CN101132595A - Energy management method for wireless network measurement - Google Patents
Energy management method for wireless network measurement Download PDFInfo
- Publication number
- CN101132595A CN101132595A CNA2007101753982A CN200710175398A CN101132595A CN 101132595 A CN101132595 A CN 101132595A CN A2007101753982 A CNA2007101753982 A CN A2007101753982A CN 200710175398 A CN200710175398 A CN 200710175398A CN 101132595 A CN101132595 A CN 101132595A
- Authority
- CN
- China
- Prior art keywords
- node
- target
- measurement
- nodes
- state
- 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
- 238000005259 measurement Methods 0.000 title claims abstract description 124
- 238000007726 management method Methods 0.000 title claims abstract description 30
- 238000004891 communication Methods 0.000 claims abstract description 25
- 230000004622 sleep time Effects 0.000 claims abstract description 14
- 238000012546 transfer Methods 0.000 claims abstract description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 50
- 238000005265 energy consumption Methods 0.000 claims description 48
- 238000005457 optimization Methods 0.000 claims description 39
- 238000000034 method Methods 0.000 claims description 31
- 238000002922 simulated annealing Methods 0.000 claims description 31
- 230000002068 genetic effect Effects 0.000 claims description 27
- 239000002245 particle Substances 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 18
- 230000007246 mechanism Effects 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000033001 locomotion Effects 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 10
- 230000001133 acceleration Effects 0.000 claims description 7
- 230000005540 biological transmission Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 5
- 230000007704 transition Effects 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 4
- 238000000342 Monte Carlo simulation Methods 0.000 claims description 2
- 238000012952 Resampling Methods 0.000 claims description 2
- 238000013459 approach Methods 0.000 claims description 2
- 230000008878 coupling Effects 0.000 claims description 2
- 238000010168 coupling process Methods 0.000 claims description 2
- 238000005859 coupling reaction Methods 0.000 claims description 2
- 238000012804 iterative process Methods 0.000 claims description 2
- 230000000644 propagated effect Effects 0.000 claims description 2
- 230000001960 triggered effect Effects 0.000 claims description 2
- 238000004088 simulation Methods 0.000 description 6
- 238000000137 annealing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 240000006829 Ficus sundaica Species 0.000 description 1
- 101100172132 Mus musculus Eif3a gene Proteins 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000008303 genetic mechanism Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- 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
- Mobile Radio Communication Systems (AREA)
Abstract
This invention relates to an energy management method capable of using experienced information to layout sleep time of nodes and optimizing measurement and comunication energies based on the awaken state of nodes including: 1, a central node collects measurement information of each round of radio sensor nodes and computes the current position of a target to forecast its track, 2, when there is not target in a measurement region, the network is at the standby state, after a target enters into the region, a network node makes sure the shortest time the target entering into the region based on its forecast position and enters into a complete sleeping state in the shortest time period, 3, selecting a suitable node from being selected ones to finish actual measurement to minimize energy loss, 4, selecting part of nodes as the transfer nodes for transferring data packets to save communication energy.
Description
Technical Field
The invention relates to the field of energy management of wireless network measurement, in particular to an energy management method for wireless network measurement, wherein a plurality of nodes in a wireless network cooperatively track a maneuvering target.
Background
A large number of intelligent nodes with sensing, processing and wireless communication functions form a wireless network in a specific area to complete complex measurement tasks. It is noted that network nodes are typically battery powered and are subject to energy constraints, and thus energy management of wireless networks is a very important issue. The basic idea of energy management is to turn off the device when not needed and wake it up when necessary. Therefore, it is necessary to design an effective node wake-up mechanism.
Currently, there are many devices with multiple energy modes. For example, the StrongARM SA-1100 processor has three energy modes: "running", "idle", and "sleeping". Similarly, bluetooth wireless devices have four different energy consumption modes: "active", "maintenance", "breathing" and "stay". A plurality of such devices are integrated in a network node, having a range of sleep states of different power, according to various combinations of device energy patterns. Consider a network node that includes a CPU, memory, sensors, and wireless communication modules, each of which can enter different energy modes as required.
Although the above modes can solve the problem of energy saving to some extent, the effect of energy saving is not good, and the problem of energy saving efficiency is still raised.
Disclosure of Invention
The invention aims to provide an energy management method for wireless network measurement, which can adopt prior information of target motion to plan network node sleep time and optimize measurement and communication energy consumption according to a node awakening state for maneuvering target tracking application of a wireless network, thereby saving energy.
The invention provides an energy management method based on maneuvering target prediction, which comprises the following steps: the method comprises the steps of predicting the movement of a maneuvering target by adopting a particle filtering algorithm, reasonably planning node sleep time according to a prediction result, designing a dynamic awakening mechanism of a network node, shortening the time of the node in a complete activation state, optimally selecting a measurement node by adopting a distributed genetic simulation annealing algorithm under the condition that nodes around the target are awakened and the target prediction position is known, minimizing measurement energy consumption on the premise of ensuring target measurement precision, and simultaneously optimizing a communication path by adopting a transfer node to reduce communication energy consumption.
The main steps of the invention are further explained as follows:
1. maneuvering target prediction
Assuming that a wireless network monitoring object is a vehicle target and has the maximum acceleration a max And a maximum velocity v max . The central node is responsible for collecting the measurement information of each round of the wireless sensing node, calculating the current position of the target and predicting the track of the target.
And describing the position, the speed and the acceleration information of the target by adopting a motion model. The equation of state is X (k + 1) = FX (k) + G 1 U(k)+G 2 V (k), observed equation Z (k + 1) = HX (k + 1) + W (k + 1). Where k is the number of measurement steps, X (k) is a state vector related to two axial positions and velocities of the target, Z (k) is an observation vector, U (k) is an input vector related to X, y axial motion accelerations of the target, V (k) and W (k) are process noise and observation noise, respectively, F is a nonlinear matrix for state vector transformation, G is the number of measurement steps, and 1 (k) For the purpose of mechanically coupling an input matrix, G 2 (k) Is the process noise input matrix and H is the model observation matrix. For two-dimensional motion, the matrix is defined as follows:
where T is the radio network measurement period.
The particle filter algorithm realizes sequential important sampling filtering based on Bayesian sampling estimation through unparameterized Monte Carlo simulation, random samples (particles) propagated in a state space are adopted to approximate a posterior probability density function of target measurement, and a sample mean value replaces integral operation to obtain state minimum variance estimation. In the stable random process, N random sample points are selected for the posterior probability density measured in the step (k-1), after the measurement information in the step (k) is obtained, the posterior probability density of N particles can be approximate to the posterior probability density measured in the step (k) through state updating. And as the number of the particles increases, the probability density function of the particles gradually approximates to the probability density function of the state, and the particle filter estimation is the optimal Bayes estimation. The next step of the target location can be estimated based on the prior probability density of the particles.
2. Dynamic wake-up mechanism design
The network node comprises a CPU, a memory, a sensor and a wireless communication module. Assuming various sleep states s k Has an energy consumption of P k The time for switching from the fully activated state to this state and for recovery is τ d，k And τ u，k (for any i > j, P j ＞P i ，τ d，i ＞τ d，j And τ is u，i ＞τ u，j ). The power consumption of the network node transitioning between sleep states employs a linear variation model. E.g. node slave state s 0 Transition to state s k Time, wireless communication, memoryAnd gradual powering down of various parts of the processor will result in a linear change in power consumption. Let the conversion time tau k ＝τ d，k ＝τ u，k Then the node goes from sleep state s p Conversion to s q The energy consumption is calculated as follows:
as shown in fig. 1 of the accompanying drawings, 5 sleep states are defined for a network node:
(1) State s of 0 Is the fully active state of a node that can sense, process, send and receive data, E tx Representing additional energy consumption for transmitting data. Is located in (x) i ，y i ) Is located at (x) j ，y j ) When the node of (D) sends data, E tx The calculation is as follows:
ψ tx ＝α 1 r+α 2 d ij n0 r (3)
where r is the data rate, n 0 Is the decay exponent.
(2) State s of 1 The node is in sensing and receiving mode, and the processor is in standby state.
(3) State s of 2 And state s 1 Similarly, the difference is that the processor is powered down and can be awakened when the sensor or radio receives data.
(4) State s of 3 The sensing only mode, except for the sensor module, is off.
(5) State s 4 Indicating a fully off state of the device, is a fully asleep state.
When no target enters the measuring area, the network is in a standby state, and the shortest time for the target to enter the sensing range can be determined according to the shortest distance from the network node to the boundary of the measuring area; when the target enters the measurement area, the network node can also determine the shortest time for the target to enter the sensing range of the target according to the target predicted position. During which the network node may enter a fully sleeping state s 4 Without losing the event. In the dynamic awakening mechanism, the node is awakened by a clock of the node according to the sleep time. The network node wakes up to s first 2 State, except for the nodes triggered by the sensor at the same time, the other nodes return to the state s directly 4 . These are in s 2 The network node in the state can respond to external information and is a candidate node for target measurement.
3. Node selection optimization
The optimization problem is to select a proper node from the candidate measuring nodes to complete the actual measuring task, and the aim is to minimize the network energy consumption on the premise of meeting the requirement of target measuring precision.
If the maneuvering target is located at the time t and can be simultaneously detected by n (n is more than 2) network nodes, each node can acquire the direction angle information of the maneuvering target, and the Direction Finding (DF) error of the maneuvering target has zero mean, gaussian distribution and constant variance. Because direction-finding errors exist, direction-finding lines of all nodes cannot intersect at one point, and therefore the target positioning of multi-node cooperation is achieved by adopting a nonlinear least square estimation method. According to the n node measurement equations, the estimation value of the current target position can be iteratively obtained by using a least square criterion. The covariance matrix of the estimated error is positive, and its associated quadratic form defines an ellipse describing the distribution of the two-dimensional error. The multi-sensor cooperative measurement error index is determined by an error ellipse. For any node selection mode, the selected node needs to transmit the measurement data to the central node, and the network node is in a state s 0 The energy consumption for data transmission can be used as energy indexAnd (4) marking.
Since the target prediction position is known, the node selection mode can be evaluated by adopting the two indexes. By utilizing the potential of wireless network distributed computation and combining the advantages of genetic algorithm and simulated annealing, a distributed genetic simulated annealing algorithm is provided for node selection optimization. From s in the current measurement period 1 Randomly selecting one node from the nodes of the state as a management node, and distributing solution individuals of the genetic algorithm to the rest s 1 A node of a state. And each node performs an SA optimization algorithm in a distributed manner, and the management node collects the distribution optimization results at fixed time intervals to perform crossing and mutation operations, redistributes the optimal individuals and obtains an optimal solution through iteration. In the optimization process, the data volume of individual communication among the network nodes is small, and the communication energy consumption can be generally ignored.
4. Communication path optimization
In each measurement period, a group of network nodes finish measurement tasks around the target and gather the acquired data and send the data to the central node. And sequentially aggregating and transmitting the data to the central node from the network node farthest from the central node according to the sequence from far to near. Of course, the target area may be far from the central node, and even sending data using the measuring node closest thereto still consumes a lot of energy. Therefore, it is considered that a part of the slave nodes is randomly selected as a transit node for transit of the data packet to save communication energy consumption. In order to balance the energy consumption of the nodes, the transit nodes are updated according to the measurement period.
The invention has the following effects: maneuvering target tracking as a typical application of a wireless network generally needs to adopt random arrangement of nodes in a wireless network measurement area, and each network node has a function of direction finding of a maneuvering target. Each sensor is set to have the same measurement range, and the wireless network performs measurement in a fixed period. Due to the density of node arrangement, multiple sensors can be used for measuring the same event at the same time, and therefore the data fusion can be carried out by using a multi-sensor cooperation measuring model. In the actual measurement process, each network node sends the measured value to the central node, the central node obtains the current position of the maneuvering target according to a nonlinear least square estimation method, and the corresponding error ellipses determine the cooperative measurement errors of the network nodes. Based on the trajectory of the maneuvering target, the position of the target at the next measurement instant can be predicted. Particle filter algorithms (PFs) are commonly used to estimate nonlinear, non-gaussian dynamics processes. The present invention uses the PF for the prediction of the position of a maneuvering target. And each network node estimates the idle time according to the priori information obtained by prediction, and performs reasonable sleep time planning, thereby realizing a dynamic awakening mechanism of the network node. The dynamic awakening mechanism ensures that network nodes around the target can be awakened to the state s in time 2 And the candidate measurement node is used as the candidate measurement node. For a given network measurement precision requirement, a group of network nodes meeting the requirement can be selected from candidate nodes according to a maneuvering target predicted position and a multi-sensor cooperative measurement model to carry out actual measurement, so that the number of nodes which need to be awakened to a fully activated state is reduced. To minimize the overall energy consumption of the selected network node for the measurement, it is necessary to use a s-basis 0 Energy index is designed according to energy consumption condition of state to select nodesAnd (6) selecting optimization. Consider some conventional combinatorial optimization algorithms such as Genetic Algorithm (GA) and Simulated Annealing (SA). GA is a kind of randomized search algorithm which uses natural selection and natural genetic mechanism in biology. For a given optimization problem, the GA genetically encodes parameters such that each string of coding individuals represents one possible solution to the problem. GA formation for a group of individualsGenerally, three operators are adopted for the operation of the population: replication, crossover, and mutation. The basic functions of these three operators are to copy individuals, exchange partial information of individuals, and change individual codes of individuals. The GA randomly uses these operators and optimizes the fitness function by iteration. SA analogizes combinatorial optimization to the process of physically annealing a solid from high to low temperatures to a minimum free energy state. When the SA is adopted to solve the optimization problem, only one random state is started, and the current state is changed to a neighboring state according to a perturbation mechanism. If the state is more optimal, accepting the state as the current state; otherwise, a random acceptance criterion (Metropolis criterion) is adopted to judge whether to accept the state. These two types of algorithms have the characteristics: (1) GA operates on a group of solutions, SA only keeps the group of solutions, GA can obtain redundant information searched in the past from a plurality of solutions, and effective parts of search results are utilized through a crossover operator, so that the global search capability is strong; (2) the SA has stronger local searching capability, and can obtain a global optimal solution which cannot be guaranteed by the GA theoretically. In addition, in a wireless network, each network node has processing functionality, providing distributed computing power. Therefore, the invention designs a distributed genetic modeling annealing algorithm (DGASA) for node selection optimization by combining the advantages of the two algorithms. Dynamic energy optimization can be performed by using the target prior information so as to reduce the node energy consumption. Each measurement node needs to transfer the acquired data to the central node. Because the energy consumption of the nodes for directly carrying out remote communication is higher, the invention provides a multi-hop communication path optimization scheme adopting the transit nodes, and the communication energy consumption is further reduced。
Experiments show that the energy management method can effectively reduce the energy consumption of the wireless network on the premise of ensuring the measurement precision.
Drawings
5 sleep states of the network node of fig. 1;
FIG. 2 is a system diagram of a method for energy management in wireless network measurements;
fig. 3 is a node arrangement of a wireless network;
FIG. 4 is a trajectory of a motorized target;
FIG. 5 illustrates a node state transition scheme using a dynamic wake-up mechanism;
FIG. 6 is a distributed genetic simulated annealing algorithm encoding scheme;
FIG. 7 is a distributed genetic simulated annealing algorithm workflow;
FIG. 8 illustrates a communication path optimization approach;
wherein, the diagram (a) is the case of not adopting a transit node; the diagram (b) is the case of using a transit node;
FIG. 9 is a diagram of target location prediction error using a particle filter algorithm;
wherein the deviation in the X-axis direction is shown in graph (a); FIG. (b) shows the Y-axis direction deviation;
FIG. 10 is a graph of measurement error optimized using three different algorithms;
wherein the graph (a) is the measurement error optimized by the genetic algorithm; graph (b) is the measurement error for simulated annealing algorithm optimization; graph (c) is the measurement error for the distributed genetic simulated annealing algorithm optimization;
FIG. 11 is a theoretical energy consumption optimization result using three algorithms, a simulated annealing algorithm and a distributed genetic simulated annealing algorithm;
fig. 12 shows the average energy consumption of the wireless network when different transfer node ratios are used;
FIG. 13 is a measurement error optimized by the distributed genetic simulated annealing algorithm when 10% transit nodes are employed;
FIG. 14 is a graph of energy consumption simulation optimized using a distributed genetic simulated annealing algorithm with 10% transit nodes and without transit nodes.
Detailed Description
The invention is further described by the drawings and the detailed description.
As shown in fig. 2, the present invention is directed to energy management for a network node comprising a sensor, a CPU, a memory, a sensor, and a wireless communication module. The modules have multiple energy modes, and the network node can enter different sleep states. The energy consumption of the network node is determined by the respective sleep state and data transmission situation. The multiple sensors of each node need cooperative measurement to obtain a target position, and the measurement error of the target position is determined by an error ellipse through nonlinear least square estimation. In the energy management process, the particle filter algorithm predicts the next position of the target according to the existing target position measurement value. The sleep time of the network node can be estimated by adopting the predicted value, so that the node is dynamically awakened. The awakened network node is a candidate node for target measurement, and a distributed genetic simulated annealing algorithm is adopted to optimize and select an actual measurement node, so that the measurement precision requirement is met, and the energy consumption in the measurement process is minimized. Meanwhile, a communication path is optimized by adopting a transfer node, and communication energy consumption is reduced by utilizing a multi-hop mode.
Embodiments of the invention are further illustrated below by examples of energy management.
Assuming that the area of the wireless network is 400m × 400m, the network includes 256 nodes randomly arranged and 1 central node located at the center of the network, as shown in fig. 3. The network node adopts a Pyroelectric Infrared (PIR) sensor, and the measurement range r sensing =60m, direction finding precision mean square value is sigma θ =2 °, network measurement period T =0.5s. Is provided withMaximum acceleration a of maneuvering target max ＝10m/s 2 Maximum velocity v max =40m/s, the movement locus thereof is generated in the measurement area as shown in fig. 4, including 120 locus points.
Next, a particle filter algorithm is used to predict the target position. Firstly, initializing a sampling point set and an importance weight, and then performing an iterative process: (1) constructing a sampling point set and updating the weight; (2) if the effective sampling scale is smaller than the theoretical value, resampling the sample; (3) and (4) weighting the particles and updating the state. In each measurement cycle, the updated state is the target position estimate at the next measurement time.
According to the target position prediction result, node sleep time planning can be carried out, and the method is mainly divided into the following two stages:
1. establishing phase
It is assumed that no maneuvering target is present in the measurement area at network initialization. According to each node to the wireless networkShortest distance d of the boundary of the network min And a target maximum velocity v max The shortest time for the target to reach the measurement range of each node can be calculated as follows: t is t min ＝(d min -r sensing )/v max . Nodes near the target predicted location need to be woken up, so the sleep time is t sleep ＝t min -T. According to the network measurement period, the sleep period number of the node is as follows:
wherein [ g ] is a rounding operation. Therefore, the nodes close to the boundary are awakened every measurement period, and are defined as boundary nodes, and the rest nodes are defined as internal nodes.
2. Tracking phase
When the target enters the network, according to the position (x) of each internal node i ，y i ) And the current position (x) of the target target ，y target ) (prediction value) estimation of idle time：t sleep ′＝(d target，i -r sensing )/v max -T. But for a new target that may enter the network, the node still needs to be prepared at t sleep The measurement is performed later, so the number of idle cycles of the node is:
the node state transition mode using dynamic wake-up mechanism is shown in fig. 5, where the network node uses a cycle number n sleep And entering sleep. The network node is awakened to s by clock triggering 2 In this state, the central node re-estimates the number of sleep cycles and sends the result to the corresponding node. The nodes needing to execute the task return s after completing measurement and transmission 4 State, other nodes return directly to s 4 The state and sleep time are determined by the received cycle value. Therefore, the sleep time of the network node is prolonged as far as possible on the premise of ensuring timely capturing of a new target and awakening of nodes around the target prediction position.
And selecting and optimizing the measuring nodes after target prediction and node awakening. In the distributed genetic simulated annealing algorithm, the number of candidate nodes is assumed to be N, the number of the candidate nodes is 1,2,
and N. As shown in fig. 6, with binary encoding, each solution is represented as a binary one-dimensional array of length N, where "0" indicates that the node on the corresponding bit is not selected, and "1" indicates that the node on the corresponding bit is selected.
And defining the length of the longer semi-axis of the multi-sensor cooperation measurement error ellipse as a measurement error index. Assume a measurement error threshold A required by a mobile target tracking application 0 =0.6m, the minimization objective function is designed as follows:
wherein A is perThe theoretical measurement error corresponding to each solution, E, is an energy index representing the total energy consumption measured, and the specific calculation method thereof will be described in the following section, E 0 Is an upper bound representing the energy index E. Therefore, the theoretical measurement error is optimized in the optimization process, and the network energy consumption is optimized after the theoretical measurement error meets the requirements.
FIG. 7 illustrates the execution flow of the distributed genetic simulated annealing algorithm. Randomly selecting a management node from the current measurement nodes, the node generating a group of solutions and distributing the solutions to the rest N s And a measuring node. Each measurementAnd executing a simulated annealing algorithm on the obtained solution by the node, collecting the distribution optimization result by the management node at fixed time intervals, executing genetic operation, and redistributing the result to the rest measurement results. In this way, an optimal solution can be obtained through a number of iterations.
Next, considering the energy consumption of the target measurement process, α in equation (3) 1 ＝50nJ/b， α 2 ＝100pJ/b/m 2 ，n 0 And =3. When the measuring nodes send data to the central node, all the nodes send wireless signals with the same energy, and the central node sorts the nodes according to the distance according to the strength of the received signals. As shown in fig. 8 (a), the nodes transmit and aggregate the data packets in a near-far order, and finally the data packets are directly transmitted to the central node by the node closest to the central node. Setting central Node as Node 0 For a group of nodes ordered from near to farThe energy consumption index is calculated as follows:
wherein d is ij Is Node i And Node j P is the packet size.
As shown in fig. 8 (b), when the measurement node set is farther from the central node, to save communication energy consumption, a certain proportion of transit nodes are selected from the internal nodes to transit data between the measurement node and the central node. The selection principle is as follows: each node randomly generates a random number in an interval [0,1], and if the random number is larger than a threshold Th, the node is selected as a transfer node. The threshold Th is calculated as follows:
in the formula, p is the proportion of the transfer nodes, r is the serial number of the current measurement period, and G is the node set which is not selected as the transfer node in the last 1/p measurement periods. The internal node wakes up to s in the measurement period selected as the transit node 2 And in the state, the central node activates the transfer node with the least energy consumption to transfer data. Set up a set of transit nodes as
The energy consumption index is:
where Δ E is the extra energy consumed to wake up the transit node.
In a simulation experiment, the total target tracking time is 1200s, the time for selecting and optimizing the nodes is 0.15s, the media access protocol of the wireless network is CSMA/CA, and the data rate is 1Mbps.
The energy saving obtained by the node sleep time planning is first analyzed separately. And (3) counting the energy consumption of the network establishment stage, and obtaining the total energy consumption of the network tracking stage under the condition of selecting all nodes in the target field as measurement nodes. At s with network node 3 Compared with the situation that the state waits for the triggering of the sensor, the overall energy consumption in the network establishing and tracking stages is lower by adopting the particle filtering track prediction and dynamic awakening mechanism.
The performance of the distributed genetic simulated annealing algorithm was then evaluated. And in each measurement period, selecting an optimal node actually measured in the target field by adopting a distributed genetic simulated annealing algorithm according to the target prediction position obtained by particle filtering. And when the selected nodes are adopted to measure and report data, the measurement precision and the energy consumption in the network tracking stage are analyzed. Meanwhile, the optimized performance of GA and SA under the same coding mode is analyzed for comparison. FIG. 9 shows the PF error on the target next round position prediction during tracking, including X-axis and Y-axis deviations. As can be seen, all deviations are within the interval [ -1,1 ]. Consider the tracking process of the target inside the network and therefore analyze only the performance of the algorithm within 10s to 50 s. FIG. 10 compares the measurement error optimization results of the GA algorithm, the SA algorithm and the DGASA algorithm, and the actual measurement error of the distributed genetic simulation annealing algorithm is the lowest superscalar of the relative error index. FIG. 11 shows the results of energy consumption optimization for various algorithms, with the highest energy consumption for GA optimization, the next highest energy consumption for SA optimization, and the lowest energy consumption for DGASA optimization.
In addition, consider the situation that uses the transit node route in the DGASA algorithm optimization, and adjust the proportion of the transit node from 0% (without using the transit node) to 30%, and carry out the simulation respectively. The influence of the transit node ratio on the theoretical average energy consumption in the measurement process is shown in fig. 12. As can be seen from the figure, the use of transit nodes may result in additional energy savings compared to the case where transit nodes are not used, and the energy consumption level is lowest when the proportion of transit nodes is 10%. The measurement error optimization result of the distributed genetic simulated annealing algorithm when the 10% transit node is adopted is shown in fig. 13, and fig. 14 compares the energy consumption optimization result of the distributed genetic simulated annealing algorithm when the 10% transit node is adopted and the transit node is not adopted.
In conclusion, the energy management method realized by using particle filter trajectory prediction and distributed genetic simulation annealing algorithm optimization in the wireless network can obviously improve the energy effectiveness on the premise of meeting the target tracking precision, and more energy can be saved by introducing the transit node path optimization mode.
Claims (10)
1. The energy management method for wireless network measurement is characterized by comprising the following steps: (1) The maneuvering target is predicted, and the central node is responsible for collecting the measurement information of each round of the wireless sensing node, calculating the current position of the target and predicting the track of the target; (2) The dynamic awakening mechanism is designed, when no target enters a measuring area, the network is in a standby state, after the target enters the measuring area, the network node determines the shortest time for the target to enter the sensing range of the target according to the target predicted position, and the network node can enter a complete sleep state within the shortest time range; (3) Selecting and optimizing nodes, namely selecting proper nodes from candidate measuring nodes to complete an actual measuring task in a measuring period, and minimizing network energy consumption on the premise of meeting the requirement of target measuring precision; (4) And optimizing the communication path, namely selecting a part of nodes as transfer nodes for transferring the data packet so as to save communication energy consumption.
2. The method of energy management for wireless network measurements as claimed in claim 1, wherein said maneuvering target prediction specifically comprises the steps of: describing position, speed and acceleration information of the target by adopting a motion model, wherein the state equation is X (k + 1) = FX (k) + G 1 U(k)+G 2 V (k), the observation equation is Z (k + 1) = HX (k + 1) + W (k + 1), where k is the number of measurement steps, X (k) is the state vector associated with two axial positions and velocities of the target, Z (k) is the observation vector, U (k) is the input vector associated with the acceleration of the X, y axial motion of the target, V (k) and W (k) are the process noise and the observation noise, respectively, F is the nonlinear matrix of state vector transitions, G (k) is the input vector associated with the acceleration of the X, y axial motion of the target, and F is the nonlinear matrix of state vector transitions 1 (k) For the purpose of mechanically coupling the input matrix, G 2 (k) Is the process noise input matrix and H is the model observation matrix. For two-dimensional motion, the matrix is defined as follows:
where T is the wireless network measurement period,
the particle filter algorithm realizes sequential important sampling filtering based on Bayesian sampling estimation through unparameterized Monte Carlo simulation, random samples (particles) propagated in a state space are adopted to approximate a posterior probability density function measured by a target, a sample mean value replaces integral operation to obtain state minimum variance estimation, N random sample points are selected for the posterior probability density measured in the k-1 step in a stable random process, after measurement information in the k step is obtained, the state is updated, the posterior probability density of N particles can be approximate to the posterior probability density measured in the k step, the probability density function of the particles gradually approaches to the probability density function of a state along with the increase of the number of the particles, the particle filter estimation is optimal Bayesian estimation, and the position of the next step of the target can be estimated according to the prior probability density of the particles.
3. The method of claim 1, wherein the dynamic wake-up mechanism design comprises the following steps: the network node defines 5 sleep states,
(1) State s 0 Is the fully active state of the node;
(2) State s 1 The node is in a sensing and receiving mode, and the processor is in a standby state;
(3) State s 2 And state s 1 Similarly, the difference is that the processor is powered off and can wake up when the sensor or radio receives data;
(4) State s of 3 The sensing only mode, except the sensor module, is off;
(5) State s 4 Indicating a fully off state of the device, being a fully asleep state;
when no target enters the measuring area, the network is in a standby state and reaches the boundary of the measuring area according to the network nodeThe shortest distance of (2) can determine the shortest time for the target to enter the sensing range; when the target enters the measuring area, the network node can also determine the shortest time for the target to enter the sensing range according to the target predicted position, and the network node enters a full sleep state s in the time 4 The node is awakened by the clock of the node according to the sleep time, and the network node is awakened to s 2 State, except for the nodes triggered by the sensor at the same time, the other nodes return to the state s directly 4 These are in s 2 The network node in the state can respond to the external message and is a candidate node for target measurement.
4. The method of claim 1, wherein the node selection optimization comprises the steps of: if the maneuvering target is located at the moment t and can be simultaneously detected by n (n is more than 2) network nodes, all the nodes read the direction angle information of the maneuvering target, the nonlinear least square estimation method is adopted to realize the target positioning of the multi-node cooperation, the least square criterion is utilized to iterate to obtain the estimation value of the current target position according to n node measurement equations, the selected node transmits the measurement data to the central node, and the network nodes are in the state s 0 And the energy consumption of data transmission is taken as an energy index; from s in the current measurement period 1 Randomly selecting one node from the nodes of the state as a management node, and distributing solution individuals of the genetic algorithm to the rest s 1 And the nodes in the state are distributed and execute an SA optimization algorithm, the management nodes collect the distribution optimization results at regular intervals to carry out cross and variation operation, the optimal individuals are redistributed, and the optimal solution is obtained through iteration.
5. The method of claim 1, wherein the communication path optimization step comprises the steps of: in each measurement period, a group of network nodes completing measurement tasks around a target gather and send collected data to a central node, the data are sequentially gathered from the network node farthest from the central node according to a sequence from far to near and are transmitted to the central node, a part of the data are randomly selected from the nodes to serve as transit nodes for transit data packets so as to save communication energy consumption, and the transit nodes are updated according to the measurement period.
6. The energy management method for wireless network measurement according to claim 1 or 2, wherein the target position is predicted by using a particle filter algorithm, a sampling point set and an importance weight are initialized, and then an iterative process is performed: (1) constructing a sampling point set and updating the weight; (2) if the effective sampling scale is smaller than the theoretical value, resampling the sample; (3) and (4) weighting the particles and updating the state. In each measurement cycle, the updated state is the target position estimate at the next measurement time.
7. The energy management method of the wireless network measurement according to claim 1 or 3, wherein the node dynamic wake-up mechanism is mainly divided into the following two stages:
(1) A building stage: assuming that no maneuvering target exists in the measurement area when the network is initialized, according to the shortest distance d from each node to the boundary of the wireless network min And a target maximum velocity v max And calculating the shortest time for the target to reach the measurement range of each node: t is t min ＝(d min -r sensing )/v max Sleep time t sleep ＝t min -T, the number of sleep cycles of a node, according to the network measurement period, is:
wherein [ g ] is rounding operation, the node close to the boundary is awakened every measurement period and is defined as the boundary node, and the other nodes are defined as the internal nodes;
(2) A tracking stage: when the target enters the network, according to the position (x) of each internal node i ，y i ) And the current position (x) of the target iarget ，y target ) (prediction value) estimated idle time: t is t sleep ′(d target，i -r sensing )/v max -T, the number of idle cycles of a node being:
network node using a number of cycles n sleep Entering sleep, the network node is awakened to s by clock trigger 2 In the state, the central node can estimate the sleep cycle number again and sends the result to the corresponding node, and the node needing to execute the task returns to s after completing measurement and transmission 4 Status, with the remaining nodes returning s directly 4 The state, sleep time is determined by the received cycle number.
8. The method for energy management of wireless network measurements according to claim 1 or 4, wherein said optimization of measurement node selection comprises the following steps: in the distributed genetic simulated annealing algorithm, assuming that the number of candidate nodes is N, the number of the candidate nodes is 1,2, L and N, binary coding is adopted, each solution is represented as a binary one-dimensional array with the length of N, wherein '0' represents that the node on the corresponding bit is not selected, and '1' represents that the node on the corresponding bit is selected, the length of the semi-major axis of the multi-sensor cooperation measurement error ellipse is defined as a measurement error index, and a measurement error threshold A required by a maneuvering target tracking application is assumed 0 =0.6m, the minimization objective function is designed as follows:
wherein A is the theoretical measurement error corresponding to each solution, E is the energy index representing the total energy consumption of the measurement, and E 0 In order to represent the upper bound of the energy index E, the theoretical measurement error is optimized in the optimization process, and the network energy consumption is optimized when the theoretical measurement error meets the requirement.
9. The method for energy management of wireless network measurement as claimed in claim 8, wherein the execution flow of the distributed genetic simulated annealing algorithm comprises the following steps: randomly selecting one from the current measurement nodesA management node generating a set of solutions and assigning them to the remaining N s And each measurement node executes a simulated annealing algorithm on the obtained solution, and the management node collects the result of distribution optimization at fixed time intervals to execute genetic operation and redistributes the result to other measurement results, so that the optimal solution can be obtained through multiple iterations.
10. The method for energy management of wireless network measurements according to claim 1, wherein said communication path optimization comprises the steps of: when the measuring nodes send data to the central node, all the nodes send wireless signals with the same energy, the central node sorts the nodes according to the distance according to the strength of the received signals, the nodes transmit and gather data packets according to the distance sequence, and finally the node closest to the central node sends the data packets to the central node through the transfer node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101753982A CN101132595A (en) | 2007-09-29 | 2007-09-29 | Energy management method for wireless network measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101753982A CN101132595A (en) | 2007-09-29 | 2007-09-29 | Energy management method for wireless network measurement |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101132595A true CN101132595A (en) | 2008-02-27 |
Family
ID=39129676
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2007101753982A Pending CN101132595A (en) | 2007-09-29 | 2007-09-29 | Energy management method for wireless network measurement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101132595A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101998598A (en) * | 2009-08-20 | 2011-03-30 | 华为技术有限公司 | Method and device for monitoring explore frame sequences |
| CN102014115A (en) * | 2010-07-09 | 2011-04-13 | 北京哈工大计算机网络与信息安全技术研究中心 | Method, device and system for anonymizing gateway node |
| CN101572936B (en) * | 2009-01-09 | 2011-06-22 | 吴鑫 | Node energy consumption management method for wireless sensor network |
| CN102332983A (en) * | 2010-06-04 | 2012-01-25 | 美国博通公司 | Utilize broadband gateway that the method and system of energy management is provided in home network |
| US8331995B2 (en) | 2010-07-07 | 2012-12-11 | Telefonaktiebolaget L M Ericsson (Publ) | Secondary radio-nodes for mobile communications networks and related methods |
| CN102077513B (en) * | 2008-08-07 | 2014-04-09 | Nec欧洲有限公司 | Method for managing network and network |
| CN104659819A (en) * | 2015-03-19 | 2015-05-27 | 武汉大学 | Wind power adsorption estimation method involving wind power prediction error |
| CN109219118A (en) * | 2018-11-27 | 2019-01-15 | 上海交通大学 | A kind of distributed consistent state estimation method based on sequence wake-up mechanism |
| WO2020029789A1 (en) * | 2018-08-06 | 2020-02-13 | 交叉信息核心技术研究院(西安)有限公司 | Energy management system and method, and electronic device, electronic apparatus and non-volatile processor |
| CN111044926A (en) * | 2019-12-16 | 2020-04-21 | 北京航天智造科技发展有限公司 | Method for predicting service life of proton exchange membrane fuel cell |
-
2007
- 2007-09-29 CN CNA2007101753982A patent/CN101132595A/en active Pending
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102077513B (en) * | 2008-08-07 | 2014-04-09 | Nec欧洲有限公司 | Method for managing network and network |
| CN101572936B (en) * | 2009-01-09 | 2011-06-22 | 吴鑫 | Node energy consumption management method for wireless sensor network |
| CN101998598B (en) * | 2009-08-20 | 2013-06-26 | 华为技术有限公司 | Method and device for monitoring explore frame sequences |
| CN101998598A (en) * | 2009-08-20 | 2011-03-30 | 华为技术有限公司 | Method and device for monitoring explore frame sequences |
| CN102332983B (en) * | 2010-06-04 | 2015-10-28 | 美国博通公司 | Utilize broadband gateway in home network, provide the method and system of energy management |
| CN102332983A (en) * | 2010-06-04 | 2012-01-25 | 美国博通公司 | Utilize broadband gateway that the method and system of energy management is provided in home network |
| US8331995B2 (en) | 2010-07-07 | 2012-12-11 | Telefonaktiebolaget L M Ericsson (Publ) | Secondary radio-nodes for mobile communications networks and related methods |
| CN102014115B (en) * | 2010-07-09 | 2013-07-10 | 北京哈工大计算机网络与信息安全技术研究中心 | Method, device and system for anonymizing gateway node |
| CN102014115A (en) * | 2010-07-09 | 2011-04-13 | 北京哈工大计算机网络与信息安全技术研究中心 | Method, device and system for anonymizing gateway node |
| CN104659819A (en) * | 2015-03-19 | 2015-05-27 | 武汉大学 | Wind power adsorption estimation method involving wind power prediction error |
| WO2020029789A1 (en) * | 2018-08-06 | 2020-02-13 | 交叉信息核心技术研究院(西安)有限公司 | Energy management system and method, and electronic device, electronic apparatus and non-volatile processor |
| US11422818B2 (en) | 2018-08-06 | 2022-08-23 | Institute for Interdisciplinary Information Core Technology (Xi'an) Co., Ltd. | Energy management system and method, electronic device, electronic apparatus, and nonvolatile processor |
| CN109219118A (en) * | 2018-11-27 | 2019-01-15 | 上海交通大学 | A kind of distributed consistent state estimation method based on sequence wake-up mechanism |
| CN111044926A (en) * | 2019-12-16 | 2020-04-21 | 北京航天智造科技发展有限公司 | Method for predicting service life of proton exchange membrane fuel cell |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101132595A (en) | Energy management method for wireless network measurement | |
| Thakur et al. | Fog computing for detecting vehicular congestion, an internet of vehicles based approach: A review | |
| Cassandras et al. | Sensor networks and cooperative control | |
| Teng et al. | Decentralized variational filtering for target tracking in binary sensor networks | |
| Cao et al. | Swarm intelligence-based performance optimization for mobile wireless sensor networks: survey, challenges, and future directions | |
| CN103052128A (en) | Wireless sensor network-based energy-efficient collaborative scheduling method | |
| CN101459915A (en) | Wireless sensor network node coverage optimization method based on genetic algorithm | |
| Liu et al. | Joint design of energy-efficient clustering and data recovery for wireless sensor networks | |
| Hsu et al. | POOT: An efficient object tracking strategy based on short-term optimistic predictions for face-structured sensor networks | |
| Ji et al. | E2PP: An Energy‐Efficient Path Planning Method for UAV‐Assisted Data Collection | |
| Nourredine et al. | A fluid approach to model and assess the energy level of autonomous devices in iot with solar energy harvesting capability | |
| Poe et al. | Minimizing the maximum delay in wireless sensor networks by intelligent sink placement | |
| CN102833812B (en) | Hybrid service-oriented wireless sensor network energy effective access method | |
| Abasi et al. | Fuzzy-clustering based data gathering in wireless sensor network | |
| John et al. | Energy consumption in wireless sensor network | |
| CN109219118B (en) | Distributed consistent state estimation method based on sequence awakening mechanism | |
| Raghunandan et al. | Hierarchical agglomerative clustering based routing algorithm for overall efficiency of wireless sensor network | |
| Li et al. | Cellular genetic algorithms for optimizing the area covering of wireless sensor networks | |
| Enami et al. | A survey on application of neural networks in energy conservation of wireless sensor networks | |
| Zhang et al. | Optimising rendezvous-based data collection for target tracking in WSNs with mobile elements | |
| Mirsadeghi et al. | Low power prediction mechanism for wsn-based object tracking | |
| Nimbalkar | Use of neural network in wsns: A survey | |
| Goudarzi et al. | An efficient clustering algorithm using evolutionary hmm in wireless sensor networks | |
| Wang et al. | Dynamic energy management with improved particle filter prediction in wireless sensor networks | |
| Li et al. | Particle swarm optimization based multi-robot task allocation using wireless sensor network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20080227 |