WO2007043001A2

WO2007043001A2 - Method and apparatus for optimizing a network

Info

Publication number: WO2007043001A2
Application number: PCT/IB2006/053704
Authority: WO
Inventors: Xin Chen; Ningjiang Chen; Willem Fontijn; Qinfeng Zhang
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2005-10-14
Filing date: 2006-10-10
Publication date: 2007-04-19
Also published as: WO2007043001A3

Abstract

The present invention provides a method and apparatus for optimizing a network. The optimizing method comprises determining the dependence relationship between at least one identification result of the network and a node of the network; and setting the priority of the node according to the dependence relationship. With the method of the present invention, the identification rate of the network can be ensured to be as high as possible in the case of insufficient bandwidth; thus, the power resources of the network can be efficiently utilized, and the encryption efficiency of data transmission also can be increased.

Description

METHOD AND APPARATUS FOR OPTIMIZING A NETWORK

FIELD OF THE INVENTION

The present invention relates to a method and apparatus for optimizing a network, and more particularly, to a method and apparatus for optimizing a sensor network.

BACKGROUND OF THE INVENTION

An application of a network, for example, a sensor network, can be widely used in the activity identification field, since it can be used to cooperatively sense, gather, and process the information of sensed objects in the geographical area covered by the network, and provide the information to the viewer.

The existing sensor network faces the following problems. Firstly, the communication ability of the sensor network is quite limited. The communication bandwidth for the sensor of the sensor network is narrow and changes frequently, with a communication-covering range of only tens to hundreds of meters, and if the data transmission exceeds the available bandwidth, a high lost-packet rate is likely to be caused, which will severely influence the identification rate (sensing accuracy) of the sensor network.

Secondly, the power resources of the sensor are quite limited. The information sensing, data processing, and communicating for the sensor network all require a great deal of energy, and sensors in the network are often disabled or discarded due to a lack of power resources; therefore, the restriction of power resources is a severe problem that limits the application of the sensor network.

Thirdly, in many cases, encryption of the data in the sensor network is required to ensure the security of the network. However, the calculation ability of the sensor network is limited, and excessive data encryption will lower the efficiency of the calculation ability. Moreover, the redundancy generated due to data encryption will influence the data transmission rate of the sensor network. Therefore, in view of the above reason, the sensor network must be optimized.

Recently, several methods for optimizing data communication in a sensor network have been provided, for example, in "A Priority-based Multi-path Routing Protocol for Sensor Networks, 15th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2004)", Barcelona, Spain, Sept. 5-8, 2004, Pp. 216-220 by Yuzhe Liu and Winston K. G. Seah, a sensor network data transmission routing protocol capable of saving energy and enhancing the error acceptance of the network is provided, wherein the priority of the sensor node indicates the power resource condition of the data path to the receiver/transmitter, and a path is selected according to the condition. The priority of the sensor node is determined by the total hop amount or the retained power resources. Although the above method optimizes the sensor network to some extent, the method considers only the influence of power consumption on the sensor network, such that the optimization effect is not desirable.

OBJECT AND SUMMARY OF THE INVENTION

The present invention provides a method and apparatus for optimizing a network, which ensures the identification rate of the network is as high as possible while reducing the data transmission amount.

According to an embodiment of the present invention, a method for optimizing a network is provided, which comprises determining the dependence relationship between at least one identification result of the network and a node of the network, and setting the priority of the node according to the dependence relationship.

According to an embodiment of the present invention, a method for optimizing a network is provided, wherein the node is a sensor, and the network is a sensor network.

According to another embodiment of the present invention, an apparatus for optimizing a network is provided, which comprises a determining device for determining the dependence relationship between at least one identification result of the network and a node of the network, and an optimizing device for setting the priority of the node according to the dependence relationship.

According to yet another embodiment of the present invention, a network is provided, which comprises: a plurality of nodes; a network controller connected with the plurality of nodes; wherein the network controller comprises a determining device for determining the dependence relationship between at least one identification result of the network and one node of the plurality of nodes and an optimizing device for setting the priority of the node according to the dependence relationship.

According to the embodiments of the present invention, the priority of the node is determined from the dependence relationship between the identification result in the network and the node, thereby achieving the optimization of the network. With the present invention, the identification rate of the network can be better ensured in case of insufficient bandwidth to optimize the utilization of the bandwidth, and meanwhile, the power resources of the network can be used efficiently, and the data encryption efficiency of nodes with higher priority can be increased as well.

Other objects and achievements of the present invention will become apparent and the present invention will be fully understood through the detailed description of the present invention and the claims with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic structural view of a sensor network according to an embodiment of the present invention;

FIG. 2 is a schematic structural view of a sensor network according to another embodiment of the present invention;

FIG. 3 is a structural view of a sensor network controller according to an embodiment of the present invention;

FIG. 4 is a schematic structural view of a determining device according to an embodiment of the present invention;

FIG. 5 is a flow chart of the identification training process based on a statistical mode and establishing the dependence relationship, according to an embodiment of the present invention;

FIG. 6 is a schematic view of the dependence relationship of a multi-path identification and multi-layer sensor fusion according to an embodiment of the present invention;

FIG. 7 is a flow chart of optimizing a sensor network by utilizing the priority of the sensor according to an embodiment of the present invention.

In all the above drawings, like numbers indicate the same, similar, or corresponding features or functions.

DETAILED DESCRIPTION OF THE INVENTION

The present invention will be described in great detail below with reference to the accompanying drawings.

An embodiment of the present invention provides a scheme for optimizing a sensor network. The priority of each sensor is obtained by determining the dependence relationship between the identification result of the sensor network and the sensors. Then, the sensor network is optimized according to the priority of each sensor, thereby better ensuring the identification rate of the sensor network in the case of lower bandwidth, and optimizing the power management and bandwidth utilization, as well as increasing the encryption efficiency.

FIG. 1 is a schematic structural view of a sensor network according to an embodiment of the present invention. The sensor network comprises a sensor network controller 110, a plurality of wireless sensor nodes 120, and a plurality of wired sensor nodes 130.

The sensors 120 and 130 are wired or wirelessly connected to the sensor network controller 110, wherein the sensor network controller 110 is used for collecting data, with a function similar to a base station in a wireless communication network, or a router in a wired communication network, or a combination of the two. The wired communication between each sensor 120, 130, and the sensor network controller 110 can be achieved through, for example, a local area network (LAN), an exclusive communication circuit, and the like. The wireless communication therebetween can be achieved through, for example, a wireless local area network (WLAN) standard, a Bluetooth standard, and the like.

Each sensor node 120, 130 transmits the information such as the detection result to the sensor network controller 110, for example, the wired transmission indicated by a solid line in the figure. Meanwhile, the sensor network controller 110 can also send a control signal to each sensor 120, 130, for example, the wired transmission indicated by the broken line in the figure, to optimize the sensors 120, 130. The optimization process comprises determining whether or not transmit the information through a certain sensor, adjusting the sampling speed of the sensor, determining whether or not to encrypt the data of the sensor, and the like. It can be seen from FIG. 1 that the sensor network controller 110 is a rendezvous point of all detection data, a processing center for processing the detection data to generate a detection result, and a control center for optimizing the network.

FIG. 2 is a schematic structural view of a sensor network according to another embodiment of the present invention. In the present embodiment, a plurality of sensors 120 and 130 is wired or wirelessly connected to a data concentrator 210, thus forming a sensor network 220. Each sensor network 220, 230, and 240 is connected to a sensor network controller 250 through the respective data concentrators. The sensor network controller 250 gathers sensor data from each sensor network 220, 230, and 240, and also sends a control command to each sensor network 220, 230, and 240, so as to carry out various optimization processes in each sensor 120, 130 in the network through a data concentrator, for example, the data concentrator 210.

FIG. 3 is a schematic view of a network controller of the above embodiment. The network controller is a sensor network controller, which comprises a determining device 310, an optimizing device 320, and a transmission controlling device 330. The determining device 310 can be an identification device that can be trained and used for gathering sensor data or training samples. The dependence relationship between the identification result and each sensor in the sensor network can be obtained by analyzing the received training samples. The specific method for determining priority according to the dependence relationship will be described below in detail. The user can also adjust the determining device 310 by inputting the user data. For example, the user can determine the priority of each sensor by inputting the data related to the dependence relationship between the identification result and each sensor.

The so-called "dependence relationship" means the corresponding relationship between the data or data range of the plurality of sensors and the identification result, that is, when the data of one or more sensors is a fixed value or falls within a fixed data range, if the probability of the occurrence of a certain identification result is larger than a predetermined threshold value, it is determined that a corresponding relationship exists between the identification result and the sensor(s). The sum of all corresponding relationships between all sensors and all identification results is the dependence relationship between the plurality of sensors and all identification results. Then, the priority of each sensor is set according to the existence or magnitude of the dependence relationship.

After the training, the optimizing device 320 optimizes the power, bandwidth, encryption, and the like of the sensor network through a transmission controlling device 330, according to the priority of each sensor determined by the determining device 310, with reference to the broken line in the figure. After the optimization, the transmission controlling device 330 gathers the real time data of the sensor in a wireless or wired manner according to the optimized encryption or/and transmission manner, and sends the data to the determining device 310 to identify the real time data using the determining device 310.

FIG. 4 is a schematic structural view of a determining device based on the statistical learning algorithm according to an embodiment of the present invention. The determining device 310 comprises a preprocessing device 410, a feature extraction/selection device 420, a calculation classification device 430, and a feedback device 440.

The preprocessing device 410 receives the detection data and pre-processes the original detection data, for example, filtering the noises such as burst error or reducing the data dimension through filtering.

The feature extraction/selection device 420 receives the pre-processed detection data, and extracts the features of the detection data, such as sound strength, color depth, and pressure. When there is a large amount of features, the features are selectively extracted or the extracted features are selected depending on the activity to be identified.

The calculation classification device 430 has two functions. One of the functions is that, if the sample input to the determining device is an activity (i.e., identification result) known sample during the training, the calculation classification device 430 functions to determine the parameters of a classification algorithm, such as the threshold condition in the linear classifier, according to the sample data, such that the classification algorithm can act as a measurement for measuring the known activity. For example, if the magnitude of the pressure sensor on an office chair falls within a certain range, e.g., 40-80 kg, the probability that someone is sitting on the chair is larger than 80%. According to the result of the classification algorithm training, the dependence relationship between at least one activity or a sub-pattern of an activity and the node can be determined. After the sensor data is pre-processed and the features are extracted and classified, a training result is obtained by establishing a classification algorithm model, and the obtained training result indicates the dependence relationship between the input sensor data and at least one activity, i.e., identification result, for example, the dependence relationship between the pressure sensor on the chair and the user's activity of sitting down. The other function of the calculation classification device 430 is classifying the real time sensor data according to the classification algorithm, and after the real time sensor data is pre-processed, and the features are extracted and classified, an identification result is obtained through a trained classification algorithm model.

According to an embodiment of the present invention, the calculation classification device 430 is further used for calculating a weight of the node with respect to at least one identification result, wherein the weight relates to the number of all nodes corresponding to the identification result, thereby determining the dependence relationship between the at least one identification result and the node according to the weight, which will be described in detail with reference to FIG. 6.

The feedback device 440 is used to adjust the preprocessing device 410, the feature extraction/selection device 420, or the calculation classification device 430 according to the user's input, for example, adjusting the parameters or preprocessing manner so as to remove more noises or accidental signals; changing the selection of features and the extracting manner; and changing the method for calculating the classification algorithm parameters so as to increase the classification accuracy.

According to an embodiment of the present invention, each function module 410, 420, 430, and 440 of the determining device 310 can also be distributed and embedded into the sensor network, i.e., the preprocessing, feature calculation, and classification are carried out by different sensor nodes, provided that different sensor nodes can communicate with each other.

According to another embodiment of the present invention, the determining device 310 also can directly receive the information related to the dependence relationship manually inputted by the user.

Generally, a simple activity can be determined directly by sensing with the sensor and gathering the specific information of the ambience. However, for a complicated activity, the activity to be identified is required to be divided into a plurality of sub-patterns for identification. For example, for determining an activity in which a person sits down on a chair, the activity can be divided into two sub-patterns: the chair has an object fallen thereon, and the height of the person is reduced. The chair having an object fallen thereon can be identified by detecting with a light sensor whether or not the light on the chair surface is shielded and by detecting the pressure applied to the chair surface with a pressure sensor. The reduction of the height of the person can be identified from the changes of the light sensor or a position sensor. According to the data sensed and gathered by each sensor, it can be determined whether or not a corresponding sub-pattern appears, and the corresponding activity can be determined by combining several corresponding sub-patterns together. Therefore, the dependence relationship between each sensor and the activity, i.e., the identification result, can be obtained through the determining device. The priority of each sensor can be determined according to the dependence relationship between each sensor and the activity i.e., the identification result, and thereby the sensor network can be optimized.

According to an embodiment of the present invention, the physical meanings of the parameters detected by each sensor in the sensor network are known. The dependence relationship between the sensor data and the activity can be obtained according to the logic relationship between the physical meaning of the detected parameters and the identification result or a sub-pattern of the identification result. The training samples are inputted into the determining device 310 for checking and adjusting the dependence relationship and determining the classification condition, and then the sensor network can be optimized according to the classification condition. This optimizing method is suitable for a sensor network where the sensor node directly corresponds to the activity to be detected.

According to another embodiment of the present invention, in case the sensor does not directly correspond to the activity to be detected, the dependence relationship between the identification result and the sensor can be obtained according to the influence of the sensor data to the identification result, to determine the priority of the sensor. For example, when identifying a determined activity, by deleting data from one sensor or a set of sensors, the influence of the data on the identification result can be obtained: if the identification rate is significantly reduced, the identification result is highly dependent upon the sensor, and the priority of the sensor should be higher; otherwise, the identification result is slightly dependent upon the sensor, and the priority of the sensor should be lower. According to the influence of the data on the identification rate, the priority of each sensor can be quantified. For example, if the data of a sensor is deleted, the correct identification result cannot be obtained, and then the priority of the sensor is 1; if the data of a sensor is deleted, the probability for obtaining the correct identification result is 20%, and then the priority of the sensor is set to 0.8 (1-20% = 0.8), and so forth.

As for a large- size sensor network with a large amount of sensor nodes, the above two methods are not suitable since there are an excessive number of sensors. According to another embodiment of the present invention, an identification model is established in a training system through an identification algorithm in the training system, such as a Decision Tree, according to a certain amount of training samples, and the identification model is modified and improved with the feedback information of the user, such that whether or not the sensor data is related to the activity sub-pattern or the activity can be determined by the trained system, thereby establishing the dependence relationship between the sensor and the activity.

According to the data gathered by the sensor, many statistical model identification technologies can be used for identifying the activity, for example, by using a Bayesian Method, including Bayesian Classification, Bayesian Network, or by using the Hidden Markov Model (HMM), and the like.

FIG. 5 shows a flow chart for an identification training process based on a statistical model that establishes the dependence relationship between the sensor and the identification result according to an embodiment of the present invention. First, in step S510, one or more training samples are gathered and received; in step S520, the training samples are preprocessed, including filtering the noises, burst errors, and the like in the sample data, so as to remove some interference signals. Several filtering methods can be used, such as inversion adaptive filtering, bandwidth noise adaptive filtering, and impulse noise filtering, etc.

In step S530, the features of the preprocessed training samples are extracted and selected. The sensor data are gathered through different types of sensors. If the sensor data corresponding to each activity is clustering in a separated area, i.e., directly related to the activity, the data can be directly used for identifying the activity, for example, sensor 7 directly corresponds to activity 3 as shown in FIG. 6.

However, the sensor data is in fact not desirable, and the data clustering areas may overlap each other. Therefore, the data need to be transmitted to another feature space by extracting the features, thereby separating the overlapped data areas. For example, voice signals can be transformed from a time field to a frequency field through Fast Fourier Transformation, and the features of the frequency field are extracted. Generally, the features of the sensor data are extracted as much as possible, so as to obtain the feature most suitable for classification and identification. However, if there is an excessively large amount of extracted features, they can be selected according to the characteristics of the practical identification activity.

In step S540, the training process is carried out with the classification algorithm. The classification algorithm training takes all the features or a part of the features of all sensors as an input, and obtains the training result by establishing a classification algorithm model. The obtained training result indicates the dependence relationship between the input sensor data and at least one activity, i.e., the identification result.

The features and classification model employed by the classification algorithm obtained in step S540 do not always meet the requirements of the identification rate, such that the identification rate is checked in step S550 to see whether it meets the requirements or not. If the identification rate does not meet the requirement, in step S560, one or more of the preprocessing, feature extracting/selecting, and classification training is adjusted, i.e., returning to step S520, S530, or S540, until the features and classification model that meet the identification rate requirements are found out.

In step 570, the dependence relationship between all sensors and at least one identification result can be obtained according to all of the above training results, and the dependence relationship between the sensor and the identification result can be indicated as a dependence relationship diagram as shown in FIG. 6.

Finally, in step 580, the priority of each sensor can be set according to the dependence relationship between all sensors and at least one identification result, which is described in great detail with reference to FIG. 6.

FIG. 6 shows a schematic view of a dependence relationship of a multi-path identification and multi-layer sensor fusion. As shown in the figure, an activity can be composed of a plurality of sub-patterns. Activity 1 consists of sub-pattern 1 and sub- pattern 2; and activity 2 consists of sub-pattern 2 and sub-pattern 3. Activity 3 is directly identified by sensor 7. The sub-patterns are directly identified by the sensor data. The dependence relationship between the sub-pattern and the activity can be determined by the logic relationship or can be further determined by the training process.

Since a single sensor can hardly ensure identification accuracy, each sub-pattern should be identified by sensors with a plurality of paths in order to enhance the robustness of the identification, and each path is referred to as an identification path. For example, in the present embodiment, sub-pattern 1 is identified by the data of sensors 1 and 2; sub- pattern 2 is identified by the data of sensors 1, 2, and 3; and sub-pattern 3 is identified by the data of sensors 1, 3, 4, and 6. The data fusion can be used to process the data originally gathered by the sensors of the sensor network, and the sensor data with the same properties are fused to enhance the reliability of the data.

The priority of the sensor can be determined by using the dependence relationship diagram. According to an embodiment of the present invention, the priority of the sensor data can be determined by the number of sub-patterns using the sensor data as an identification path and by the importance of the sensor data for identifying the sub-pattern. For example, to simplify the calculation, it is assumed that the importance of all sub- patterns for the identification result is the same, and the importance of all sensor data for identifying the sub-patterns is the same.

As shown in FIG. 6, the weight marked beside each sloped line between a sensor and a sub-pattern indicates the importance of the sensor for identifying the corresponding sub- pattern. For example, sensors 1 and 2 are both used for identifying the sub-pattern 1, thus the weights of them for sub-pattern 1 are 1/2; sensors 1, 2, and 3 are all used for identifying the sub-pattern 2, thus the weights of them for sub-pattern 2 are all 1/3; and similarly, the weights of sensors 1, 3, 4, and 6 for identifying sub-pattern 3 are all 1/4. Of course, in a practical application, the importance of each sensor can be different, and if there is a significant contrast between the influences of two identification paths as for the parent identification result, the identification path with the smaller influence can even be neglected.

The priority of the sensor data is obtained by adding the importance i.e., the weight of the sensor data for the identification result. For example, sensor 1 influences the identification of sub-patterns 1, 2, and 3, thus its priority is 1/2+1/3+1/4 = 13/12; sensor 2 only influences the identification of sub-patterns 1 and 2, thus its priority is 1/2+1/3=5/6; obviously, the priority of sensor 1 is larger than that of sensor 2. As for sensors 2 and 3, they both influence the identification of two sub-patterns, i.e., sensor 2 influences sub- patterns 1 and 2, and sensor 3 influences sub-patterns 2 and 3; however, since the identification path of sub-pattern 3 is much more than that of sub-pattern 1, the priority of sensor 3 is 1/3+1/4=7/12, which is less than that of sensor 2; similarly, the priority of other sensors can be obtained. The importance of each sensor to at least one identification result can be conveniently compared according to the priority of each sensor, such that the whole sensor network can be more desirably optimized.

It should be appreciated by those skilled in the art that the dependence relationship provided in the present invention as shown in FIG. 6 also can be directly set by the user when the sensor network is constructed, without the identification training process as described in FIG. 5.

FIG. 7 shows a flow chart for optimizing a sensor network by using the priority of the sensors according to an embodiment of the present invention. First, in step S710, the priority of each sensor is determined according to the above-mentioned method. Next, each performance of the sensor network is optimized. The optimizing steps described below can be carried out sequentially, or simultaneously.

In step S720, since the power resources of the sensor power are limited, it is determined whether or not the power management strategy of the sensor network is required to be optimized; if so, in step 730, more detection data are transmitted through the sensor nodes with lower sensor priority, thereby saving the power utility of the sensors with higher priority, prolonging their service life; otherwise, power optimization will be neglected.

In step S740, when the bandwidth is limited, it is determined whether or not the bandwidth utility is required to be optimized. If the bandwidth optimization is required, in step 750, the sensor data with lower priority can be discarded, or the sampling rate of the sensor with lower priority is reduced; thus, the damage to the activity identification result can be reduced as much as possible.

In step S760, it is determined whether or not the encryption efficiency should be enhanced when encrypting the data in the network; if so, in step 770, only the sensor data with higher priority are encrypted; thus, the addition of redundant data can be reduced and the encryption efficiency of the sensor network can be improved, while ensuring the network security. After all optimization processes of the sensor network have been finished, the whole flow chart ends.

It should be understood by those skilled in the art that the method and apparatus for optimizing a sensor network provided in the present invention not only can be applied in the sensor network, various modifications without departing from the content of the present invention can be made as well, so as to be applied in other types of networks, such as Bluetooth piconet. Through the principle of the present invention, the dependence relationship between an identification result of the network and at least one node can be determined, wherein the at least one node is included in the network, and then the priority of the at least one node can be set according to the dependence relationship.

The technical content and characteristics of the present invention have been disclosed above; however, various alternations and modifications without departing from the spirit of the present invention can still be made by those skilled in the art based on the teaching and disclosure of the present invention. Therefore, the protective scope of the present invention is not limited to the disclosure of the present invention, but includes various alternations and modifications without departing from the present invention, and is contemplated by the claims.

Claims

CLAIM:

1 . A method for optimizing a network, comprising:

(a) determining the dependence relationship between at least one identification result of the network and a node of the network;

(b) setting the priority of the node according to the dependence relationship.

2 . The method as claimed in Claim 1, wherein the node is a sensor, and the network is a sensor network.

3 . The method as claimed in Claim 1, wherein step (a) comprises determining the dependence relationship between the identification result and the node according to a logic relationship between a physical meaning of a piece of data of the node and the at least one identification result.

4 . The method as claimed in Claim 1, wherein step (a) comprises determining the dependence relationship between the identification result and the node according to a logical relationship between a physical meaning of a piece of data of the node and a sub- pattern of the at least one identification result.

5 . The method as claimed in Claim 1, wherein step (a) comprises: training a classification algorithm model according to a set of training samples of all nodes in the network; determining the dependence relationship between the at least one identification result and the node according to the training result.

6 . The method as claimed in Claim 1, wherein step (a) comprises: calculating a weight of the node corresponding to the at least one identification result, wherein the weight relates to the number of all nodes corresponding to the identification result; determining the dependence relationship between the at least one identification result and the node according to the weight.

7 . An apparatus for optimizing a network, comprising: a determining device, for determining the dependence relationship between at least one identification result of the network and a node of the network; an optimizing device, for setting the priority of the node according to the dependence relationship.

8 . The apparatus as claimed in Claim 7, wherein the node is a sensor, and the network is a sensor network.

9 . The apparatus as claimed in Claim 7, wherein the determining device is used for determining the dependence relationship between the identification result and the node according to a logic relationship between a physical meaning of a piece of data of the node and the at least one identification result.

10 . The apparatus as claimed in Claim 7, wherein the determining device is used for determining the dependence relationship between the at least one identification result and the node according to a logic relationship between the physical meaning of a piece of data of the node and a sub-pattern of the at least one identification result.

11 . The apparatus as claimed in Claim 7, wherein the determining device comprises: a calculation classification device, for training a classification algorithm model according to a set of training samples of all nodes in the network, thereby determining the dependence relationship between the at least one identification result and the node according to the training result.

12 . The apparatus as claimed in Claim 7, wherein the determining device comprises: a calculation classification device, for calculating a weight of the node corresponding to the at least one identification result, wherein the weight relates to the number of all nodes corresponding to the identification result, thereby determining the dependence relationship between the at least one identification result and the node according to the weight.

13 . A network, comprising: a plurality of nodes; a network controller connected with the plurality of nodes; wherein, the network controller comprises: a determining device, for determining the dependence relationship between at least one identification result of the network and one node of the plurality of nodes; an optimizing device, for setting the priority of the node according to the dependence relationship.