CN103974276A - Method for extracting signal feature parameters of wireless sensor network in distributed mode - Google Patents

Abstract

The invention discloses a method for extracting signal feature parameters of a wireless sensor network in a distributed mode. The method comprises the following steps that (A) a center node of the wireless sensor network sends feature extraction instructions to N monitoring nodes in the wireless sensor network; (B) each monitoring node establishes a cluster according to the corresponding feature extraction instruction, wherein each cluster comprises one monitoring node and one or more neighbor nodes, the node with the maximum energy in each cluster serves as a cluster head node, and other nodes in each cluster serve as cluster member nodes; (C) the nodes in each cluster are used for collecting different signal feature parameters respectively, and storing the collected signal feature parameters in the corresponding cluster head nodes; (D) the cluster head nodes of the N clusters send the signal feature parameters of the cluster head nodes to the center node, so that the signal feature parameters are classified and identified by the center node. The method can be used for extracting the signal feature parameters in the distributed mode, the overall energy consumption of the wireless sensor network is steadily reduced in the extracting process, and the service life of the wireless sensor network is prolonged to the greatest extent.

Description

A kind of wireless-sensor network distribution type signal characteristic parameter extracting method

Technical field

The present invention relates to wireless sensor network, particularly a kind of wireless-sensor network distribution type signal characteristic parameter extracting method.

Background technology

At present less to the signal Modulation identification technology research based on wireless sensor network, modulation signal characteristic parameter extraction is also limited to single node carries out, and single node feature extraction is to carry out the extraction of signal characteristic parameter by a node.

Appearance along with the development and futures new business of small-power short-range radio communications business, as wireless Internet, wireless sense network and wireless mobile network etc., all shown the wilderness demand to electromagnetic spectrum, and an urgent demand to effective spectrum access, traditional electromagnetic spectrum monitoring network cannot meet and adapt to these requirements, as the important content of electromagnetic spectrum monitoring, the identification of signal modulation system is towards distributed, community system, fine granularity and the future development that becomes more meticulous.Distributed nature parameter extraction is mutually collaborative by the signal being distributed in certain area, and common cooperation extracted characteristic parameter.Its advantage is the complexity while reducing a node processing, meet become more meticulous, short haul connection business development.

Liu Aisheng, Zhu Qi have proved the discrimination when Modulation Identification rate of a plurality of sensor node distributed collaboratives is all carried out Modulation Identification higher than single-sensor single node in < < multisensor node distributed collaborative Modulation Recognition > >.Wang Jinkui has provided distributed frequency spectrum sensor network conceptual design, application model, system core technology in < < distributed frequency spectrum sensor network design > >.Li Zan, learn equality proposed to build the electromagnetic spectrum monitoring net of new generation based on wireless sensor network in the design of < < a new generation electromagnetic spectrum monitoring net and application > >, inquire into the composition 26S Proteasome Structure and Function effect of network, analyzed the meaning of building electrical measurement spectrum monitoring net of new generation.

At present, traditional characteristic parameter extraction is mainly to carry out based on a central station.Due to the development of short haul connection business, proposed the demand of distributed Modulation Identification, so distributed nature parameter extraction is an innovative point.But owing to carrying out based on wireless sensor network, brought again the limited problem of network energy consumption, the present invention mainly solves the reduction network energy consumption problem in distributed nature leaching process.

Summary of the invention

Unresolved the problems referred to above, the invention provides a kind of wireless-sensor network distribution type signal characteristic parameter extracting method.

A kind of wireless-sensor network distribution type signal characteristic parameter extracting method provided by the invention, comprising:

A) Centroid of wireless sensor network sends feature extraction instruction to N monitoring node in described network;

B) each monitoring node is set up one bunch according to described feature extraction instruction, and its each bunch comprises a monitoring node and one or more neighbor node, and using bunch in a node of energy maximum as leader cluster node, other node is as a bunch member node;

C) bunch in, each node gathers respectively different signal characteristic parameters, and by the signal characteristic Parameter storage collecting to leader cluster node;

D) leader cluster node of N bunch is sent to Centroid by signal characteristic parameter separately respectively, for Centroid, carries out Classification and Identification.

Preferably, described steps A) comprising:

Centroid is chosen N monitoring node, and to described N monitoring node, sends feature extraction instruction by broadcast mode.

Preferably, described step B) comprising:

B1) each monitoring node, according to described feature extraction instruction, sends network to its neighbor node and sets up request, and forms one bunch with one or more neighbor nodes of the described network foundation of response request;

B2) each node of described bunch, by mutual positive energy exchange information, obtain having the node of ceiling capacity, and using the described node with ceiling capacity as leader cluster node, other node is as a bunch member node.

Preferably, at described step B1) in, described monitoring node forms one bunch with M node of formerly response.

Preferably, described step C) comprising:

C1) leader cluster node bunch in assigned characteristics extract task;

C2) bunch in, each node exchanges signal to noise ratio information mutually, and according to signal to noise ratio information, gathers corresponding signal characteristic parameter in described feature extraction task.

Preferably, described step C2) comprising:

In bunch, each node, according to the size sequence of its signal to noise ratio, gathers other signal characteristic parameter of respective priority in described feature extraction task.

Preferably, described step D) comprising:

D1) set up for the parameter of signal characteristic separately of N leader cluster node being passed to the routing relation of Centroid;

D2) each bunch of head, according to described routing relation, determined signal characteristic parameter is delivered to the bang path of upper level leader cluster node, and signal characteristic parameter is sent to described upper level leader cluster node;

D3) described upper level leader cluster node, by determining its bang path, forwards the signal characteristic parameter of receiving, and in repeating process, adds the signal characteristic parameter of oneself, thereby the signal characteristic parameter of N leader cluster node is passed to Centroid.

Preferably, described step D1) comprising:

Communication cost between each leader cluster node calculating and other leader cluster node;

Each leader cluster node utilizes described communication cost, sets up the local routing information table that comprises one or more upper level leader cluster nodes to be selected, thereby forms the routing relation from N leader cluster node to Centroid.

Preferably, described step D2) comprising:

Each leader cluster node is according to the communication cost of itself and one or more upper level leader cluster nodes to be selected, calculate the selection probability of described one or more upper level leader cluster nodes to be selected, and according to the size of described selection probability, determine that signal characteristic parameter is delivered to the bang path of upper level leader cluster node.

Compared with prior art, beneficial effect of the present invention is:

The present invention can carry out distributed extraction to signal characteristic parameter, and in leaching process, realizes the steady reduction of network in general energy consumption, extends to greatest extent the life-span of network.

Accompanying drawing explanation

Fig. 1 is a kind of wireless-sensor network distribution type signal characteristic parameter extracting method theory diagram provided by the invention;

Fig. 2 is a kind of wireless-sensor network distribution type signal characteristic parameter extraction flow chart that the embodiment of the present invention provides.

Embodiment

Below in conjunction with accompanying drawing, to a preferred embodiment of the present invention will be described in detail, should be appreciated that following illustrated preferred embodiment, only for description and interpretation the present invention, is not intended to limit the present invention.

Fig. 1 is a kind of wireless-sensor network distribution type signal characteristic parameter extracting method theory diagram provided by the invention, as shown in Figure 1, comprising:

The Centroid of step 101, wireless sensor network sends feature extraction instruction to N monitoring node in described network.

In described step 101, Centroid is chosen N monitoring node in network at random, and to described N monitoring node, sends feature extraction instruction by broadcast mode, starts characteristic parameter extraction process.

Step 102, each monitoring node are set up one bunch according to described feature extraction instruction, its each bunch comprises a monitoring node and one or more neighbor node, and using bunch in a node of energy maximum as leader cluster node, other node is as a bunch member node.

In described step 102, each monitoring node, according to described feature extraction instruction, sends network to its neighbor node and sets up request, and forms one bunch with one or more neighbor nodes of the described network foundation of response request; Each node of described bunch, by mutual positive energy exchange information, obtains having the node of ceiling capacity, and using the described node with ceiling capacity as leader cluster node, other node is as a bunch member node.

Further, the monitoring node of receiving feature extraction instruction can form cluster with M node of response at first, this M+1 node exchanges the energy information of self mutually, choose the node of energy value maximum as leader cluster node, soon wherein the node of energy maximum is as leader cluster node, and other M node is as a bunch member node.

Step 103, bunch in each node gather respectively different signal characteristic parameters, and by the signal characteristic Parameter storage collecting to leader cluster node.

In described step 103, leader cluster node bunch in assigned characteristics extract task, in bunch, each node exchanges signal to noise ratio information mutually, and according to signal to noise ratio information, gathers corresponding signal characteristic parameter in described feature extraction task.

Further, in bunch, each node, according to the size sequence of its signal to noise ratio, gathers other signal characteristic parameter of respective priority in described feature extraction task.Specifically, the node that signal to noise ratio is higher can gather the signal characteristic parameter that priority level is lower.

Take M=5 as example, and leader cluster node assigned characteristics in network is extracted task, and each node determines to gather which signal characteristic parameter according to the state of signal-to-noise of self environment of living in, according to priority R _a> σ _af> σ _aa> γ _max> σ _dp> σ _ap, the node that signal to noise ratio is minimum gathers R _a, the node that signal to noise ratio is the highest will gather σ _ap, other by that analogy.All nodes send to leader cluster node by signal characteristic parameter one jumping collecting.In this way can be so that whole recognition correct rate be improved.

Wherein: civil time the signal characteristic parameter of frequency domain comprise:

1, the spectrum density maximum γ of normalize and center instantaneous amplitude _max

γ _max=max|FFT[a _cn(i)]| ²/N _s

In formula, N _sfor number of sampling, a _cn(i) be normalize and center instantaneous amplitude, by following formula, calculated:

a _cn(i)=a _n(i)-1

In formula, and for the mean value of instantaneous amplitude a (i), object instantaneous amplitude being normalized with mean value is in order to eliminate the impact of channel gain.γ _maxbe used for the changes in amplitude of characterization signal.

2, the standard deviation of the non-weak signal section of zero center instantaneous phase nonlinear component _dp

${\mathrm{\&sigma;}}_{\mathrm{dp}}=\sqrt{\frac{1}{c}\left[\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}{\mathrm{\&phi;}}_{\mathrm{NL}}^2\left(i\right)\right]-{\left[\frac{1}{c}\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}{\mathrm{\&phi;}}_{\mathrm{NL}}\left(i\right)\right]}^2}$

In formula, a _tbe an amplitude decision threshold level of judgement weak signal, c is the number of non-weak signal value in gross sample data, φ _nL(i) be the nonlinear component of instantaneous phase after zero centerization is processed, when carrier wave Complete Synchronization, have

In formula for instantaneous phase.

3, the standard deviation of the non-weak signal section of zero center instantaneous phase nonlinear component absolute value _ap

${\mathrm{\&sigma;}}_{\mathrm{ap}}=\sqrt{\frac{1}{c}\left[\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}{\mathrm{\&phi;}}_{\mathrm{NL}}^2\left(i\right)\right]-{\left[\frac{1}{c}\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}\right|{\mathrm{\&phi;}}_{\mathrm{NL}}\left(i\right)\left|\right]}^2}$

σ _apwith σ _dpdifference be that the former is the standard deviation of phase place absolute value, and the latter is the standard deviation of Direct Phase (non-absolute value phase place).σ _apbeing used for the phase place of characterization signal changes.

4, the standard deviation of normalize and center instantaneous amplitude absolute value _aa

${\mathrm{\&sigma;}}_{\mathrm{aa}}=\sqrt{\frac{1}{N_s}\left[\underset{i=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{a}_{\mathrm{cn}}^2\left(i\right)\right]-{\left[\frac{1}{N_s}\underset{i=1}{\overset{N_s}{\mathrm{\&Sigma;}}}\right|a_{\mathrm{cn}}\left(i\right)\left|\right]}^2}$

5, the standard deviation of the non-weak signal section of normalize and center instantaneous frequency absolute value _af

${\mathrm{\&sigma;}}_{\mathrm{af}}=\sqrt{\frac{1}{c}\left[\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}{f}_N^2\left(i\right)\right]-{\left[\frac{1}{c}\underset{a_n\left(i\right)>a_t}{\mathrm{\&Sigma;}}{f}_N\left(i\right)\right]}^2}$

In formula, f _m(i)=f (i)-m _f, r wherein _sfor the character rate of digital signal, the instantaneous frequency that f (i) is signal.

6, the quadratic sum variance ratio R of signal envelope average _amathematic(al) representation as follows:

$R_a=\frac{{\mathrm{\&mu;}}^2}{{\mathrm{\&sigma;}}^2}$

Wherein, μ represents the average of envelope, σ ²represent the variance of envelope.R _athe envelope of parameter reflected signal rises and falls big or small.

The leader cluster node of step 104, N bunch is sent to Centroid by signal characteristic parameter separately respectively, for Centroid, carries out Classification and Identification.

In described step 104, set up for the parameter of signal characteristic separately of N leader cluster node being passed to the routing relation of Centroid; Each bunch of head, according to described routing relation, determined signal characteristic parameter is delivered to the bang path of upper level leader cluster node, and signal characteristic parameter is sent to described upper level leader cluster node; Described upper level leader cluster node, by determining its bang path, forwards the signal characteristic parameter of receiving, and in repeating process, adds the signal characteristic parameter of oneself, thereby the signal characteristic parameter of N leader cluster node is passed to Centroid.

Further, communication cost between each leader cluster node calculating and other leader cluster node, and utilize described communication cost, set up the local routing information table that comprises one or more upper level leader cluster nodes to be selected, thereby form the routing relation from N leader cluster node to Centroid.Each leader cluster node is according to the communication cost of itself and one or more upper level leader cluster nodes to be selected, calculate the selection probability of described one or more upper level leader cluster nodes to be selected, and according to the size of described selection probability, determine that signal characteristic parameter is delivered to the bang path of upper level leader cluster node.

N leader cluster node can carry out transfer of data according to energy multipath routing protocols, and each leader cluster node, in the process of forwarding information, joins the characteristic parameter of this bunch of collection in parameter set.

In the present invention, leader cluster node is according to selecting probability to select a upper level leader cluster node as route next jump node in local routing information table, and the computing formula of described selection probability is:

$P_{N_j,N_i}=\frac{1/C_{N_j,N_i}}{\underset{k\&Element;F{T}_j}{\mathrm{\&Sigma;}}1/C_{N_j,N_k}}$

for node N _jselect node N _iprobability, be the communication cost of two hops, its computing formula is:

$C_{N_j,N_i}=\mathrm{Cost}\left(N_i\right)+\mathrm{Metric}(N_j,N_i)$

Wherein, Cost (N _i) be defined as the mean value to destination node cost via node in local path information table, that is:

$\mathrm{Cost}\left(N_i\right)=\underset{k\&Element;F{T}_i}{\mathrm{\&Sigma;}}{P}_{N_i,N_j}{C}_{N_i,N_k}$

Metric (N _j, N _i) representing the communication energy consumption between two nodes, computing formula is as follows: Metric (N _j, N _i)=e _ij ^ar _i ^β

E _ij ^arepresent the energy consumption of direct communication between node, R _i ^βrepresent node N _idump energy.α, β are constants, there is no fixing span, and this module has considered energy consumption and two aspects of residue energy of node of nodes transmission.

FT _irepresent leader cluster node N _iby leader cluster node N _jthe condition that adds local routing table.

$F{T}_i=\left\{j\right|C_{N_i,N_j}\&le;\mathrm{\&gamma;}\left(\underset{k}{\min }\left(C_{N_i,N_k}\right)\right)\}$

Wherein, γ is greater than 1 system parameters.

Fig. 2 is a kind of wireless-sensor network distribution type signal characteristic parameter extraction flow chart that the embodiment of the present invention provides, and as shown in Figure 2, step comprises:

1, suppose to have the sensor network of 100 nodes, Centroid is chosen at random 10 monitoring node broadcast feature in network and is extracted instruction, starts characteristic parameter extraction process.

2,10 monitoring nodes receiving described feature extraction instruction are set up request to neighbor node transmission network.

3, each monitoring node and 5 nodes responding at first form one have 6 nodes bunch, positive energy exchange information mutually between bunch interior nodes, chooses the node of energy value maximum as leader cluster node, other 5 nodes are as a bunch member node.

4, bunch in, each node extracts signal characteristic parameter according to state of signal-to-noise, according to priority R _a> σ _af> σ _aa> γ _max> σ _dp> σ _ap, the node that signal to noise ratio is minimum gathers R _a, the node that signal to noise ratio is the highest will gather σ _ap, other by that analogy.All nodes send to leader cluster node by signal characteristic parameter one jumping collecting.In this way can be so that whole recognition correct rate be improved.

After signal characteristic parameter acquisition completes, leader cluster node storage signal characteristic parameter information, a bunch member node enters resting state.

5,10 leader cluster nodes send data to Centroid according to energy multipath routing protocols.

First, each leader cluster node is according to above-mentioned formula set up local routing information table.

Then, each leader cluster node is according to above-mentioned selection new probability formula from local routing information table, select upper level leader cluster node, i.e. down hop route.

Finally, leader cluster node, in the process of Centroid signal transmission characteristic parameter, adds the signal characteristic parameter of self node storage in the signal characteristic parameter forwarding, and is sent to a selected upper level leader cluster node.

6, Centroid is received after the signal characteristic parameter from 10 leader cluster nodes, carries out Classification and Identification, obtains recognition result.

In sum, the present invention has following technique effect:

1, the distributed self-organized clustering of each node in the present invention, and by leader cluster node based on state of signal-to-noise assigned characteristics parameter acquisition task, each node only need extract a characteristic parameter, has reduced to greatest extent computation complexity.

2, the present invention combines the collection of signal characteristic parameter with energy multi-path routing algorithm, has considered communication energy and the dump energy of node, thereby reaches the object of energy efficient.

3, the present invention has considered consumed energy and the dump energy on communication path, and the route of communication also with energy correlation, communication energy consumption can be distributed on mulitpath, thereby realize the steady reduction of network in general energy consumption, extend to greatest extent the life-span of network, the average life span that is about to wireless sensor network extends 40%.

4, the present invention is without global information and clock synchronous.

Although above the present invention is had been described in detail, the invention is not restricted to this, those skilled in the art of the present technique can carry out various modifications according to principle of the present invention.Therefore, all modifications of doing according to the principle of the invention, all should be understood to fall into protection scope of the present invention.

Claims (9)

1. a wireless-sensor network distribution type signal characteristic parameter extracting method, is characterized in that, comprising:

A) Centroid of wireless sensor network sends feature extraction instruction to N monitoring node in described network;

B) each monitoring node is set up one bunch according to described feature extraction instruction, and its each bunch comprises a monitoring node and one or more neighbor node, and using bunch in a node of energy maximum as leader cluster node, other node is as a bunch member node;

C) bunch in, each node gathers respectively different signal characteristic parameters, and by the signal characteristic Parameter storage collecting to leader cluster node;

D) leader cluster node of N bunch is sent to Centroid by signal characteristic parameter separately respectively, for Centroid, carries out Classification and Identification.

2. method according to claim 1, is characterized in that, described steps A) comprising:

Centroid is chosen N monitoring node, and to described N monitoring node, sends feature extraction instruction by broadcast mode.

3. method according to claim 1, is characterized in that, described step B) comprising:

B1) each monitoring node, according to described feature extraction instruction, sends network to its neighbor node and sets up request, and forms one bunch with one or more neighbor nodes of the described network foundation of response request;

B2) each node of described bunch, by mutual positive energy exchange information, obtain having the node of ceiling capacity, and using the described node with ceiling capacity as leader cluster node, other node is as a bunch member node.

4. method according to claim 3, is characterized in that, at described step B1) in, described monitoring node forms one bunch with M node of formerly response.

5. method according to claim 1, is characterized in that, described step C) comprising:

C1) leader cluster node bunch in assigned characteristics extract task;

C2) bunch in, each node exchanges signal to noise ratio information mutually, and according to signal to noise ratio information, gathers corresponding signal characteristic parameter in described feature extraction task.

6. method according to claim 5, is characterized in that, described step C2) comprising:

In bunch, each node, according to the size sequence of its signal to noise ratio, gathers other signal characteristic parameter of respective priority in described feature extraction task.

7. method according to claim 1, is characterized in that, described step D) comprising:

D1) set up for the parameter of signal characteristic separately of N leader cluster node being passed to the routing relation of Centroid;

D2) each bunch of head, according to described routing relation, determined signal characteristic parameter is delivered to the bang path of upper level leader cluster node, and signal characteristic parameter is sent to described upper level leader cluster node;

D3) described upper level leader cluster node, by determining its bang path, forwards the signal characteristic parameter of receiving, and in repeating process, adds the signal characteristic parameter of oneself, thereby the signal characteristic parameter of N leader cluster node is passed to Centroid.

8. method according to claim 7, is characterized in that, described step D1) comprising:

Communication cost between each leader cluster node calculating and other leader cluster node;

Each leader cluster node utilizes described communication cost, sets up the local routing information table that comprises one or more upper level leader cluster nodes to be selected, thereby forms the routing relation from N leader cluster node to Centroid.

9. method according to claim 8, is characterized in that, described step D2) comprising:

Each leader cluster node is according to the communication cost of itself and one or more upper level leader cluster nodes to be selected, calculate the selection probability of described one or more upper level leader cluster nodes to be selected, and according to the size of described selection probability, determine that signal characteristic parameter is delivered to the bang path of upper level leader cluster node.