JP5923856B2 - Sensor network control device and control method - Google Patents

Description

  The present invention relates to a sensor network that arranges a plurality of sensor nodes and measures data.

  Sensor networks that can be used for weather observation, security crime prevention, building management, energy management, water quality surveys, etc. are attracting attention. A sensor network is a system that arranges a plurality of sensor nodes and measures data (see Patent Documents 1 to 7).

  In general, a sensor node of a sensor network operates with a battery. In that case, since the lifetime of the sensor node is determined by the life of the battery, power saving of each sensor node is a major issue.

  It is known that the process with the largest power consumption in the sensor node is a network process with communication. For this reason, various attempts have been made to reduce network processing. In particular, in wireless sensor networks (WSNs: Wireless Sensor Networks), power consumption is about the same between signal reception and transmission.

  In order to save the power of the sensor nodes, for example, there is a method of putting each node into a dormant state and operating only the minimum necessary nodes that cover the sensor area (Span of MIT (Massas Institutes of Technology)). There is also a routing method (LEACH: Low-Energy Adaptive Clustering Hierarchy) that reduces the communication amount itself.

JP 2004-226157 A JP 2006-190247 A JP 2010-245725 A JP-T-2005-525061 Special table 2008-532460 gazette Special table 2009-529187 Special table 2010-506480

  However, if only a minimum number of nodes are operated in order to extend the life of the sensor node, there is a risk that important sense data may be missed. In addition, when trying to record important data without leaking, there is a trade-off that the number of sensor nodes to be operated increases.

  For example, in Span, the network is autonomously configured by selecting the smallest number of sensor nodes that can cover an area. The network is configured mainly depending on the network status. As a result, there is an effect of reducing the power consumption of the sensor node and extending the life of the network. However, since the characteristics of sensor data are not necessarily reflected in the network configuration, there is a risk that characteristic data may be missed in an environment where measured sensor data is spatially or temporally biased.

  For example, assume a sensor network along a road. When the cliff facing the road is about to collapse, it is considered that sensor nodes that can detect the event are unevenly distributed in a specific part.

  If the sampling interval of the sensor node is shortened or the number of active nodes is increased to increase the total operation time of the sensor node in the sensor network, such an event can be captured with high accuracy. However, increasing the total operating time increases the power consumption of the sensor node and shortens the life of the sensor node. As a result, the lifetime of the entire network is shortened.

  In contrast, events that the user wants to observe on the sensor network are defined in advance, the sensor node utilization rate is increased around the sensor node where the event is measured, and the event changes are tracked sensitively. There is a method to do. This method is effective when a query from a user is clarified in advance. However, an event cannot be defined in advance if the query from the user is not clear in advance. For example, when the measured sensor data is to be mined later, the sensor network is shared by multiple users, and various events are assumed. It is difficult to set an event in advance, and it is difficult to obtain an expected result.

  As another system, there is a system in which highly spatially and temporally related data acquired by a plurality of nearby sensor nodes is processed by in-network processing and sensor nodes to be operated are aggregated. According to this method, the total communication amount in the sensor network is reduced, and power consumption is reduced. However, since the network configuration does not link with environmental features, the accuracy of the obtained data is not improved. Rather, by consolidating the sensor nodes, it is difficult to grasp the data distribution in the environment, and the operation time of the sensor nodes may not be reduced.

  An object of the present invention is to provide a technology that enables both power saving of a sensor network and high-precision data measurement.

In order to achieve the above object, the sensor network control device of the present invention provides:
Features of the sensor data in the area based on unique information including information indicating respective positions of the plurality of sensor nodes arranged in a predetermined area and sensor data measured in each of the plurality of sensor nodes A feature calculation unit for calculating
And a network control unit that controls the plurality of sensor nodes based on the characteristics.

The sensor network control method of the present invention includes:
Features of the sensor data in the area based on unique information including information indicating respective positions of the plurality of sensor nodes arranged in a predetermined area and sensor data measured in each of the plurality of sensor nodes To calculate
The plurality of sensor nodes are controlled based on the characteristics.

  According to the present invention, it is possible to save power in the sensor network while maintaining high accuracy of data measurement of the sensor network.

It is a block diagram which shows the structure of the sensor network of this embodiment. It is a block diagram which shows the structure of the sensor node of this embodiment. It is a block diagram which shows the structure of the sensor network by a 1st Example. It is a block diagram which shows the structure of the sensor network in a 3rd Example.

  An embodiment according to a basic configuration of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram illustrating a configuration of a sensor network according to the present embodiment. FIG. 2 is a block diagram showing the configuration of the sensor node of this embodiment. Although FIG. 2 shows the configuration of the sensor node 1, the sensor nodes 2 and 3 basically have the same configuration.

  Referring to FIG. 1, the sensor network of the present embodiment has a plurality (three in this case) of sensor nodes 1 to 3 that are interconnected. In this embodiment, as an example, a cluster is formed by a plurality of sensor nodes by clustering, and one of the sensor nodes controls another sensor node as a cluster head.

  Any sensor node 1-3 can be a cluster head.

  A sensor node that is not a cluster head notifies the other sensor nodes in the surrounding area of its own unique information and its measured sensor data.

  On the other hand, the sensor node that has become the cluster head includes unique information including information indicating the positions of the sensor nodes 1 to 3, including information notified from a plurality of sensor nodes in the peripheral area, and a plurality of sensors. Based on the sensor data measured at each of the nodes 1 to 3, the characteristics of the sensor data in the area are calculated. And the sensor node of a cluster head controls the several sensor nodes 1-3 based on the characteristic.

  Referring to FIG. 2, the sensor node 1 includes a sensor 11 and a control unit 15. Here, attention is paid to the case where the sensor node 1 operates as a cluster head, and the control unit 15 includes a feature calculation unit 17 and a network control unit 18.

  The feature calculation unit 17 is measured by specific information including information indicating the positions of the plurality of sensor nodes 1 to 3 arranged in the peripheral area forming the cluster, and each of the plurality of sensor nodes 1 to 3. Based on the sensor data, the characteristics of the sensor data in the area are calculated.

  Specifically, the feature calculation unit 17 calculates at least one of a spatial gradient and a temporal gradient of the sensor data in the area as a feature. This gradient indicates a spatial or temporal change in sensor data, and a large absolute value of the gradient means that the sensor data changes drastically. At this time, the feature calculation unit 17 may form a scalar field by interpolating between sensor data in the area by approximation, and may calculate a gradient in the scalar field as a feature.

  The network control unit 18 controls the plurality of sensor nodes 1 to 3 based on the feature calculated by the feature calculation unit 17.

  As a specific example, the network control unit 18 calculates the respective schedules of the sensor nodes 1 to 3 based on the characteristics, and operates the sensor nodes 1 to 3 according to the calculated schedule. The schedule includes timing information for operating or pausing the sensor node by TDMA (Time Division Multiple Access). For example, the schedule defines the sampling timing for executing the measurement of each of the sensor nodes 1 to 3, and the sensor nodes 1 to 3 operate only at the time of sampling and are stopped at other times.

  As described above, here, as an example, the feature calculated by the feature calculation unit 17 is the gradient of the sensor data in the area. The network control unit 18 calculates the schedule so that the sensor node located at the position where the absolute value of the gradient is large operates at a high operation rate.

  As another specific example, the network control unit 18 may move the sensor nodes 1 to 3 based on the feature calculated by the feature calculation unit 17. Again, as an example, it is assumed that the feature is the gradient of sensor data in that area. And the network control part 18 moves the sensor nodes 1-3 so that the density of a sensor node becomes high, so that the absolute value of a gradient is large.

  As described above, according to the present embodiment, the plurality of sensor nodes 1 to 3 are controlled based on the characteristics of the sensor data calculated from the respective positions and sensor data of the plurality of sensor nodes 1 to 3. The power saving of the sensor network can be achieved while keeping the accuracy of the data measurement of the sensor network high.

  Subsequently, a more specific embodiment will be described.

(First embodiment)
FIG. 3 is a block diagram showing the configuration of the sensor network according to the first embodiment. Referring to FIG. 3, the sensor network is composed of a plurality (three in this case) of sensor nodes 1 to 3. The sensor nodes 1 to 3 can communicate with each other via the communication medium 4. The communication medium 4 is, for example, a wireless line or a wired line.

  The sensor node 1 includes a sensor 11, a communication module 12, a power supply 13, a unique information generation unit 14, and a control unit 15. The sensor nodes 2 and 3 have basically the same configuration as the sensor node 1.

  The sensor 11 measures some item with respect to the external environment and sends the value of the measurement result to the control unit 15. There are various measurement items depending on the application of the sensor network. Examples of sensor network applications include weather observation, security crime prevention, building management, energy management, and water quality surveys. The temperature, light, color, sound, vibration, moving object, etc. used for those applications can be the measurement object by the sensor 11.

  The communication module 12 performs communication between the sensor nodes according to an instruction from the control unit 15 and acquires information regarding the other sensor nodes 2 and 3. Alternatively, information on the own node is given to the other sensor nodes 2 and 3.

  The power supply 13 supplies power to each part of the sensor node 1.

  The unique information generation unit 14 is a physical ID of the own node obtained by using an ID unique to the own node for identifying each sensor node, position information indicating a logical position of the sensor node, GPS (Global Positioning System), and the like. Position information indicating the position is generated and sent to the control unit 15 as unique information of the own node.

  The control unit 15 is a control unit realized by a processor executing a program. The control unit 15 communicates with the nearby sensor nodes 2 and 3 via the communication module 12, and adjusts the operation timing of the nearby sensor nodes 2 and 3 and the own node by executing the process of In-network processing. . Thereafter, the control unit 15 controls the power supply 13 to shift the own node to the standby mode, and maintains the standby mode until the next operation timing.

  Next, the operation of the sensor nodes 1 to 3 in the first embodiment will be described in detail.

  The sensor network of the present embodiment is based on LEACH (Low Energy Adaptive Clustering Hierarchy), which is an existing technology.

  In LEACH, an appropriate cluster head ratio p in a certain area is determined. The cluster head ratio is a value that determines how much of the sensor nodes are to be cluster heads. Each of the sensor nodes 1 to 3 performs the following processing as one “round” and repeats the processing.

  First, the control unit 15 of each sensor node generates a random number k (0 <k <1) by itself and compares the candidate value T (p) that can be calculated from p with k. If k <T (p), the sensor node becomes a cluster head, and advertises itself to nearby sensor nodes.

  Next, a sensor node that is not a cluster head issues a cluster participation request to a nearby cluster head.

  After that, the cluster head performs TDMA (Time Division Multiple Access) scheduling of the operation timing of each neighboring sensor node, and adjusts the operating time of each sensor node by notifying each neighboring sensor node of the scheduling result information. To do.

  When a certain period of time elapses, this round is completed, and as the next round, the random number k is generated again and the cluster head is extracted.

  In the present embodiment, a method for requesting participation in a cluster and a method for setting TDMA scheduling are different from those of a normal LEACH. Hereinafter, the operation in the present embodiment will be described in detail.

  Here, it is assumed that the sensor node 1 is selected as the cluster head in the sense network of FIG. Assume that the sensor node 2 in the vicinity of the sensor node 1 makes a cluster participation request to the sensor node 1.

  The control unit 25 of the sensor node 2 acquires the latest sensor data Dp of the sensor 21 and the unique information unique to the sensor node 2 generated by the unique information generation unit 24, and uses the communication module 22 to acquire the sensor node 1. This information is transmitted. The sensor node 1 receives similar information from a plurality of nearby sensor nodes i.

  The control unit 15 of the sensor node 1 that has become the cluster head determines, based on the obtained unique information, the position Li of the nearby sensor node i by two-dimensional coordinates (xi, yi) or three-dimensional coordinates (xi, yi, zi) is mapped to an area (sensor area) covered by the sensor network.

  Next, the control unit 15 maps the sensor data Dpi collected from each sensor node i as a scalar value for the position of each sensor node i in the sensor area. Further, the control unit 15 interpolates between the sensor data by approximation to form a scalar field Dp = φ (L).

  In addition, when mapping the position of the sensor node i on the sensor area, the control unit 15 does not necessarily use the physical position of the sensor node i, but the virtual position calculated using the radio wave intensity or the like is used. May be used.

  Subsequently, the control unit 15 calculates the gradient at each position of the scalar field Dp = φ (L) formed on the sensor area. Then, the control unit 15 sets TDMA scheduling of the sensor node that has requested cluster participation based on the absolute value | φ (L) | of the gradient, and adjusts the operation rate.

  For example, when the absolute value of the gradient at the position of the sensor node is greater than or equal to a certain value, the control unit 15 sets the sampling interval in the scheduling of the sensor node to the shortest value. Conversely, the control unit 15 sets the sampling interval to a time longer than the shortest value in the scheduling of the sensor node at a position where the absolute value of the gradient is equal to or less than a certain value. For example, a sampling interval that is twice or three times the shortest value may be set. As a result, the sensor node can be kept in a dormant state until the next round of cluster head selection.

  A small absolute value of the gradient of the sensor data at a certain position means that the position is in an area having close values in all directions. Therefore, in the present embodiment, the lifetime of the entire sensor network can be extended by increasing the sampling interval of sensor data at such positions and shortening the operation time of the sensor node. Conversely, in a position or area where the absolute value of the gradient of the sensor data is large and the change of the sensor data is steep, sampling is performed at a short sampling interval to obtain detailed sensor data. In this way, in this embodiment, it is possible to achieve both power saving of the sensor network and highly accurate data measurement.

  Further, in this embodiment, feature information obtained by processing sensor data by in-network processing is used for the sensor network configuration, so that power can be efficiently and efficiently saved according to the environment to be observed regardless of the sensor node density. To extend the life of the sensor network.

  In addition, since characteristic areas are intensively measured from characteristic information obtained by in-network processing, the quality of sensor data acquired by the sensor network is improved.

  Moreover, since the measurement of the area having similar sensor data is thinned out from the feature information obtained by in-network processing, the amount of communication data of the sensor network can be efficiently reduced.

  In addition, since data aggregation that assumes a specific query is not performed in in-network processing, the amount of communication data in the sensor network is reduced, and a sensor with the same accuracy for a query that is not assumed in advance. Data can be provided.

  Moreover, since the total amount of sensor data is reduced by thinning out similar data from the feature information obtained by in-network processing, the response to the query to the sensor network can be speeded up.

  Even if there is no sensor node at the query position, the estimated value of the sensor data is returned from the gradient obtained by interpolating between the sensor data by In-network processing, so that the query to the specific position of the sensor network is returned. The accuracy of sensor data in response can be improved.

(Second embodiment)
The basic configuration of the sensor network in the second embodiment is the same as that of the first embodiment shown in FIG.

  In the first embodiment, each sensor node sends the latest sensor data Dp when sending a cluster join request to a nearby cluster head, and the cluster head is based on the spatial characteristics of the latest sensor data of each sensor node. Scheduling. On the other hand, in the second embodiment, each sensor node sends not only the latest sensor data but also sensor data within a certain period from the past to the present to the cluster head, and the cluster head performs a certain period of time for each sensor node. Scheduling based on spatial and spatial characteristics. As a result, the TDMA schedule can be determined in consideration of the gradient of temporal change.

  In the second embodiment, first, sensor data Dt1, Dt2 within a certain period (t1, t2,..., Tn) from the past to the present are detected by the sensor nodes that are cluster heads. , ..., Dtn is sent.

  The control unit 15 of the sensor node 1 serving as the cluster head determines the position Li and the time tj of the neighboring sensor node i based on the unique information obtained from the neighboring sensor node and the time when the sensor data is acquired. The three-dimensional coordinates including the time axis (xi, yi, tj) or the four-dimensional coordinates (xi, yi, zi, tj) are mapped to the sensor area.

  Next, the control unit 15 maps the sensor data Dt1, Dt2,..., Dtn measured at each sensor node as a scalar value with respect to the position and time of each sensor node i in the sensor area. Further, the control unit 15 interpolates between the sensor data by approximation to form a scalar field.

  In addition, when mapping the position of the sensor node i on the sensor area, the control unit 15 does not necessarily use the physical position of the sensor node i, but the virtual position calculated using the radio wave intensity or the like is used. May be used.

  In the second embodiment, since time is taken into account in the scalar field, an efficient sensor network can be configured by extracting features with respect to time changes. For example, if the sensor data spatial change in the entire area managed by a certain cluster head is small but suddenly changes in time, the sampling interval in the TDMA schedule can be shortened to obtain detailed measurement data. It becomes possible.

  Further, in the second embodiment, using the correlation between the spatial change due to the gradient and the temporal change, the movement of the rapidly changing area (hot spot) is predicted, and scheduling is performed so as to anticipate the movement. Is also possible. As a result, it is possible to follow a hot spot that moves faster.

(Third embodiment)
FIG. 4 is a block diagram showing the configuration of the sensor network in the third embodiment. Referring to FIG. 4, the third embodiment differs from the first embodiment shown in FIG. 3 in that the sensor nodes 1 and 2 have moving mechanisms 16 and 26.

  In the first and second embodiments, the sensor node that becomes the cluster head sets a short sampling interval in the schedule in an area where the absolute value of the gradient of the sensor data is large. For this reason, when an area (hot spot) that changes rapidly exists spatially and fixedly in a certain part, there is a possibility that the life of the sensor node in that part may be exhausted faster than the other part.

  In the third embodiment, instead of shortening the sampling interval of the schedule of the sensor node in the vicinity of the hot spot, the sensor node that has become the cluster head instructs the movement to the sensor node in the vicinity of the hot spot. Increase sensor node density. Thereby, the sensor data of the hot spot can be measured in detail without shortening only the lifetime of the sensor node near the hot spot.

  In the third embodiment, since the sensor node moves autonomously according to the characteristics of the sensor data, the sensor node is automatically arranged optimally in an environment in which changes are spatially biased. Can do. For example, when sensor data is arranged on a volcano or the like, the density of sensor nodes in a portion to be noted such as a crater can be automatically increased even if sensor nodes are uniformly distributed over a wide area.

  As mentioned above, although embodiment and the Example of this invention were described, this invention is not limited only to these embodiment and Example, In the range of the technical thought of this invention, these are combined. It may be used or a part of the configuration may be changed.

DESCRIPTION OF SYMBOLS 1, 2 Sensor node 11, 21 Sensor 12, 22 Communication module 13, 23 Power supply 14, 24 Specific information generation part 15, 25 Control part 16, 26 Moving mechanism 17 Feature calculation part 18 Network control part 4 Communication medium

Claims (7)

Features of the sensor data in the area based on unique information including information indicating respective positions of the plurality of sensor nodes arranged in a predetermined area and sensor data measured in each of the plurality of sensor nodes A feature calculation unit for calculating
Based on the feature, have a, a network control unit for controlling the data measuring operation of said plurality of sensor nodes,
The network control unit is a sensor network control device that calculates a schedule of the sensor node based on the characteristics, and operates the sensor node according to the schedule,
The characteristic is a slope of the sensor data in the area;
The network control unit calculates the schedule so that the first sensor node has a higher operation rate than the second sensor node at a position where the absolute value of the gradient is smaller than the position of the first sensor node. A sensor network control device. Features of the sensor data in the area based on unique information including information indicating respective positions of the plurality of sensor nodes arranged in a predetermined area and sensor data measured in each of the plurality of sensor nodes A feature calculation unit for calculating
A network control unit that controls data measurement operations of the plurality of sensor nodes based on the features;
The network controller, based on the feature, moves the sensor nodes, sensor network controller. The feature calculation unit calculates at least one of the spatial gradient and the temporal gradient of the sensor data in the area as the feature, the sensor network controller according to claim 1 or 2. The feature calculation unit, the scalar field is formed by interpolating the approximation between the sensor data in the area, calculates the gradient in the scalar field as the feature, according to any one of claims 1 3 Sensor network controller. The characteristic is a slope of the sensor data in the area;
The network control unit is configured so that the density of the sensor nodes at the first position is higher than the density of the sensor nodes at the second position where the absolute value of the gradient is smaller than that of the first position. The sensor network control device according to claim 2 , wherein the sensor network is moved. When itself is the cluster head that controls the other sensor nodes, the feature calculation unit calculates the feature, the network control unit is a sensor node for controlling said plurality of sensor nodes, of claims 1 to 5 The sensor network control device according to any one of the above. The sensor network control device according to claim 6 , wherein when the network control unit itself is not the cluster head, the network control unit notifies the other sensor nodes of the unique information and the sensor data measured by the network control unit.

Images