JP2014169906A - Battery lifetime prediction method of wireless sensor network in state monitor system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a battery lifetime prediction method using information on a total number of packets transmitted within a fixed period in sensor nodes in order to suppress a power load required for battery lifetime prediction as further as possible.SOLUTION: The battery lifetime prediction method of a wireless sensor network in a state monitor system comprises: estimating a packet flow from the number of transmitted packets in wireless sensor nodes; calculating a packet flow probability; computing a remaining battery capacitance of a wireless sensor node in which battery exhaustion first occurs, up to a time point when such a sensor node occurs; then performing update to a packet flow probability in removing the sensor node in which battery exhaustion occurs; computing the remaining battery capacitance of a wireless sensor in which battery exhaustion occurs up to the moment when such a sensor node is next generated; and then repeating the operations until battery exhaustion occurs in all the wireless sensor nodes.

Description

  The present invention relates to a battery life prediction method for a wireless sensor network, and more particularly to a battery life prediction method for a wireless sensor network in a state monitoring system.

  2. Description of the Related Art Conventionally, with the rapid development of information communication technology and sensor technology in recent years, research and development related to sensor networks in which a large number of sensors are arranged and networked in various social environments have been extensively conducted. Among them, a wireless sensor network (WSN) composed of a large number of sensor nodes having a wireless communication function has been attracting attention in recent years (see Patent Document 1 below). The range of application has been expanded to include research in fields such as the environment, the military, and industry.

  The following Patent Document 2 is an example of battery replacement of a wireless sensor, the following Patent Document 3 is an example of a battery-driven node, and the following Patent Document is an example of control for power saving in a multi-hop wireless network. 4 has been proposed.

JP 2000-304531 A JP 2008-039196 A Registered Utility Model No. 3145353 JP 2009-206749 A

Adam Dunkels, Fredrik O (with umlauts) sterlind, Nicolas Tsifters, Zhitao He, "Software-based Online enthorth," 28-32 Announced June 25-26, 2007

  In WSN, the sensor node is generally battery powered. In particular, the WSN for the purpose of condition monitoring is premised on long-term operation, so that the problem of sensor battery depletion occurs, and battery replacement is required. However, since battery replacement requires a large amount of money, it is required to formulate an appropriate battery replacement strategy in order to improve the long-term operation of the WSN. In order to formulate a battery replacement strategy, it is important to predict battery life dynamically because the battery power load of each sensor node is affected by dynamically changing installation environments and routing protocols. It becomes a difficult task.

Usually, in order to dynamically predict the battery life, it is necessary to acquire route information. However, constant acquisition of route information increases traffic and increases the power load of the entire network.
Therefore, in view of the above situation, the present invention provides a battery life prediction method using total packet number information transmitted in a certain period in each sensor node in order to suppress the power load necessary for battery life prediction as much as possible. For the purpose.

In order to achieve the above object, the present invention provides
[1] In a battery life prediction method of a wireless sensor network in a state monitoring system, a packet flow amount is estimated from the number of transmitted packets of a wireless sensor node, a packet flow probability is obtained, and a battery is firstly determined based on the estimated packet flow probability. When a depleted wireless sensor node occurs, calculate the remaining battery level of the wireless sensor node up to that point, and then update to the packet flow probability when the depleted wireless sensor node is removed Then, based on the updated packet flow probability, when a wireless sensor node that has run out of battery occurs, the remaining battery level of the wireless sensor node up to that point is calculated. It is characterized by repeating operations until all wireless sensor nodes are exhausted. .

  [2] The battery life prediction method for a wireless sensor network in the state monitoring system according to [1] above, wherein the wireless sensor node is replaced according to the state of each step. .

According to the present invention, in order to suppress the power load necessary for battery life prediction as much as possible, it is possible to perform battery life prediction using the total number of packets information transmitted in a certain period in each sensor node.
Further, the battery life manager can reflect the battery life prediction described above in the battery replacement guideline in consideration of each step.

It is a figure which shows the structural example of the network which shows the Example of this invention. It is a figure which shows the aspect of the battery life prediction of the network which shows the Example of this invention. It is explanatory drawing of the input / output information and restrictions for estimating the other packet flow amount of this invention.

  The battery life prediction method of the wireless sensor network in the state monitoring system estimates the packet flow amount from the number of transmitted packets of the wireless sensor node, obtains the packet flow probability, and first causes the battery depletion based on the estimated packet flow probability. If a sensor node is generated, the remaining battery level of the wireless sensor node until that time is calculated, and then updated to the packet flow probability when the sensor node that has run out of battery is removed, and the updated Next, based on the packet flow probability, if a sensor node that has run out of battery occurs, calculate the remaining battery level of the wireless sensor node up to that point, then repeat the above operation to Repeat until the wireless sensor nodes are exhausted.

Hereinafter, embodiments of the present invention will be described in detail.
FIG. 1 is a diagram showing a configuration example of a network showing an embodiment of the present invention, FIG. 2 is a diagram showing a battery life prediction mode of the network, FIG. 2 (a) is step 1 and FIG. 2 (b). FIG. 2 is a diagram showing Step 2 and FIG.
Here, first, terms are defined.

(1) Sensor node: It is assumed that it is battery-driven and has a sensor function, a wireless function, and a calculation function. Further, it is not always necessary to perform sensing, and it is possible to operate as a relay node that only transmits data.
(2) Gateway (GW): It is driven by AC power, and gathers information from sensor nodes.

(3) Node: When the sensor node and the gateway are not distinguished, they are simply called nodes.
(4) Edge: A communication path between nodes is called an edge, and the edge is represented by a directed side and indicates that data can be transferred from the start point to the end point.
(5) Path: A node sequence in which the end point is a gateway and all others are sensor nodes, and when there is an edge in all adjacent nodes included in this node sequence, this node sequence is called a path.

The wireless sensor network in the present invention assumes a network composed of sensor nodes and gateways. Further, when data is transmitted along the path, no packet loss or loop occurs.
An example of the network configuration targeted here is shown in FIG. In this figure, sensor node A indicates that data can be transmitted to sensor node B and sensor node C.

  When the function of the sensor node stops due to a cause such as battery exhaustion, the network topology changes. In addition, the change in the network topology affects the power load at each sensor node. For example, in FIG. 1, when the sensor node A is transmitting data to the sensor node B and the sensor node C, when the sensor node B stops, the data that has passed through the sensor node B until now passes through the sensor node C. As a result, the power load of the sensor node C increases.

Normally, route information is not included in a packet transmitted from a sensor node, and in an ad hoc network, it is difficult to estimate to which sensor node a power load is concentrated. On the other hand, as described above, constant acquisition of route information increases the power load of the entire network.
Therefore, in the present invention, each sensor node counts the number of packets transmitted in a certain period, and uses this information to predict the battery life of each sensor node in consideration of the effect of sensor node stop due to battery depletion. Propose. In the present invention, since battery consumption is particularly large when packets are transmitted at each sensor node, the battery life can be predicted by managing the number of packets transmitted.

Here, an outline of the battery life prediction method will be described below.
First, the packet flow amount of each edge is estimated from the number of packets of each sensor node transmitted in each sensor node for a certain period. Next, based on the estimated amount, the power consumption of each sensor node is estimated, and the battery life is predicted from the power consumption. The battery life prediction procedure will be described later.

First, input / output information and constraint conditions for estimating the packet flow amount will be described.
(1) Input:
-Node set-Edge set-Number of transmitted packets in each sensor node for a certain period-Number of sensing data in each sensor node for a certain period (2) Restrictions:
In all sensor nodes, the sum of the number of sensing data and the number of received packets for a certain period is equal to the number of transmitted packets.

(3) Output:
-Packet flow amount at each edge Next, a battery life prediction procedure will be described with reference to FIGS.
First, the number of packet transmissions at each sensor node is acquired, and the remaining battery level of each sensor node at the current time t _c is calculated [FIGS. 2 (a) and 3 (a)]. The number on the sensor node is the number of packet transmissions of each sensor node.

Next, the packet flow amount is estimated from the number of transmitted packets (packet flow probability), and based on the estimated packet flow probability, the sensor node at which the battery is depleted first and its time t ₁ are predicted, and the time t ₁ The remaining battery levels of the other sensor nodes are calculated [Fig. 2 (b), Fig. 3 (b)]. The numbers are the packet flow amounts on each side.
Next, the packet flow probability is updated by calculating the packet flow amount in the network from which the sensor node depleted of the battery is removed, and then the sensor node depleted of the battery and its time t based on the updated packet flow probability. ₂ is predicted, and the remaining battery levels of the other sensor nodes at time t ₂ are calculated [FIG. 2 (c), FIG. 3 (c)]. Here, when the node is stopped, the topology changes and the load on other nodes changes. The numbers on the sensor node are the number of packet transmissions, and the others are the packet flow amounts on each side.

This operation is repeated until the estimation of the time t ₃ (FIG. 2C) at which the batteries of all sensor nodes are depleted is obtained.
In equipment state monitoring (for example, grasping of age-related deterioration) assumed to be long-term operation, the battery drain of the sensor node becomes a problem, so an efficient battery replacement strategy is required. Therefore, the present invention has proposed a method for estimating the battery life necessary for formulating a battery replacement strategy.

Hereinafter, the battery life prediction method of the wireless sensor network in the state monitoring of the present invention will be described.
The overall flow is as follows:
(1) Input given parameters.
(2) Solve the problem (P) with a mathematical optimization solver, estimate the flow amount of each edge, and calculate the packet flow probability.

(3) Based on the estimated flow amount, the remaining battery level of each sensor node is calculated.
Each item is described in detail below.
[1] Input parameters-Sensor communication possible side-Sensing frequency of each sensor node-Battery capacity [2] Next, formulation of problem (P) will be described.

First, the definition of symbols will be described.
・ Graph G (V, A)
・ Sensor set S
・ Relay set R
・ Gateway set T
Node set V = S∪R∪T
Edge set A = {(i, j) | i∈S∪R, j∈V, i ≠ j, packet can be transmitted from i to j}
・ Sensing frequency s _i (≧ 0, ∈i∈S)
Each sensor node transmits one packet for each sensing.

Number of transmitted packets Pi at node i ∈ S∪R (≧ 0, ∀i∈S∪R)

<Decision variable> Packet flow amount between nodes x _ij [≧ 0, ∀ (i, j) ∈A] <Restriction condition>

Packet flow probability The packet flow probability on each side is determined to be equal to or less than Pr (i, j).
Pr (i, j) = (x _ij / P _i ) [∀ (i, j) ∈A]
[3] Next, the battery remaining capacity curve will be described.

[A] Input (1) Parameters related to networking ・ Current time t _c (see FIG. 2)
Node set V _c in, the current time t _c
• The current put in time t _c edge set A _c
-Graph G '(V', A ') at time t
The weight Pr (i, j) of the arc (i, j) in the graph G ′ (V ′, A ′)
・ Sensing frequency ・ Power capacity C _i , i∈S∪R of node i∈S∪R
-Number of transmitted packets p _i (≧ 0), i∈S∪R
And time t _c cumulative transmission packets Received in a cumulative transmission packet transmission number and time t _c in (2) necessary power calculating parameter supply voltage H
-Packet size-Current consumption of the processor (per packet with the packet size used this time) _Im
・ Product of the operating time per packet and the number of packets processed during processor operation (others omitted) t _m
・ Processor current consumption during standby I _l
At the time of transmission of-radio consumption current I _t
At the time of reception of-radio consumption current I _r
・ Current consumption of other components (sensors, LEDs, etc.) consisting of i pieces I _Ci
[B] Battery remaining capacity curve of each node in the period from the output time t _C until the battery of all the nodes is depleted [C] Calculation algorithm Node survival rate for each confirmation period (the number of transmitted packets of each node is sent) At each timing):

Step S1 V ′: = V _C , A ′: = A _C , t: = t _C. Also, the battery remaining capacity curve is initialized.
Step S2 From the cumulative number of transmitted / received packets at time t and the power consumption calculation formula (1), the remaining battery level K _i of the node iεV ′ \ T is calculated by the following formula.
K _i = C _i − [power consumption of node i∈V ′ \ T defined by equation (1)), i∈V ′ \ T
At this time, if K _i ≦ 0, V ′: = V ′ \ {i ′}, all arcs connected from A to {i ′} are removed, and G ′ is updated. In addition, the remaining battery level curve is updated with the remaining battery level of the node i ′ at time t being zero. If an unconnected node is generated as a result of this operation, G ′ and the battery remaining capacity curve are updated by the same operation as described above. At this time, if V ′ is not 0, the process is terminated.

Step S3: The weight of G ′ is updated using the output solution of the problem (P).
Step S4 Considering the graph G ′, the sensing frequency of each sensor, the remaining power of each node and the number of received packets, the node i ^{* at} which battery depletion first occurs and its time t ^* are obtained. The transition of the battery remaining amount from time t to time t ^* calculated in this process is added to the battery remaining amount curve. V ′; = V ′ \ {i ^* }, remove all arcs connected from A ′ to {i ^* }. At this time, if V ′ is not 0, the process is terminated.

Step S5: Determine the cumulative number of transmission / reception packets of node iεV ′ \ T at time t ^*, and return to Step S2 as t = t ^* (this is already calculated from Step S4 and is a formal description). .
[D] Calculation of power consumption in each node (see Non-Patent Document 1)
_{E C / H = I m t} m + I l t l + I r t r + ΣI Ci t Ci ... (1)
However, E _C is power consumption.

  In addition, this invention is not limited to the said Example, Based on the meaning of this invention, a various deformation | transformation is possible and these are not excluded from the scope of the present invention.

  The battery life prediction method of the wireless sensor network in the state monitoring system of the present invention can be used as a battery life prediction method of the wireless sensor network that calculates the remaining battery level of the wireless sensor based on the number of transmission / reception of each sensor node of the wireless sensor. It is.

A, B, C, D Sensor node GW Gateway t _C Current time t ₁ Battery depletion time of sensor node C t ₂ Battery depletion time of sensor node A t ₃ Battery depletion time of all sensor nodes

Claims (2)

  Estimate the packet flow amount from the number of transmitted packets of the wireless sensor node, determine the packet flow probability, and if a sensor node that has run out of battery first occurs based on the estimated packet flow probability, up to that point The remaining battery level of the wireless sensor node is calculated, and then updated to the packet flow probability when the sensor node depleted of the battery is removed, and then the battery is depleted based on the updated packet flow probability. When a sensor node occurs, the remaining battery level of the wireless sensor node up to that point is calculated, and then the above operation is repeated until all wireless sensor nodes are exhausted. A battery life prediction method for a wireless sensor network in a monitoring system.   The wireless sensor network battery life prediction method in the state monitoring system according to claim 1, wherein the wireless sensor node is replaced according to the state of each step. A battery life prediction method for a wireless sensor network.

Images