CN107105466A - A kind of mobile Sink methods of data capture based on enhancing learning algorithm - Google Patents
A kind of mobile Sink methods of data capture based on enhancing learning algorithm Download PDFInfo
- Publication number
- CN107105466A CN107105466A CN201710149149.XA CN201710149149A CN107105466A CN 107105466 A CN107105466 A CN 107105466A CN 201710149149 A CN201710149149 A CN 201710149149A CN 107105466 A CN107105466 A CN 107105466A
- Authority
- CN
- China
- Prior art keywords
- data
- mobile sink
- learning algorithm
- value
- enhancing learning
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/023—Limited or focused flooding to selected areas of a network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
- H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
-
- 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
The invention discloses a kind of mobile Sink methods of data capture based on enhancing learning algorithm, mobile Sink paths planning methods proposed by the present invention help to lift Data Collection quality, overcome the problem of traditional wireless sense network data collection plan only focuses on the quantity for collecting data and ignores the quality of gathered data, by introducing cost function formula, the data of sensor collection are considered in terms of initial value and rate of decay two, can be on the premise of time delay be fixed, the quality of Data Collection is improved, the utilization rate of data is maximized.
Description
Technical field
The invention belongs to wireless sensor network data assembling sphere, more particularly to a kind of shifting based on enhancing learning algorithm
Dynamic Sink methods of data capture.
Background technology
Wireless sensor network is constituted by being deployed in substantial amounts of cheap microsensor node in monitored area, passes through nothing
The network system of one multi-hop ad hoc of line communication mode formation, the purpose is to collaboratively perceive, gather and handle network to cover
The information of object is perceived in cover area, and is sent to observer.
Under normal circumstances, the data that sensor is gathered are transferred into base station has two ways:Multi-hop transmission between sensor
With use mobile data collection device (referred to as mobile Sink).Multi-hop transmission is that the data of sensor node monitoring are sensed by other
It is transmitted, Monitoring Data may be by multiple node processings in transmitting procedure, the final road after multi-hop device node hop-by-hop
By being forwarded to base station.It is to arrange a mobile node by the way of mobile Sink, according to the path planned or in data
Dynamic selection access target, is moved near sensor node during collection, and by high-bandwidth communication mode, (such as light wave leads to
Letter) collecting sensor node data, be then back to base station.In terms of wireless sensor network data collection, both data
Collection mode cuts both ways.In sensor multi-hop transmission mode, if the data volume of collection is smaller, then met in bandwidth
In the case of the data that collect be transferred to the delay of base station can be very short, decision-maker can be facilitated to utilize data as soon as possible;But
It is that, if data volume is larger, simultaneously because the limitation of sensor self-energy and bandwidth, the efficiency of data transfer will be very low, very
Exhausted to the node energy that may result in some cases close to base station is too fast so that when reducing the survival of sensor network
Between.Mobile Sink collects data and increases Data Collection delay to a certain extent, but can avoid some sensor nodes
Energy expenditure excessively dead situation, so as to be conducive to extending on the whole the time-to-live of sensor network.In addition, using
The mobile Sink of single-hop, which collects data, can simplify the hardware design and software construction of sensor node, reduce the deployment of whole network
Cost.
In the wireless sense network of data is collected by mobile Sink, mobile Sink path planning is influence Data Collection
The key factor of performance.On the premise of network delay is given, how effectively layout data collecting path is so that sensing
It is the key for designing wireless sensor network data collection algorithm that device data, which can in real time, fully be transferred to base station,.The present invention
A kind of new Sink paths planning methods based on enhancing learning algorithm are proposed, being capable of effectively collecting sensor data.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of based on enhancing learning algorithm for the not enough of background technology
Mobile Sink methods of data capture.
The present invention uses following technical scheme to solve above-mentioned technical problem
A kind of mobile Sink methods of data capture based on enhancing learning algorithm, are specifically comprised the following steps:
Step 1:Monitored area is divided into by several square areas according to the communication radius of sensor node;
Step 2:Value of the movements the Evaluations matrix Q, wherein m that mobile Sink builds a m rows n row in internal memory are state
Number, n is behavior number;
Step 3:Mobile Sink sends probe messages, the data of requesting query adjacent domain to flood mode to adjacent domain
Value;
Step 4:In each adjacent domain, each sensor node selects cluster according to the descending sequence of self rest energy
Head node, by perception data multi-hop transmission to leader cluster node, each leader cluster node calculates data value according to value formula;
Step 5:Data value is returned to mobile Sink by each leader cluster node;
Step 6:Mobile Sink updates matrix after the data value of adjacent states is received according to enhancing learning algorithm formula
Q, then calculates the adjacent domain set that remaining time can be accessed, finally according to ε-greedy strategies, with ε probability selections
The maximum behavior of element value in the corresponding a line of current state in matrix Q, randomly chooses other (n-1) with 1- ε probability and plants behavior;
Step 7:When mobile Sink reaches a new region, the data that leader cluster node is perceived are collected;
Step 8:Calculate mobile Sink available remaining time, if it is possible to support next round is accessed then to go to step 3, otherwise
Mobile Sink returns to base station and reclaims collected data, and terminates epicycle data-gathering process.
As a kind of further preferred scheme of the mobile Sink methods of data capture based on enhancing learning algorithm of the present invention,
In step 1, the region of planning is that the length of side isSquare area, wherein, R be sensor node communication radius, and
Number consecutively 1,2,3 ..., N, wherein, N is number of regions, and N=m.
As a kind of further preferred scheme of the mobile Sink methods of data capture based on enhancing learning algorithm of the present invention,
In step 4, value formula is FVOI(t)=Ae-Bt, wherein A represents data initial value, and B represents to be worth rate of decay, t tables
Registration is according to the time begun to pass through from collection.
As a kind of further preferred scheme of the mobile Sink methods of data capture based on enhancing learning algorithm of the present invention,
In step 2, matrix Q is value of the movements Evaluations matrix.
As a kind of further preferred scheme of the mobile Sink methods of data capture based on enhancing learning algorithm of the present invention,
In step 6, enhancing learning algorithm formula is
Wherein, s represents current state, and a represents current behavior;α represents learning rate;(s a) is represented when movement R
Sink takes the reward immediately that action a can be obtained when being in state s;For current state s NextState,For current state s
Next behavior;γ is discount factor, for weighing reward R (s, a) with the income in memory immediately
The present invention uses above technical scheme compared with prior art, with following technique effect:
Mobile Sink paths planning methods proposed by the present invention help to lift Data Collection quality, overcome traditional nothing
Line Sensor Network data collection plan only focuses on the amount for collecting data, and ignores the quality problems of gathered data, by introducing
Cost function formula, sensor sensing data are considered in terms of initial value and rate of decay two, can be solid in time delay
On the premise of fixed, the quality of Data Collection is improved, the utilization rate of data is maximized.
Brief description of the drawings
Fig. 1 is inventive network zoning plan.
Fig. 2 is the implementation model figure of the present invention.
Fig. 3 is path planning algorithm flow chart of the present invention based on enhancing study.
Embodiment
Technical scheme is described in further detail with reference to Figure of description and specific embodiment.Tool
Body embodiment is described as follows:
Enhancing learning algorithm in it is stateful, act and award these three key elements.Intelligent body needs to be adopted according to current state
Action is taken, is obtained after awarding accordingly, then goes to improve these actions so that when next time arrives equal state again, intelligent physical efficiency is done
Go out more excellent action.Mobile Sink is analogized to intelligent body by the present invention, below to each key element using wireless sensor network as the back of the body
It is specifically described under conditions of scape:
State:Whole sensor network domains are divided into several identical square regions according to node communication radius
Comprising the sensor node that quantity is approximate in domain, each region, then each region is just equivalent to a state.
Action:When mobile Sink is in the data in the complete region of some state collection, it needs to determine next visit
The region or stop asked are in situ, and this acts the transformation for causing state.
Award:When mobile Sink has multiple actions to be contemplated that, the foundation which is selected act is exactly to award,
In sensor network background, the sensing data value of quantization is analogized to award by us.
State and action corresponding to intelligent body, enhancing learning algorithm have an action utility function, for evaluating in spy
Determine the quality that intelligent body under state takes some to act.Under normal circumstances, effectiveness can be used when intelligent body is in some state
Optimal action, but can so cause to move Sink be able to can not be explored to some by yoke among conventional experience always
Preferably selection.
Of the invention to determine to move Sink action using ε-greedy strategies, ε-greedy strategies are i.e. except according to optimal
Value of utility determines that next step is acted, and can also consider random action with certain probability, can so keep one to explore unknown shape
State and the balance before between experience.
Whenever mobile Sink reaches a certain state, after having collected data in the region, just updated according to enhancing study formula
Mobile Sink utility function.Then judge that can remaining time support subsequent region to access again, if can if continue iteration,
Otherwise terminate access and return to base station transmission data.Enhancing learns formula:
Wherein s and a represent current state and action respectively;(s a) represents the effect for taking action a to obtain under state s to Q
With evaluation;α represents learning rate, and learning rate is bigger, and the effect learnt before reservation is fewer;(s a) is represented when movement R
Sink takes the reward immediately that action a can be obtained when being in state s;WithNextState and behavior for current state s;γ is
Discount factor, for weighing reward R (s, a) with the interests in memory immediatelyIt represents mobile Sink memories
In next state action in value of utility maximum.It can be seen that discount factor is bigger, mobile Sink will more pay attention to
Toward experience, on the contrary mobile Sink only pay attention to immediate interest R (s, a).
Describe for convenience, it is assumed that having following application example, as shown in Figure 2.Base station is located at monitored area edge, just
Sink is moved during the beginning from base station, it is assumed that an acquisition time is set to T, mobile Sink does not have energy and data capacity limit,
Its constant translational speed is v, and identical with the sensor node and mobile Sink of communication range can calculate it by location technology
So as to calculate a certain moment, it returns to base station required time with the distance of terminal base station.
As shown in figure 3, a kind of process of the mobile Sink methods of data capture based on enhancing learning algorithm is as follows:
Step 1:In order to cover monitored area, the sensor network of deployment is divided into several length of sides isJust
Square region, the length of side isSquare area it is as shown in Figure 1;Wherein, R is sensor node communication radius, and successively
Numbering 1,2,3 ..., N, wherein, N is number of regions.
Step 2:Mobile Sink builds the matrix Q (m is status number, and n is behavior number) of a m rows n row in its internal memory, and
It is initialized as null matrix.Q is value of the movements Evaluations matrix, can for evaluating the quality for taking some to act in a particular state
With the knowledge base for being interpreted as moving Sink by it.
Step 3:Mobile Sink sends probe messages, the data valency of requesting query node to flood mode to adjacent domain
Value.
Step 4:In each adjacent domain, each sensor node sends dump energy inquiry request of data to adjacent node
Bag, then according to the descending sequence of dump energy, the maximum leader cluster node for being chosen as the region of dump energy is responsible for collection
All data in region;Meanwhile, each leader cluster node calculates value of the data at the T moment in the region according to value formula.
Wherein, value formula is FVOI(t)=Ae-Bt, A represents data initial value, and B represents to be worth rate of decay, and t represents data from receipts
Collect the time begun to pass through.
Step 5:Data value is returned to mobile Sink by each leader cluster node.
Step 6:Mobile Sink updates matrix after the data value of adjacent states is received according to enhancing learning algorithm formula
Q.Then this is calculated according to built-in timer and collects process remaining times, calculated using location technology and obtain mobile Sink
The distance between with terminal, with reference to speed v so as to calculate the adjacent domain set that can be accessed, then according to ε-greedy
Strategy, is conducted interviews with the region that ε probability selection Q values are maximum, with the other adjacent domains of 1- ε probability random access.
Step 7:Mobile Sink reaches new region center, collects the data that leader cluster node is perceived.
Step 8:Mobile Sink calculates the excess-col time according to built-in timer, if the remaining time can be supported down
One wheel adjacent domain Data Collection then goes to step 3, otherwise moves Sink and returns to the collected data of base station recovery, and terminates epicycle
Data-gathering process.
Claims (5)
1. a kind of mobile Sink methods of data capture based on enhancing learning algorithm, it is characterised in that:Specifically comprise the following steps:
Step 1:Monitored area is divided into by several square areas according to the communication radius of sensor node;
Step 2:Value of the movements the Evaluations matrix Q, wherein m that mobile Sink builds a m rows n row in internal memory are status number, and n is
Behavior number;
Step 3:Mobile Sink sends probe messages, the data valency of requesting query adjacent domain to flood mode to adjacent domain
Value;
Step 4:In each adjacent domain, each sensor node selects cluster head section according to the descending sequence of self rest energy
Point, by perception data multi-hop transmission to leader cluster node, each leader cluster node calculates data value according to value formula;
Step 5:Data value is returned to mobile Sink by each leader cluster node;
Step 6:Mobile Sink updates matrix Q, so after the data value of adjacent states is received according to enhancing learning algorithm formula
The adjacent domain set that remaining time can be accessed is calculated afterwards, finally according to ε-greedy strategies, with ε probability selection matrixes
The maximum behavior of element value in the corresponding a line of current state in Q, randomly chooses other (n-1) with 1- ε probability and plants behavior;
Step 7:When mobile Sink reaches a new region, the data that leader cluster node is perceived are collected;
Step 8:Calculate mobile Sink available remaining time, if it is possible to support next round is accessed then to go to step 3, otherwise move
Sink returns to base station and reclaims collected data, and terminates epicycle data-gathering process.
2. a kind of mobile Sink methods of data capture based on enhancing learning algorithm according to claim 1, its feature exists
In:
In step 1, the region of planning is that the length of side isSquare area, wherein, R is sensor node communication half
Footpath, and number consecutively 1,2,3 ..., N, wherein, N is number of regions, and N=m.
3. a kind of mobile Sink methods of data capture based on enhancing learning algorithm according to claim 1, its feature exists
In:In step 4, value formula is FVOI(t)=Ae-Bt, wherein A represents data initial value, and B represents to be worth rate of decay, t
Represent the time that data are begun to pass through from collection.
4. a kind of mobile Sink methods of data capture based on enhancing learning algorithm according to claim 1, its feature exists
In:In step 2, matrix Q is value of the movements Evaluations matrix.
5. a kind of mobile Sink methods of data capture based on enhancing learning algorithm according to claim 1, its feature exists
In:In step 6, enhancing learning algorithm formula is
Wherein, s represents current state, and a represents current behavior;α represents learning rate;(s a) is represented when mobile Sink is in shape R
The reward immediately for taking action a to obtain during state s;For current state s NextState,For current state s next behavior;
γ is discount factor, for weighing reward R (s, a) with the income in memory immediately
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710149149.XA CN107105466A (en) | 2017-03-14 | 2017-03-14 | A kind of mobile Sink methods of data capture based on enhancing learning algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710149149.XA CN107105466A (en) | 2017-03-14 | 2017-03-14 | A kind of mobile Sink methods of data capture based on enhancing learning algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107105466A true CN107105466A (en) | 2017-08-29 |
Family
ID=59675583
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710149149.XA Pending CN107105466A (en) | 2017-03-14 | 2017-03-14 | A kind of mobile Sink methods of data capture based on enhancing learning algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107105466A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109275099A (en) * | 2018-09-26 | 2019-01-25 | 河海大学常州校区 | More AUV efficient data collection methods in underwater wireless sensor network based on VOI |
CN109445437A (en) * | 2018-11-30 | 2019-03-08 | 电子科技大学 | A kind of paths planning method of unmanned electric vehicle |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100149983A1 (en) * | 2008-12-16 | 2010-06-17 | At&T Intellectual Property I, Lp | Broadcast latency optimization in multihop wireless networks |
CN101835237A (en) * | 2010-05-14 | 2010-09-15 | 南京邮电大学 | Data aggregation method in wireless sensor network |
CN104270796A (en) * | 2014-10-23 | 2015-01-07 | 湘潭大学 | Data collection method for selecting Sink routes based on markov model |
CN106358203A (en) * | 2016-08-30 | 2017-01-25 | 湖南大学 | Method for spectrum allocation in distributed cognition wireless sensor network on basis of Q study |
-
2017
- 2017-03-14 CN CN201710149149.XA patent/CN107105466A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100149983A1 (en) * | 2008-12-16 | 2010-06-17 | At&T Intellectual Property I, Lp | Broadcast latency optimization in multihop wireless networks |
CN101835237A (en) * | 2010-05-14 | 2010-09-15 | 南京邮电大学 | Data aggregation method in wireless sensor network |
CN104270796A (en) * | 2014-10-23 | 2015-01-07 | 湘潭大学 | Data collection method for selecting Sink routes based on markov model |
CN106358203A (en) * | 2016-08-30 | 2017-01-25 | 湖南大学 | Method for spectrum allocation in distributed cognition wireless sensor network on basis of Q study |
Non-Patent Citations (1)
Title |
---|
周淑俐;章韵;陈志;扈罗全;岳文静: "基于Q学习的多Sink节点无线传感网路由机制研究", 《传感技术学报》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109275099A (en) * | 2018-09-26 | 2019-01-25 | 河海大学常州校区 | More AUV efficient data collection methods in underwater wireless sensor network based on VOI |
CN109275099B (en) * | 2018-09-26 | 2020-06-09 | 河海大学常州校区 | VOI-based multi-AUV (autonomous Underwater vehicle) efficient data collection method in underwater wireless sensor network |
CN109445437A (en) * | 2018-11-30 | 2019-03-08 | 电子科技大学 | A kind of paths planning method of unmanned electric vehicle |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Nakas et al. | Energy efficient routing in wireless sensor networks: A comprehensive survey | |
Angin et al. | Agrilora: a digital twin framework for smart agriculture. | |
Cambra et al. | An IoT service-oriented system for agriculture monitoring | |
Chatzigiannakis et al. | Sink mobility protocols for data collection in wireless sensor networks | |
CN103702387B (en) | Social network-based vehicle-mounted self-organization network routing method | |
CN104010336B (en) | A kind of underwater wireless sensor network routing method of two-stage heterogeneous clustering | |
Khelifa et al. | Smart irrigation using internet of things | |
CN102118450B (en) | Betweenness centrality-based opportunistic network P2P (peer-to-peer) information query method | |
CN108924897A (en) | A kind of mobile sink paths planning method based on deeply learning algorithm | |
Geng et al. | An Agricultural Monitoring System Based on Wireless Sensor and Depth Learning Algorithm. | |
CN102438291B (en) | Data aggregation method for increasing capacity of wireless sensor network | |
CN107205254B (en) | Mobile sink node access and interference rejection method and system based on tight delay constraint | |
CN107295453A (en) | A kind of wireless sensor network data fusion method | |
CN109787699A (en) | A kind of wireless sensor network routing link trend prediction method based on interacting depth model | |
CN107105466A (en) | A kind of mobile Sink methods of data capture based on enhancing learning algorithm | |
Lampin et al. | Efficient route redundancy in dag-based wireless sensor networks | |
Bucur et al. | Optimizing groups of colluding strong attackers in mobile urban communication networks with evolutionary algorithms | |
Bharany et al. | Energy Efficient Clustering Protocol for FANETS Using Moth Flame Optimization. Sustainability 2022, 14, 6159 | |
CN106290772A (en) | A kind of sewage monitoring system | |
Dong et al. | HARVEST: A task-objective efficient data collection scheme in wireless sensor and actor networks | |
Ying | Intelligent rural tourism environmental suitability evaluation system based on a wireless sensor network | |
Yadav et al. | A novel algorithm for wireless sensor network routing protocols based on reinforcement learning | |
Dong et al. | Performance Evaluation of the Transmission Quality of the Internet of Things in Farmland Environmental Monitoring Based on ns-3 | |
Ajofoyinbo et al. | Directional antennas MAC protocol for energy‐efficient wireless sensor networks with mobile sink | |
Jiang | Application of hybrid ant colony algorithm in wireless sensor network coverage |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170829 |
|
RJ01 | Rejection of invention patent application after publication |