set；
3.2：According to SVR formula, suitable kernel function is selected to calculate corresponding w vector sums b vectors, according to y=wK (x)+b
Fit corresponding curve；Wherein, y indicates that predicted value, K (x) indicate that kernel function time series, w expression parameters, b indicate deviation,
B i.e. W=[w, b] included in the matrix of w；
3.3：AUV reconfigures the node with identical anticipation trend to form new cluster, to avoid the excessive caused section of scale
It is communicated between point, the problems such as energy consumption, for arrange parameter Hop=n to control cluster scale, wherein n indicates node hop count.
4. AUV mobile data collection methods in the underwater sensing network according to claim 1 based on data prediction, special
Sign is：Traversal whole network again in the step (3), specially：
4.1：AUV broadcasts initial position message, and leader cluster node is calculated at a distance from AUV after receiving this information, calculated i-th
Cluster and AUV initial distances：
Wherein (a, b, c) initial AUV position coordinates, (xi,yi,zi) it is i-th of cluster position coordinates；
Data volume total in cluster and the dump energy of leader cluster node are obtained simultaneously, these information deposit control message packet is carried out
Broadcast request；
4.2：AUV receives these control message packets, and the coefficient of competition of request cluster is calculated according to formula, calculates coefficient of competition：
Wherein num indicates that num cluster, M are network boundary length, and F (i) indicates i-th of cluster data flow, andIt represents entire
All cluster total data flow E of network0Node primary power, E (i) are then i-th of cluster dump energies；α and χ coefficients；Meet α ＞ χ
＞ (1- α-χ)；
Select the maximum cluster of coefficient of competition as path starting point, using shortest path first -- Dijkstra's algorithm progress time
It goes through, while prediction model is transmitted to corresponding cluster node, this path is as original path；
4.3：Second leg, AUV traverse all clusters for not receiving prediction model and carry out data collection again, form new path.
5. AUV mobile data collection methods in the underwater sensing network according to claim 1 based on data prediction, special
Sign is：The step (4) updates prediction model：
5.1：Maximum tolerance error threshold δ and adjustable delay sensitive threshold gamma are set；
5.2：Node calculates the square error and J of the time series data perceived, if J ∈ [0, δ), acquiescence need not be more
Newly, whereas if J ∈ [δ ,+∞), necessarily update；
5.3：Calculate whole section of time series data square error and while, calculate individual data predicted value square error, if
More than δ, abnormal data is inserted into abnormal data queue (EQ), if size (EQ) ＞ γ, prediction model is updated immediately and leads to
It crosses transmission solicited message and informs AUV.
6. AUV mobile data collection methods in the underwater sensing network according to claim 1 based on data prediction, special
Sign is：The step (5) is specially：
6.1：Cluster is asked to search for the positions AUV from both direction clockwise and anticlockwise along original path；
6.2：The request that AUV once receives a direction currently cruises pause, considers request cluster with planning path again, together
When current location be arranged an agent node, once another direction request message reach, will obtain agent node message,
To packet loss, do not continue to search, request terminates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810390515.5A CN108683468B (en) | 2018-04-27 | 2018-04-27 | AUV mobile data collection algorithm in underwater sensor network based on data prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810390515.5A CN108683468B (en) | 2018-04-27 | 2018-04-27 | AUV mobile data collection algorithm in underwater sensor network based on data prediction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108683468A true CN108683468A (en) | 2018-10-19 |
| CN108683468B CN108683468B (en) | 2020-09-22 |
Family
ID=63801687
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810390515.5A Active CN108683468B (en) | 2018-04-27 | 2018-04-27 | AUV mobile data collection algorithm in underwater sensor network based on data prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108683468B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109506767A (en) * | 2018-10-24 | 2019-03-22 | 西北工业大学 | A kind of real-time detection method causing sound field exception to underwater intrusion target |
| CN109982283A (en) * | 2019-02-15 | 2019-07-05 | 江苏商贸职业学院 | A kind of industrial cloud and mist framework communication system for transmitting energy consumption towards expectation |
| CN110392040A (en) * | 2019-06-12 | 2019-10-29 | 东南大学 | A kind of underwater mobile node re-authentication method based on trust chain |
| CN110602723A (en) * | 2019-08-27 | 2019-12-20 | 华侨大学 | Two-stage bidirectional prediction data acquisition method based on underwater edge equipment |
| CN111010704A (en) * | 2019-12-03 | 2020-04-14 | 沈阳化工大学 | Underwater wireless sensor network data prediction optimization method based on exponential smoothing |
| CN111542020A (en) * | 2020-05-06 | 2020-08-14 | 河海大学常州校区 | Multi-AUV cooperative data collection method based on region division in underwater acoustic sensor network |
| CN111600774A (en) * | 2020-05-13 | 2020-08-28 | 北京奇艺世纪科技有限公司 | Consumption delay determination method, system, device, equipment and readable storage medium |
| CN113825219A (en) * | 2021-11-10 | 2021-12-21 | 慕思健康睡眠股份有限公司 | Human body data collecting method and device |
| CN115243212A (en) * | 2022-07-20 | 2022-10-25 | 青岛科技大学 | Ocean data acquisition method based on AUV (autonomous Underwater vehicle) assistance and improved cross-layer clustering |
| CN115470936A (en) * | 2022-09-23 | 2022-12-13 | 广州爱浦路网络技术有限公司 | NWDAF-based machine learning model updating method and device |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5894450A (en) * | 1997-04-15 | 1999-04-13 | Massachusetts Institute Of Technology | Mobile underwater arrays |
| CN102833160A (en) * | 2012-08-17 | 2012-12-19 | 北京航空航天大学 | Contact predication based large-scale mobile delay tolerant network cluster-based routing method and system thereof |
| CN103686922A (en) * | 2013-12-18 | 2014-03-26 | 浙江树人大学 | Optimization method for survival time of multi-Sink-node movement wireless sensor network |
| CN106059920A (en) * | 2016-01-28 | 2016-10-26 | 中国电子科技集团公司第十研究所 | Routing method adapting to make-and-break connection data transmission of spatial network link |
| CN106209261A (en) * | 2016-07-21 | 2016-12-07 | 河海大学常州校区 | The mobile data collection method of three-dimensional UASNs based on probability neighborhood grid |
| CN106231636A (en) * | 2016-07-21 | 2016-12-14 | 河海大学常州校区 | The mobile data collection method of the three-dimensional UASNs of collection is covered based on probability neighborhood |
| CN107276684A (en) * | 2017-07-19 | 2017-10-20 | 河海大学常州校区 | Method of data capture based on AUV position predictions in underwater sensor network |
| CN107548029A (en) * | 2017-08-21 | 2018-01-05 | 河海大学常州校区 | AUV methods of data capture in a kind of underwater sensing network based on sea water stratification |
| CN107919918A (en) * | 2017-11-20 | 2018-04-17 | 中国人民解放军陆军工程大学 | The reliable acquisition method of Internet of Things data under a kind of mobile node auxiliary water |
-
2018
- 2018-04-27 CN CN201810390515.5A patent/CN108683468B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5894450A (en) * | 1997-04-15 | 1999-04-13 | Massachusetts Institute Of Technology | Mobile underwater arrays |
| CN102833160A (en) * | 2012-08-17 | 2012-12-19 | 北京航空航天大学 | Contact predication based large-scale mobile delay tolerant network cluster-based routing method and system thereof |
| CN103686922A (en) * | 2013-12-18 | 2014-03-26 | 浙江树人大学 | Optimization method for survival time of multi-Sink-node movement wireless sensor network |
| CN106059920A (en) * | 2016-01-28 | 2016-10-26 | 中国电子科技集团公司第十研究所 | Routing method adapting to make-and-break connection data transmission of spatial network link |
| CN106209261A (en) * | 2016-07-21 | 2016-12-07 | 河海大学常州校区 | The mobile data collection method of three-dimensional UASNs based on probability neighborhood grid |
| CN106231636A (en) * | 2016-07-21 | 2016-12-14 | 河海大学常州校区 | The mobile data collection method of the three-dimensional UASNs of collection is covered based on probability neighborhood |
| CN107276684A (en) * | 2017-07-19 | 2017-10-20 | 河海大学常州校区 | Method of data capture based on AUV position predictions in underwater sensor network |
| CN107548029A (en) * | 2017-08-21 | 2018-01-05 | 河海大学常州校区 | AUV methods of data capture in a kind of underwater sensing network based on sea water stratification |
| CN107919918A (en) * | 2017-11-20 | 2018-04-17 | 中国人民解放军陆军工程大学 | The reliable acquisition method of Internet of Things data under a kind of mobile node auxiliary water |
Non-Patent Citations (9)
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109506767A (en) * | 2018-10-24 | 2019-03-22 | 西北工业大学 | A kind of real-time detection method causing sound field exception to underwater intrusion target |
| CN109506767B (en) * | 2018-10-24 | 2020-12-08 | 西北工业大学 | Real-time detection method for sound field abnormity caused by underwater invasion target |
| CN109982283A (en) * | 2019-02-15 | 2019-07-05 | 江苏商贸职业学院 | A kind of industrial cloud and mist framework communication system for transmitting energy consumption towards expectation |
| CN110392040A (en) * | 2019-06-12 | 2019-10-29 | 东南大学 | A kind of underwater mobile node re-authentication method based on trust chain |
| CN110392040B (en) * | 2019-06-12 | 2021-09-07 | 东南大学 | Underwater mobile node re-authentication method based on trust chain |
| CN110602723B (en) * | 2019-08-27 | 2022-05-03 | 华侨大学 | Two-stage bidirectional prediction data acquisition method based on underwater edge equipment |
| CN110602723A (en) * | 2019-08-27 | 2019-12-20 | 华侨大学 | Two-stage bidirectional prediction data acquisition method based on underwater edge equipment |
| CN111010704A (en) * | 2019-12-03 | 2020-04-14 | 沈阳化工大学 | Underwater wireless sensor network data prediction optimization method based on exponential smoothing |
| CN111010704B (en) * | 2019-12-03 | 2023-06-02 | 沈阳化工大学 | Underwater wireless sensor network data prediction optimization method based on exponential smoothing |
| CN111542020A (en) * | 2020-05-06 | 2020-08-14 | 河海大学常州校区 | Multi-AUV cooperative data collection method based on region division in underwater acoustic sensor network |
| CN111600774A (en) * | 2020-05-13 | 2020-08-28 | 北京奇艺世纪科技有限公司 | Consumption delay determination method, system, device, equipment and readable storage medium |
| CN113825219A (en) * | 2021-11-10 | 2021-12-21 | 慕思健康睡眠股份有限公司 | Human body data collecting method and device |
| CN115243212A (en) * | 2022-07-20 | 2022-10-25 | 青岛科技大学 | Ocean data acquisition method based on AUV (autonomous Underwater vehicle) assistance and improved cross-layer clustering |
| CN115243212B (en) * | 2022-07-20 | 2023-08-08 | 青岛科技大学 | Ocean data acquisition method based on AUV assistance and improved cross-layer clustering |
| CN115470936A (en) * | 2022-09-23 | 2022-12-13 | 广州爱浦路网络技术有限公司 | NWDAF-based machine learning model updating method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108683468B (en) | 2020-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108683468A (en) | AUV mobile data collection algorithms in underwater sensing network based on data prediction | |
| Zhuo et al. | AUV-aided energy-efficient data collection in underwater acoustic sensor networks | |
| Arafat et al. | A survey on cluster-based routing protocols for unmanned aerial vehicle networks | |
| Rani et al. | Energy efficient chain based routing protocol for underwater wireless sensor networks | |
| Han et al. | An AUV location prediction-based data collection scheme for underwater wireless sensor networks | |
| Zhou et al. | E-CARP: An energy efficient routing protocol for UWSNs in the internet of underwater things | |
| Wang et al. | Mobility management algorithms and applications for mobile sensor networks | |
| Khan et al. | An energy-efficient data collection protocol with AUV path planning in the Internet of Underwater Things | |
| Khan et al. | AUV-aided energy-efficient clustering in the Internet of underwater things | |
| TW200408232A (en) | Intelligent communication node object beacon framework(ICBF) with temporal transition network protocol (TTNP) in a mobile AD hoc network | |
| Khedr et al. | Successors of PEGASIS protocol: A comprehensive survey | |
| CN111542020B (en) | Multi-AUV cooperative data collection method based on region division in underwater acoustic sensor network | |
| CN109275099B (en) | VOI-based multi-AUV (autonomous Underwater vehicle) efficient data collection method in underwater wireless sensor network | |
| CN103298055B (en) | Based on the greedy routing method of space lattice Region dividing in underwater sensor network | |
| CN107276684A (en) | Method of data capture based on AUV position predictions in underwater sensor network | |
| CN103249110B (en) | A kind of wireless sense network method for tracking target based on dynamic tree | |
| Senel et al. | Autonomous deployment of sensors for maximized coverage and guaranteed connectivity in underwater acoustic sensor networks | |
| CN111836327B (en) | Routing data transmission method for underwater sensor network and underwater sensor network | |
| CN113759971A (en) | Path planning method for unmanned aerial vehicle cooperative reconnaissance | |
| CN104539542A (en) | Low-energy-consumption routing tree pruning method based on mobile Sink data collection | |
| CN103297339A (en) | Spatial region division based routing method in underwater sensor network | |
| Rady et al. | Comprehensive survey of routing protocols for Mobile Wireless Sensor Networks | |
| CN107222900A (en) | A kind of wireless sensor network node collaboration method based on dynamic chain | |
| CN105228212A (en) | The underwater sensor network method for routing that a kind of many mobile sink node location are auxiliary | |
| Shah et al. | Water rippling shaped clustering strategy for efficient performance of software define wireless sensor networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |