CN110167054A - A kind of QoS CR- LDP method towards the optimization of edge calculations node energy - Google Patents
A kind of QoS CR- LDP method towards the optimization of edge calculations node energy Download PDFInfo
- Publication number
- CN110167054A CN110167054A CN201910418848.9A CN201910418848A CN110167054A CN 110167054 A CN110167054 A CN 110167054A CN 201910418848 A CN201910418848 A CN 201910418848A CN 110167054 A CN110167054 A CN 110167054A
- Authority
- CN
- China
- Prior art keywords
- node
- algorithm
- activ
- value
- automatic machine
- 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.)
- Withdrawn
Links
Classifications
-
- 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
- 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
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/12—Communication route or path selection, e.g. power-based or shortest path routing based on transmission quality or channel quality
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
- H04W52/0203—Power saving arrangements in the radio access network or backbone network of wireless communication 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
A kind of QoS CR- LDP method towards the optimization of edge calculations node energy, belong to internet of things field, the present invention considers three end-to-end delay, transmission reliability and energy consumption QoS constraint conditions, innovatively with the relevant technologies in edge calculations and machine learning, sensor network is configured to a multi-constrained optimal path model, proposes the method for the present invention.The algorithm accelerates algorithmic statement by using the mode of automatic machine and environmental interaction, and the modes such as suspend mode state of activation of control node consume to reduce network energy.By experiment, MQEN algorithm and QMCRA-DP, CMQRA, LQOR algorithm are compared, compare other three kinds of algorithms, MQEN algorithm can satisfy requirement of the multi-QoS constraint to end-to-end delay and reliability services while significantly reducing network energy consumption.
Description
Technical field
The invention belongs to internet of things field, and in particular to a kind of QoS constraint road towards the optimization of edge calculations node energy
By method.
Background technique
Mobile edge calculations technology (Mobile Edge Computing, MEC) be by by computational load from core cloud number
Mitigate network burden to the mobile edge side close to user terminal according to central transference.MEC server can be in access point (such as base
Stand, router etc.) cloud service is provided for mobile subscriber in coverage area, traditional cloud computing has been zoomed in the net of user side by it
Network edge responds rapidly to realize in marginal end so that access content, services and applications are accelerated.MEC server exists
For user provide highly distributed calculate environment while, also accelerate the distribution of the network information, by provide auxiliary calculate and
Storage resource is come the response speed of the service of improving and program, to reduce time delay, saves Internet resources.
With being constantly progressive for society, scientific and technical fast development, centralized cloud computing model cannot be very before
, then there is a kind of novel computation model, edge calculations in the good current problem encountered of solution.Edge calculations technology
(Edge Computing, EC), which refers to, handles data at the edge of network, can reduce request response time in this way, promote electricity
Pond cruising ability, and while reducing network bandwidth guarantee data safety and privacy.
With the arrival in 5G epoch, many relevant multimedia application are had already appeared on internet, such as Video chat, net
Network live streaming, Internet phone-calling, long-distance education etc..Internet gradually started from single data transmission to data, voice,
The comprehensive transmission of the multimedia messages such as image, video develops.However it transmits these applications and requires different QoS (Quality
Of Service) it requires, so multi-QoS constraint routing becomes a kind of effective way to solve the problems, such as these QoS.
At present on network in terms of the research of QoS routing is broadly divided into two, one is multipath QoS Routing Protocols
Research, another kind is the research of the single-hop path QoS Routing Protocol about node control.Multipath can guarantee information transmission
Reliability and it is end-to-end between delay, but more expend energy.QoS routing as a kind of Novel road by research algorithm, it
Compared with traditional routing mechanism, it can significantly improve the handling capacity of network and the degeneration of network performance, optimize allocation of resources, put down
Weigh network load, the optimization of network global resource utilization rate is realized, so that maximization network receives the energy of other qos parameter demands
Power.Therefore the transmission service of (best-effort-service) of doing one's best only is provided for traditional routing, portfolio is as early as possible
Transmission, without specific time and guaranteed reliability, and existing besteffort service is also unable to satisfy various applications to net
The problem of difference of network transmission quality requires situation, by considering three end-to-end reliability, time delay and energy consumption indexs, wound
The mechanism such as convergence and node energy perception wake-up property are newly accelerated using the pretreatment of node initial data, prune rule to reduce net
The overall power consumption of network proposes a kind of QoS CR- LDP method MQEN towards the optimization of edge calculations node energy
(Multi-QoS constrained routing algorithm for Edge computing and Node energy
optimization).By Experimental comparison, MQEN algorithm is guaranteeing that end arrives compared to LQOR, CMQRA and QMCRA_DP algorithm
Under the premise of holding reliability and delay, the energy consumption of network can be reduced significantly, extend network lifecycle.
Summary of the invention
The purpose of the present invention is finding the Routing Protocol for meeting multi-QoS constraint, only provides and do the best for traditional routing
And the transmission for being is serviced without specific time and guaranteed reliability, and the transmission service done one's best also is unable to satisfy respectively
Kind of application the problems such as requiring different situations to network transmission quality, proposes a kind of towards the optimization of edge calculations node energy
QoS CR- LDP method MQEN.The algorithm is by considering three reliability, time delay and energy consumption indexs end to end, innovation
The mechanism such as algorithmic statement and node energy perception wake-up are accelerated to type using the pretreatment of node initial data, prune rule to reduce
The overall power consumption of network.Learning automaton is deployed in each node, each node by present invention application edge calculations technology
Being distributed for task optimizes configuration according to objective environment by corresponding learning automaton, promotes each node to the maximum extent
Operational capability and storage capacity.In addition, the collected initial data of sensor node is carried out preprocessing processing, so that reducing has
The transmission time and transmission energy consumption for imitating data, guarantee communication quality, extend network lifecycle, improve data transfer efficiency,
With certain practical value.
QoS CR- LDP method provided by the invention towards the optimization of edge calculations node energy, main includes such as ShiShimonoseki
Key step:
1st, initial phase；It specifically includes that
1.1st, edge calculations network model is established；
1.2nd, packet reception rate PRR, end-to-end time delay DT (l), node are defined and sends data consumption energy in T time
Enode；
1.3rd, the maximum number of iterations T that initialization algorithm terminatesk, the node set that is activated during algorithms selection
Minimum node set MIN_ACTIV_SET (initial value is the node in whole network) in ACTIV_SET, ACTIV_SET set,
The threshold value threshold of the product of ACTIV_SET set interior joint probability is judged, for recording the counter m of algorithm iteration number；
2nd, behavior aggregate formation stages；First node being activated sends initial message to other nodes in network,
To which each node forms the behavior aggregate of oneself；It specifically includes:
2.1st, source node broadcasts one article of carrying address and synchronous letter to the single-hop neighbor node of oneself in communication range
The data packet of breath after neighbor node receives data packet, returns to the data packet comprising oneself address h and synchronizing information, then
Source node sends an ACK message packet to the neighbor node again after receiving data packet, indicates to have built up connection between them, can
Mutually to transmit information.Source node will have sent the node receiveed the response and make after the response for receiving multiple single-hop neighbor nodes
For the optional behavior aggregate of oneself next hop information transmission；
2.2nd, after the automatic machine of source node is activated, a behavior will be randomly choosed from the behavior aggregate of oneself,
It sends information to activate the automatic machine of selected behavior corresponding node, selected node forms the optional of oneself in the same way
Behavior aggregate, and then the automatic machine by activating randomly chooses the transmission that a movement carries out next hop information.Repeat, until certain
Until aggregation node is accessed in the selected movement of the automatic machine of one node；
2.3rd, in edge calculations network, as soon as the node that is often activated, is written set ACTIV_SET for the node
In, the path slave source node to aggregation node elected after the completion of algorithm each round iteration is had recorded in ACTIV_SET；
3rd, the rewards and punishments stage；Judge whether algorithm meets termination condition: if the product of set ACTIV_SET interior joint probability
Value greater than threshold or m is greater than Tk, then algorithm terminates.It, be according to specific rule if being unsatisfactory for termination condition
Then automatic machine is punished or rewarded, the process of automatic machine and environmental interaction is as shown in Fig. 1；It specifically includes:
The 3.1st, if set ACTIVE_SET interior joint quantity and the value of DT (l) are both less than the value of last round of iteration, and
And PDR is greater than or equal to the value of last round of iteration, then the value of β (n) is that 1, ENV just rewards automatic machine, and update probability is dynamic
Make matrix P (n)；
3.2nd, on the contrary, if set ACTIV_SET interior joint quantity and the value of DT (l) are both greater than last round of iteration
Value, and PDR is less than the value of last round of iteration, then the value of β (n) is that 0, ENV just punishes automatic machine, and update probability acts square
Battle array P (n)；
3.3rd, ENV provides an enhancing signal to the automatic machine being activated in last round of, and by the value of m from increasing 1, according to
Step 1 restarts the selection of next round node；
4th, the self-training stage.Behavior aggregate is carried out from trimming, it is specified that the node elected using prune rule
It cannot be selected again by the automatic machine of other nodes.Allow behavior aggregate carry out from build can solve automatic machine randomly select it is dynamic
When making, the closed loop phenomenon that movement is mutually chosen may be led to the problem of among the nodes.
5th, the stage is transmitted.The node elected according to MQEN carries out the transmission of data, if found in transmission PDR <
100%, then illustrate there is the phenomenon that packet loss in transmission process, needs to retransmit data at this time.
The advantages and positive effects of the present invention:
A kind of QoS CR- LDP method towards the optimization of edge calculations node energy of major design of the present invention, for wireless
Conventional routing protocols in sensor network only provide the transmission service done one's best and but protect without specific time and reliability
Barrier, although and multipath QoS Routing Protocol can guarantee information transmission reliability and it is end-to-end between delay, more
The problem of expending energy, MQEN algorithm considers three reliability, time delay and energy consumption indexs end to end, innovatively adopts
Accelerate algorithmic statement and node energy perception wake-up mechanism with the pretreatment of node initial data, prune rule to reduce network
Overall power consumption.It is compared with existing LOQR, CMQRA with QMCRA_DP agreement, MQEN algorithm is guaranteeing end-to-end reliability
Under the premise of delay, the energy consumption of network can be reduced significantly, extends network lifecycle, and there is certain practical valence
Value.
Detailed description of the invention
Fig. 1 is automatic machine and environmental interaction schematic diagram；
Fig. 2 is edge calculations network model figure；
Fig. 3 is edge calculations network node distributed model figure；
Fig. 4 is edge calculations network simulation scene figure；
Fig. 5 is algorithm iteration number and data packet delivery fraction relationship comparison diagram；
Fig. 6 is algorithm iteration number and end-to-end delay relationship comparison diagram；
Fig. 7 is algorithm iteration number and surviving node number relationship comparison diagram；
Fig. 8 is algorithm iteration number and node average energy consumption relationship comparison diagram；
Fig. 9 is algorithm iteration number and network total energy consumption relationship comparison diagram；
Figure 10 is sensor data acquisition analogous diagram；
Figure 11 is analog sensor communication port schematic diagram；
Figure 12 is wisdom agricultural system data analysis chart.
Specific embodiment
The method of the present embodiment design is to carry out emulation experiment to the method for the present invention by MATLAB2014a developing instrument.
This method and LQOR, CMQRA and QMCRA_DP method are compared and analyzed.Under identical test environment and test parameter,
Analysis compares transmission reliability, end-to-end time delay, network average energy consumption and the network gross energy of these four different methods
Parameter is consumed, wherein indicating these four algorithms with different labels in the accompanying drawings.Referring to attached drawing 1, specific implementation process is described in detail such as
Under:
Step 1, system model are established:
Step 1.1 establishes edge calculations network model
It is applied with wisdom farm as background, edge calculations network model, such as attached drawing is constructed in conjunction with QoS route restriction condition
Shown in 2.Define edge calculations network: the node in edge calculations network is initialized, network by N number of arbitrary arrangement section
Point constitutes V={ v0,v1,……,vn, wherein v0It is the sink node in network.Each node has a sensing range RSWith
One communication range RC, primary power E0, and each node has certain computing capability NCWith storage sensing capability
NS.Sensor network is configured to a multi-constrained optimal path model, the topology diagram of network is G=(V, E), wherein V
It is the vertex set in figure G, E is the set on side in figure G.A weight w is set in each edge (u, v) ((u, v) ∈ E) of G
(u, v) sets the QoS constrained parameters in each edge as qk(u, v), k=1,2,3, then the QoS routing of edge calculations network is asked
Topic translates under k QoS constraint condition, finds a shortest path l from source node to sink node, so that it is full
Foot:
qk(l)=∑(u,v)∈lqk(u,v) qk≤ c (u, v), k=1,2,3
W (l)=∑(u,v)∈lw(u,v) (1)
Wherein w (u, v) is the smallest weight in all feasible paths for meet above-mentioned condition.
Step 1.2 defines edge calculations network parameter
The processing environment ENV: the processing environment towards edge calculations learning automaton for defining learning automaton is one three
Tuple < Ea,Eb, C >:
Ea={ α1,α2,…,αrBe environment input behavior set；
Eb={ β1,β2,…,βmBe environment output behavior set；
C={ cij=Pr { β (n)=βj| α (n)=αi}} 1≤i≤r,1≤j≤m
If value be not the variation with the time and change that we just claim the environmental variance to be stable；Otherwise claim
The environmental variance is unstable.
Define packet reception rate PRR: the data packet sum and sending node that receiving node receives within certain time are sent
The ratio of data packet sum, calculation formula are as follows:
Wherein, PRec_PacketsIndicate the data packet sum that receiving node receives in a period of time, PS_PacketsIndicate this section
The data packet sum that sending node is sent in time.
Define data packet delivery fraction PDR: indicating the PRR between any single-hop node with R (u, v), then from source node to
The data packet delivery fraction of the path l of sink node are as follows:
MQEN algorithm calculates the reliability of end-to-end transmission using PDR.
Define end-to-end time delay: in the application environment of wireless sensor network, what data packet was sent between end-to-end
Delay mainly includes transmission delay and propagation delay, and propagation delay is (usual compared to can be ignored for transmission delay
It is nanometer unit grade).Therefore main to consider transmission delay, calculation formula is as follows, and wherein dt (u, v) indicates the biography of node-to-node
Defeated delay:
Definition node sends data capacity consumption: in wireless sensor network, the energy consumption of node is mainly derived from
Data transmission on wireless communication module, so each sensor node disappears within the time as k-bit information transmitting range d
The energy of consumption are as follows:
Enode(vi)=T (Eeleck+εampkdn) d < Rc (5)
Wherein, EelecIt indicates emitter on sensor and receives the energy consumption that circuit sends or receives unit；εampIndicate hair
Penetrate the energy that amplifier is consumed every unit of transfer square metre；D indicates the node single-hop node adjacent thereto where sending device
Receive the distance between circuit；N is propagation attenuation index, usual 2 < n < 5.
Step 2, initial phase:
Parameter used in algorithm is initialized.The parameter initialized includes: the maximum that algorithm terminates
The number of iterations Tk；The node set ACTIV_SET being activated during algorithms selection；In dynamically recording ACTIV_SET set most
The MIN_ACTIV_SET set of trifle points, the set are initialized as the node in whole network；Algorithm termination condition threshold value
Threshold, when the product of set ACTIV_SET interior joint probability is greater than the threshold value, algorithm is terminated；Counter m, for recording
The number of algorithm iteration.
After algorithm starts execution, a carrying T can be sent to whole networkk, the message of m and threshold initial value, use
All nodes in notice network, algorithm bring into operation.
Step 3, behavior aggregate formation stages:
In edge calculations network, source node (for sink node) is first node being activated, information
It is transmitted since source node, sink node refers to the aggregation node of information transmission, connects the gateway of outer net.
Step 3.1, source node construct behavior aggregate
Source node is in communication range RcOneself introversive single-hop neighbor node broadcasts one and carries address and synchronizing information
Data packet after its neighbor node receives data packet, can equally return to the data packet comprising oneself address and synchronizing information,
Source node can send an ACK packet to neighbor node again after receiving data packet, notify neighbor node, and both sides have built up connection
Information can be transmitted mutually each other in system.Source node will have sent after receiving the receiveing the response of multiple single-hop neighbor nodes
The optional behavior aggregate that the node receiveed the response is transmitted as oneself next hop information.
Step 3.2, next-hop node construct behavior aggregate
After the automatic machine of source node is activated, a behavior will be randomly choosed from the behavior aggregate of oneself, send letter
Breath activates the automatic machine of selected behavior corresponding node, and then this node forms the movement of oneself in the same way again
Collection, and then the automatic machine by activating randomly chooses the transmission that a movement carries out next hop information.Repeat this operation, directly
Until sink node is accessed in movement selected by the automatic machine of a certain node.
Step 3.3, the path for generating information transmission
In this process, as soon as whenever there is a node to be activated, which is written in set ACTIV_SET, ACTIV_
Had recorded in SET after the completion of algorithm each round iteration it is selected slave source node to the path of sink node.
As shown in Fig. 3, it is assumed that V is source node, and J is sink node.V node is added in set ACTIV_SET and logical
The interior data packet that carrying an address and synchronizing information are broadcasted to its single-hop neighbor node (B, D, E, F) of letter range Rc, B, D,
E, after F receives data packet, can equally return to V mono- include oneself address and synchronizing information data packet, V receive B, D, E,
An ACK packet can be returned after the data packet of F, had built up and contacted for informing B, D, E, F, between us, it can be mutually
Mutually transmission information.The movement defecate collection of V node forms Ev={ α1,α2,α3,α4}.V concentrates one movement of random selection from movement
(such as α3), α3Corresponding node is E.Set ACTIV_SET is added in E node, and then the learning automaton on E is activated, E
Node equally also carries out behavior aggregate and establishes Ee={ α1,α2(algorithm has been carried out from trimming, institute due to using prune rule
With there is no the movements for being directed toward V node in the behavior aggregate of E), the automatic machine of E randomly chooses a movement (such as α2), α2It is corresponding
Node is H, and set ACTIV_SET is added in H node, and the automatic machine of H node is activated at this time, and H equally also carries out building for behavior aggregate
Vertical Eh={ α1,α2}.H concentrates one movement (such as α of random selection from movement1), α1Corresponding node is J.Set is added in J node
ACTIV_SET, at this point, first round circulation terminates.The path chosen are as follows: V → E → H → J.
Step 4, rewards and punishments stage
Step 4.1, the size for judging set ACTIV_SET and set MIN_ACTIV_SETd:
The size for first judging set ACTIV_SET and set MIN_ACTIV_SETd, if set ACTIV_SET interior joint
Quantity is less than MIN_ACTIV_SET, then the node in set ACTIV_SET is just assigned to set MIN_ACTIV_SET.
Step 4.2 judges whether the termination condition of algorithm meets:
The termination condition of algorithm: if the product of set ACTIV_SET interior joint probability is big greater than threshold m value
In Tk, then algorithm terminates.At this point, the node in set MIN_ACTIV_SET is the optimal path l that MQEN is selected.Otherwise, it holds
Row step 4.3.
Step 4.3, environment are rewarded or are punished to automatic machine:
If set ACTIV_SET interior joint quantity and the value of DT (l) are both less than the value of last round of iteration, and PDR is big
In or equal to last round of iteration value, then at this time the value of β (n) be 1, ENV just automatic machine is rewarded, according to formula (5)
Probability matrix P (n) is acted to it to be updated.
Wherein, m and n0 is respectively perturbation parameter and analytic parameter.
On the contrary, being acted generally according to formula (6) to it if the value of β (n) at this time is that 0, ENV just punishes automatic machine
Rate matrix is updated.
kj(n+1)=kj(n),1≤j≤r (6)
At this time, it is also necessary to indicate automatic machine at the n moment according to formula (7) (8) Lai Gengxin estimator D (n) and U (n), D (n)
Certainty (reward) estimator vector, U (n) indicate automatic machine the n moment random estimator vector.
WhereinIt is to obey equally distributed stochastic variable in section.
ENV provides an enhancing signal to the automatic machine being activated in last round of, and by the value of m from increasing 1, according to step 1
Restart the election of next round node.
Step 5, self-training stage
In the behavior aggregate forming process of step 2, automatic machine is when selection acts, may be due to its randomness
Generation acts the closed loop phenomenon mutually chosen between node.So it is carried out using prune rule from trim, it is specified that by
The node chosen cannot be chosen again by the automatic machine of other nodes.
In addition, the node elected can be constantly in active state when carrying out information transmission, compared to other suspend mode sections
Point more expends energy, so introducing a kind of waking up nodes mechanism of Energy-aware: when the energy of live-vertex is reduced to 50%
E0(E0For the primary power of node) when, then take strategy that it is made to be converted to sleep state, then from other remaining nodes again into
The selection of row optimal path.
Step 6, data transfer phase
Illustrate to pass if finding PDR < 100% in transmission process according to the transmission that the node that MQEN is selected carries out data
There is packet loss phenomenon during defeated, needs to retransmit data at this time.
MQEN pseudo-code of the algorithm:
Algorithm is analyzed in terms of time complexity and space complexity two:
In the worst case, because there is N-2 node between source node and sink node, and information transmission will pass through institute
Some intermediate nodes could communicate, so the worst time complexity of algorithm is O (n).It is selected in learning automaton optimal
After node, number of nodes is less than total nodal point number amount N, if the optimal number of nodes selected is aN (0 < a < 1), at this time source node
There is aN-2 node between sink node, the time complexity of algorithm is aO (n) (0 < a < 1).
For the random distribution situation of sensor node, the information between node and node be suitble to the adjacency matrix of figure come
Indicate the connection between node.Each node will store the induction range and routing induction shape of neighbor node in this algorithm
State, therefore the space complexity of algorithm is O (n2).
Emulation and the experiment of specific example application scenarios are tested.
In order to prove that proposed MQEN algorithm has preferably performance, using MATLAB emulation tool to MQEN algorithm into
Row analog simulation compares experimental result and three typical multi-QoS constraint agreements, including LQOR agreement, CMQRA agreement
With QMCRA_DP agreement.In order to guarantee the accuracy and integrality of experiment, emulation experiment is arranged in the same network topology knot
It is carried out under structure, and uses identical parameter setting.
The setting of edge calculations network parameter: network size is the region of 200m*200m, as shown in Fig. 4, all sensings
In the entire network, each node has a sensing range R to device node random placementSWith a communication range RC, primary power
For E0, each node has certain computing capability NCWith storage capacity NS, all nodes in network are all random schedule fortune
Capable, and dynamic adaptive updates maintenance is carried out during network work.There is a base station outside network scenarios, covering is entire
Network, for carrying out the processing of information, sending and receiving.Design parameter is shown in Table 1:
1 simulation parameter of table
The experimental results of this example are as follows:
1. MQEN agreement proposed by the present invention is during carrying out data transmission it can be seen from attached drawing 5, with four kinds
After carrying out more wheel iteration, data transmission credibility all increases algorithm, this is because each algorithm changes having carried out more wheels
Dai Hou, every kind of algorithm all have selected the optimal transmission paths for being suitble to respective agreement.In MQEN algorithm, sensor node is collected
After data, data can be pre-processed, reduce the transmission of invalid data；Secondly, after the iteration that algorithm has carried out more wheels,
It has elected a small amount of and efficient node to carry out data transmission, has improved the reliability of network traffic.So MQEN algorithm
Transmission reliability be better than other three kinds of algorithms.
2. end-to-end delay is also constantly reducing, each with the raising of algorithm iteration number it can be seen from attached drawing 6
Algorithm is all transmitting collected data packet as soon as possible.MQEN algorithm is pre- due to having carried out to the collected initial data of sensor
Processing, the quantity for transmitting data packet become smaller, therefore delay is less than other algorithms in the initial stage.Automatic machine in MQEN algorithm
After receiving the enhancing signal of environment, searching of just doing the best always postpones the smallest node to transmit data, so MQEN is calculated
For method with the increase of the number of iterations, it is little that variation is estimated in delay.MQEN algorithm is compared to other three kinds of algorithms, hence it is evident that reduces letter
Cease propagation delay time.
3. QMCRA_DP algorithm begins to node energy occur when iteration is less than 1000 times it can be seen from attached drawing 7
It measures totally, this is because the business that QMCRA_DP algorithm is preferably high priority provides preferably service, so node energy
Consume larger, network lifecycle is shorter.And MQEN algorithm is to use node as few as possible to carry out data transmission, and pass
Defeated data volume is relatively fewer, while using Energy-aware mechanism keeps the energy consumption of each node in network more average,
So MQEN algorithm extends the operating time of node, network lifecycle is improved, the remaining survival number of node is more.
4. it can be seen from attached drawing 8 the node average energy consumption of MQEN agreement compared to LQOR agreement, CMQRA agreement and
QMCRA_DP agreement is minimum, because MQEN agreement is avoided using the waking up nodes mechanism and prune rule of Energy-aware
The energy consumption of extra node, while accelerating the convergence of algorithm.
5. when iterating to about 400 wheel, the total power consumption of network aobvious MQEN agreement occurs it can be seen from attached drawing 9
The reduction of work, this is because automatic machine acts in the early stage consumes more energy on invalid action in selection course, and
Later period since the probability that invalid action is selected reduces, avoids the consumption of excess energy, to reduce network total energy consumption, prolongs
Network lifecycle is grown.
6. attached drawing 10 is the image emulated using proteus software to sensor node in network, wherein simulating
The process of Temperature Humidity Sensor, optical sensor and wind transducer acquisition data, and adopted using C Plus Plus to write data
The code of collection, while collected initial data is pre-processed, the transmission of invalid data is reduced, efficiency of transmission is improved.
7. attached drawing 11 is the port communicated for the sensor node of emulation being arranged at the end PC, so that sensor section
Point simulation model can be communicated with wisdom agricultural monitoring system server-side, send system service for collected data
End, so that system analyzes data, so that the equipment controlled in farm responds.
8. attached drawing 12 is the end Web of wisdom agricultural monitoring system, the collected data of sensor are analyzed at the end Web
Later, processing result is graphically shown, system manager is facilitated to understand running situation.
Claims (4)
1. a kind of QoS CR- LDP method towards the optimization of edge calculations node energy, it is characterised in that this method mainly includes
Following steps:
1st, initial phase, the maximum number of iterations T terminated including initialization algorithmk, the section that is activated during algorithms selection
Point set ACTIV_SET, records the MIN_ACTIVE_SET of minimum node number in ACTIV_SET set, and algorithm terminates threshold value
Threshold, for recording the counter m of algorithm iteration number；
2nd, behavior aggregate formation stages, source node broadcast a carrying address to the single-hop neighbor node of oneself in communication range
With the data packet of synchronizing information, after neighbor node receives data packet, it equally can also return to one and include oneself address and synchronous letter
The data packet of breath, source node will send the node receiveed the response as oneself after the response for receiving multiple single-hop neighbor nodes
The optional behavior aggregate of next hop information transmission；
3rd, the rewards and punishments stage, if algorithm is unsatisfactory for termination condition: the product of set ACTIV_SET interior joint probability is greater than threshold value
Threshold m value is greater than Tk, then will automatic machine be punished or be rewarded；If saved in set ACTIV_SET
Point quantity and the value of DT (l) are both less than the value of last round of iteration, and PDR is greater than or equal to the value of last round of iteration, then this
When β (n) value be 1, ENV just automatic machine is rewarded；If the value of β (n) is that 0, ENV just punishes automatic machine；
4th, self-training stage, automatic machine, due to its randomness, may act in generation among the nodes when selection is acted
The closed loop phenomenon mutually chosen, so prune rule is taken to be carried out to it from trimming, it is specified that the node being selected can not
To be chosen again by the automatic machine of other nodes；
5th, transmit the stage, the node elected according to MQEN carries out the transmission of data, if found in transmission process PDR <
100%, then illustrate there is packet loss phenomenon in transmission process, needs to retransmit data at this time.
2. the QoS CR- LDP method according to claim 1 towards the optimization of edge calculations node energy, feature exist
In behavior aggregate formation stages described in step 2 include:
2.1st, source node broadcasts one article to the single-hop neighbor node of oneself in communication range and carries address and synchronizing information
Data packet, neighbor node receive equally also return to after data packet one include oneself address and synchronizing information data packet, source section
Point can return again to an ACK packet after receiving data packet, for notifying node to have built up connection, can transmit each other information, source
After node receives the response of neighbor node, the optional movement that the node receiveed the response will be sent is transmitted as oneself next hop information
Collection；
2.2nd, after node automatic machine is activated, a movement is randomly choosed from the behavior aggregate of oneself, then sending information will
The automatic machine activation of selected movement corresponding node, this node form the behavior aggregate of oneself in the same way again, repeat,
Until aggregation node is accessed in the movement selected by the automatic machine of a certain node；
2.3rd, in behavior aggregate forming process, as soon as the node is put into ACTIV_SET set whenever a node is activated
In, terminate to algorithm, what is stored in ACTIV_SET set is exactly the node experienced from source node to aggregation node.
3. the QoS CR- LDP method according to claim 1 towards the optimization of edge calculations node energy, feature exist
Include: in, rewards and punishments stage described in step 3
If the 3.1st, set ACTIV_SET interior joint quantity is less than MIN_ACTIV_SET, just will set ACTIV_SET
In node be assigned to set MIN_ACTIV_SET；If the product of set ACTIV_SET interior joint probability is greater than threshold, or
Person's m value is greater than Tk, then algorithm terminates, and the node in MIN_ACTIV_SET is that the optimal path that elects of MQEN algorithm is passed through
The node crossed；
If the 3.2nd, set ACTIV_SET interior joint quantity and the value of DT (l) are both less than the value of last round of iteration, and PDR
More than or equal to the value of last round of iteration, then the value of β (n) is that 1, ENV just rewards automatic machine；Otherwise, β (n) value is
0, ENV just punishes automatic machine；
3.3rd, ENV provides an enhancing signal to the automatic machine being activated in last round of time, and then the value of m is pressed from increasing 1
Restart the selection of next round node according to step 1.
4. the QoS CR- LDP method according to claim 1 towards the optimization of edge calculations node energy, feature exist
In: step 4 includes the waking up nodes mechanism of Energy-aware: when the energy of live-vertex is reduced to 50%E0When, then take strategy
So that it is converted to sleep state, then re-starts the selection of optimal path from other remaining nodes；Wherein E0For the first of node
Beginning energy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910418848.9A CN110167054A (en) | 2019-05-20 | 2019-05-20 | A kind of QoS CR- LDP method towards the optimization of edge calculations node energy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910418848.9A CN110167054A (en) | 2019-05-20 | 2019-05-20 | A kind of QoS CR- LDP method towards the optimization of edge calculations node energy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110167054A true CN110167054A (en) | 2019-08-23 |
Family
ID=67631459
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910418848.9A Withdrawn CN110167054A (en) | 2019-05-20 | 2019-05-20 | A kind of QoS CR- LDP method towards the optimization of edge calculations node energy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110167054A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110809291A (en) * | 2019-10-31 | 2020-02-18 | 东华大学 | Double-layer load balancing method of mobile edge computing system based on energy acquisition equipment |
| CN112380008A (en) * | 2020-11-12 | 2021-02-19 | 天津理工大学 | Multi-user fine-grained task unloading scheduling method for mobile edge computing application |
| WO2021047346A1 (en) * | 2019-09-11 | 2021-03-18 | 北京智芯微电子科技有限公司 | Route orchestration method for edge computing network, and route orchestrator and system |
| CN112738756A (en) * | 2021-01-13 | 2021-04-30 | 中国地质大学（北京） | Internet of things equipment data collection method and device |
| WO2022063391A1 (en) * | 2020-09-22 | 2022-03-31 | Telefonaktiebolaget Lm Ericsson (Publ) | Graph-based systems and methods for controlling power switching of components |
| WO2023274304A1 (en) * | 2021-06-30 | 2023-01-05 | 中兴通讯股份有限公司 | Distributed routing determining method, electronic device, and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100220653A1 (en) * | 2007-11-01 | 2010-09-02 | Hwang So-Young | Multi-path routing method in wireless sensor network |
| WO2011040659A1 (en) * | 2009-09-29 | 2011-04-07 | 경희대학교 산학협력단 | Data transmission method for considering qos and energy consumption in wireless sensor network |
| CN102083162A (en) * | 2011-01-21 | 2011-06-01 | 河海大学常州校区 | Greedy multi-path routing method of wireless multimedia sensor network based on QoS (Quality of Service) |
-
2019
- 2019-05-20 CN CN201910418848.9A patent/CN110167054A/en not_active Withdrawn
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100220653A1 (en) * | 2007-11-01 | 2010-09-02 | Hwang So-Young | Multi-path routing method in wireless sensor network |
| WO2011040659A1 (en) * | 2009-09-29 | 2011-04-07 | 경희대학교 산학협력단 | Data transmission method for considering qos and energy consumption in wireless sensor network |
| CN102083162A (en) * | 2011-01-21 | 2011-06-01 | 河海大学常州校区 | Greedy multi-path routing method of wireless multimedia sensor network based on QoS (Quality of Service) |
Non-Patent Citations (2)
| Title |
|---|
| 安宁: "《基于元胞自动机的无线传感器网络节能路由协议的研究》", 《中国优秀博硕士学位论文全文数据库》 * |
| 江文: "《学习自动机算法与应用》", 《万方数据知识服务平台》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021047346A1 (en) * | 2019-09-11 | 2021-03-18 | 北京智芯微电子科技有限公司 | Route orchestration method for edge computing network, and route orchestrator and system |
| CN110809291A (en) * | 2019-10-31 | 2020-02-18 | 东华大学 | Double-layer load balancing method of mobile edge computing system based on energy acquisition equipment |
| CN110809291B (en) * | 2019-10-31 | 2021-08-27 | 东华大学 | Double-layer load balancing method of mobile edge computing system based on energy acquisition equipment |
| WO2022063391A1 (en) * | 2020-09-22 | 2022-03-31 | Telefonaktiebolaget Lm Ericsson (Publ) | Graph-based systems and methods for controlling power switching of components |
| CN112380008A (en) * | 2020-11-12 | 2021-02-19 | 天津理工大学 | Multi-user fine-grained task unloading scheduling method for mobile edge computing application |
| CN112380008B (en) * | 2020-11-12 | 2022-04-15 | 天津理工大学 | Multi-user fine-grained task unloading scheduling method for mobile edge computing application |
| CN112738756A (en) * | 2021-01-13 | 2021-04-30 | 中国地质大学（北京） | Internet of things equipment data collection method and device |
| WO2023274304A1 (en) * | 2021-06-30 | 2023-01-05 | 中兴通讯股份有限公司 | Distributed routing determining method, electronic device, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110167054A (en) | A kind of QoS CR- LDP method towards the optimization of edge calculations node energy | |
| Zhang et al. | Novel PEECR-based clustering routing approach | |
| Al-Kashoash et al. | Congestion control in wireless sensor and 6LoWPAN networks: toward the Internet of Things | |
| Di Francesco et al. | Reliability and energy-efficiency in IEEE 802.15. 4/ZigBee sensor networks: An adaptive and cross-layer approach | |
| Nuggehalli et al. | Energy-efficient caching strategies in ad hoc wireless networks | |
| Mohanty et al. | Energy efficient structure-free data aggregation and delivery in WSN | |
| CN102026329B (en) | Wireless communication network and self-adaptive route selecting communication method thereof | |
| Althamary et al. | A survey on multi-agent reinforcement learning methods for vehicular networks | |
| Yau et al. | Application of reinforcement learning to wireless sensor networks: models and algorithms | |
| CN114138373B (en) | Edge computing task unloading method based on reinforcement learning | |
| Wu et al. | Resource allocation for delay-sensitive vehicle-to-multi-edges (V2Es) communications in vehicular networks: A multi-agent deep reinforcement learning approach | |
| Al-Abiad et al. | Energy-efficient resource allocation for federated learning in noma-enabled and relay-assisted internet of things networks | |
| Chen et al. | Reinforcement learning-based sensor access control for WBANs | |
| Galluccio et al. | A MAC/Routing cross-layer approach to geographic forwarding in wireless sensor networks | |
| Bouachir et al. | EAMP-AIDC-energy-aware mac protocol with adaptive individual duty cycle for EH-WSN | |
| Akbaş et al. | Lightweight routing with dynamic interests in wireless sensor and actor networks | |
| Tham et al. | A load balancing scheme for sensing and analytics on a mobile edge computing network | |
| Lehong et al. | A survey of LoRaWAN adaptive data rate algorithms for possible optimization | |
| Qiu et al. | Maintaining links in the highly dynamic fanet using deep reinforcement learning | |
| Waqas et al. | A game theoretical approach for topology control in wireless ad hoc networks with selfish nodes | |
| Kordafshari et al. | A joint duty cycle scheduling and energy aware routing approach based on evolutionary game for wireless sensor networks | |
| CN102202344B (en) | Cross-layer congestion control method based on game theory in wireless Mesh network | |
| Patil et al. | A novel AEB-AODV based AADITHYA cross layer design hibernation algorithm for energy optimization in WSN | |
| Peng et al. | Real-time transmission optimization for edge computing in industrial cyber-physical systems | |
| Lodhi et al. | Transient Multipath routing protocol for low power and lossy 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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190823 |
|
| WW01 | Invention patent application withdrawn after publication |