Title of Invention: COMMUNICATION DEVICE, COMMUNICATION METHOD AND COMMUNICATION SYSTEM
Technical Field
[0001] The present invention relates to a communication device that constitutes a wireless multi-hop network. Background Art
[0002] A power failure has been conventionally notified by use of a wireless multi-hop network, when an electricity meter detects interruption of power supply from a primary power source, by transmitting a power-failure notification message through activating a radio function provided in the electricity meter by a secondary power source provided in the electricity meter. In this manner, by activating the radio function by the secondary power source, notification (power failure notification) of occurrence of a power failure is provided to a host device (for example, Patent Literature 1 and Patent Literature 2).
In Patent Literature 1, a wireless unit which can detect a power failure is installed in a transformer and a substation in addition to an electricity meter. Then, when a power failure occurs, by transmitting a power-failure notification message to an electricity meter connected to a system other than a power system where the power failure occurs by using information on the power system, the probability of making the power-failure notification reach a host device is improved.
Further, in Patent Literature 2, when there is no power supply from a primary power source for a fixed period of time, a node makes a judgement that power supply

fails, and transmits a power failure notification to its upper node during a reporting period. At that time, many nodes transmit notifications simultaneously; hence congestion and collision of traffic occur. Therefore, the power failure notification disappears, and in order to prevent this, the reporting period of power-failure notification is divided (reporting window = reporting period/(hop+l)) in accordance with the number of hops from its master unit. Then, by transmitting a power failure notification at a random transmission standby time in the window, a transmission opportunity of nodes far from the master unit (the number of hops is large) is increased. Further, the upper node decreases congestion of notification transmission by merging a power failure notification received from its lower node with its own notification. Citation List Patent Literature [0003] Patent Literature 1: US 2014/0085105 A
Patent Literature 2: US 8970394 B Summary of Invention Technical Problem
[0004] Each of the conventional techniques is a technique to improve an arrival rate of a power failure notification to an upper device, wherein behavior according to a scale (unit count) of a power failure is not considered. As for a power failure notification,
i there is a case wherein a time limit required for the power failure notification to reach an upper device may differ between a case of a power failure in a single node and a case of a power failure in a plurality of nodes. For example, when there is a power failure in a single node, an instant notification completion within several tens of seconds is required, whereas when there is a power failure in a plurality of nodes, a notification
i completion after several minutes is required. However, in the conventional techniques,

it is impossible to judge whether a power failure occurs in a single node or a plurality of nodes; hence, there is a necessity to transmit a notification completion so as to reach an upper device in a short period of time. Therefore, when a power failure occurs in a plurality of nodes, there is a possibility of occurrence of congestion as a result, an excess of a time limit to report a power failure notification, and decrease in an arrive rate of the power failure notification reaching a master unit.
Further, since a node where a power failure does not occur exists in a small scale of power failure, meter reading data (measured value data) of normal power measurement and a power failure notification may collide with each other.
The present invention is performed so as to resolve the problems as described above, and is aimed at adjusting the time required for a power failure notification in accordance with a scale of a power failure occurred. Solution to Problem
[0005] A communication device to constitute a wireless multi-hop mesh network according to one aspect of the present invention, the communication device includes:
a power-failure detection unit to detect a power failure of a power source that is supplied to the communication device;
a power-failure information processing unit to divide, when the power failure is detected by the power-failure detection unit, a power-failure notification period set into a first period and a second period, to generate and output power-failure information to transmit a power-failure notification message by broadcasting in the first period, and to generate and output power-failure information to transmit the power-failure notification message by unicasting in the second period; and
a wireless communication control unit to transmit a frame including the power-failure information to another communication apparatus.

Advantageous Effects of Invention
[0006] When a power failure is detected, a communication device of the present invention divides a power failure notification period determined into a first period and a second period, broadcast a power-failure notification message in the first period, and unicast the power-failure notification message again in the second period; therefore, it is possible to make it unnecessary to transmit of a confirmation communication message to confirm whether the power-failure notification message transmitted in the first period reaches a master unit, and to reduce communication loads. Brief Description of Drawings
[0007] Fig. 1 is a configuration diagram of a system of an automatic meter reading system 600 to be applied a communication device (slave unit) 1 according to a first embodiment of the present invention;
Fig. 2 is a configuration diagram of hardware of the slave unit 1 according to the first embodiment of the present invention;
Fig. 3 is a diagram of a function configuration of the slave unit 1 according to the first embodiment of the present invention;
Fig. 4 is a hardware configuration diagram of a communication device (master unit 21) according to the first embodiment of the present invention;
Fig. 5 is a diagram of a function configuration of the master unit 21 according to the first embodiment of the present invention;
Fig. 6 is a flowchart illustrating an operation of a power-failure detection unit 8 of the slave unit 1 according to the first embodiment of the present invention;
Fig. 7 is a flowchart illustrating an operation of a power-failure information processing unit 9 of the slave unit 1 according to the first embodiment of the present invention;

Fig. 8 is an explanatory diagram of a time period for the slave unit 1 to report a power failure according to the first embodiment of the present invention;
Fig. 9 is a flowchart illustrating an operation of the power-failure information processing unit 9 of the slave unit 1, according to the first embodiment of the present invention;
Fig. 10 is a diagram illustrating a state wherein the slave unit 1 makes a power failure notification reach a master unit 21 by broadcasting, according to the first embodiment of the present invention;
Fig. 11 is a diagram illustrating a state wherein the slave unit 1 makes a power failure notification reach the master unit 21 by unicasting, according to the first embodiment of the present invention;
Fig. 12 is a diagram illustrating a timing for the slave unit 1 to transmit a power-failure notification message in a power-failure notification time period, according to the first embodiment of the present invention;
Fig. 13 is a flowchart illustrating an operation of a power-failure information processing unit 28 of the master unit 21 according to the first embodiment of the present invention;
Fig. 14 is a diagram illustrating a state wherein a plurality of slave units 1 report a power failure by broadcasting, according to the first embodiment of the present invention;
Fig. 15 is a diagram illustrating a state wherein the plurality of slave units 1 report a power failure by unicasting, according to the first embodiment of the present invention;
Fig. 16 is a diagram illustrating a timing for the plurality of slave units 1 to report a power-failure notification message in a power-failure notification time period,

according to the first embodiment of the present invention;
Fig. 17 is a flowchart illustrating an operation in a case wherein a power-failure notification message is received from another slave unit 1, when the slave unit 1 itself has not detected a power failure, according to the first embodiment of the present invention;
Fig. 18 is a flowchart illustrating an operation in a case wherein a power-failure notification message is received from another slave unit 1 as well when the slave unit 1 itself has detected a power failure, according to the first embodiment of the present invention;
Fig. 19 is a flowchart illustrating an operation to transmit a power-failure notification message only in a large-scale period, in a case wherein the slave unit 1 itself has detected a power failure, according to the first embodiment of the present invention; and
Fig. 20 is a flowchart illustrating an operation to transfer a power-failure notification message only in a large-scale period, in a case wherein the slave unit 1 itself has not detected a power failure, according to the first embodiment of the present invention.
Description of Embodiments
[0008] Hereinafter, an embodiment of a communication device according to the present invention will be described in detail based on diagrams. Note that the present invention is not limited to the present embodiment. [0009] First Embodiment
Fig. 1 is a diagram of a system configuration of an automatic meter reading system 600 to be applied a communication device (slave unit) 1 according to a first embodiment of the present invention.

In Fig. 1, the automatic meter reading system 600 of a power meter is composed of a wireless mesh network A 100-a, a wireless mesh network B100-b, a wireless mesh network ClOO-c, a wide area network (WAN) 200 and a server 300.
Additionally, in the wireless mesh network A 100-a, with a master unit (communication device) A21-a as an apex, subordinate slave units 1-a through 1-n are connected via a wireless network having tree structure. Further, a wireless mesh network BlOO-b and a wireless mesh network ClOO-c have similar structures as that of the wireless mesh network A100-a. In the wireless mesh network B100-b, with a master unit B21-b as an apex, slave units are connected via a wireless network having tree structure, and in the wireless mesh network ClOO-c, with a master unit C21-c as an apex, slave units are connected via a wireless network having tree structure. In addition, the master unit A21-a, the master unit B21-b and the master unit C21-c are connected to the server 300 via the WAN 200.
[0010] Fig. 2 is a configuration diagram of hardware of the slave unit 1 according to the first embodiment of the present invention.
In Fig. 2, the slave unit 1 is composed of a central operation unit 2 to perform calculations, a memory 3 to store programs or data, a transceiver 4 and an antenna 5 to perform wireless transmission, a secondary power source 6 to activate the slave unit 1 temporarily when power supply to the slave unit 1 from a primary power source (not shown) is blocked, and a measurement device 7 to measure electric energy (power consumption) used by apparatuses using the primary power source. [0011] Further, Fig 3 is a diagram of a function configuration of the slave unit 1 according to the first embodiment of the present invention.
In Fig. 3, the slave unit 1 is composed of a power-failure detection unit 8, a power-failure information processing unit 9, a measurement information processing unit

10, a mesh network control unit 12, a wireless media access control unit 13 and a wireless communication control unit 14. Further, the power-failure information Drocessing unit 9 and the measurement information processing unit 10 are collectively :alled a message processing unit 11.
Then, the processes of the wireless communication control unit 14 and the wireless media access control unit 13 are performed by the transceiver 4 of Fig. 2. As for the processes performed by the mesh network control unit 12, the measurement information processing unit 10, the power-failure information processing unit 9 and the power-failure detection unit 8, programs stored in the memory 3 are read and executed by the central operation unit 2. Further, measured values of the measurement device 7 df Fig. 2 are input in the measurement information processing unit 10. The power-failure detection unit 8 controls activation of the secondary power source 6 of Fig. 2. [0012] Hereinafter, each configuration of the slave unit 1 will be described.
The power-failure detection unit 8 is connected to the primary power source and the secondary power source of Fig. 2, and the power-failure information processing unit 9. The power-failure detection unit 8 detects a power failure of the primary power source by detecting voltage reduction of the primary power source supplied to the slave unit 1 itself. Then, the secondary power source of Fig. 2 is activated, and additionally, a power-failure start timer is activated; and when power supply of the primary power source is not restored even when the power-failure start timer is timed out, an instruction to process a notification of power-failure (power-failure notification process) is output to the power-failure information processing unit 9. [0013] The power-failure information processing unit 9 is connected to the power-failure detection unit 8 and the mesh network control unit 12. When an

instruction of the power-failure notification process is input from the power-failure detection unit 8, the power-failure information processing unit 9 divides a power-failure notification period set beforehand into a small-scale period (corresponding to a first period of the former half) and a large-scale period (corresponding to a second period of the latter half). The small-scale period is a period in which completion of a notification of a single small-scale power failure is required, which is set beforehand. Then, power-failure information to broadcast a power-failure notification message to peripheral terminals (another slave unit 1 or the master unit 21) in a random transmission standby time within the small-scale period is generated, and the power-failure information generated is output to the mesh network control unit 12. At this time, when an unreceived power-failure notification message is received by means of broadcasting from another slave unit 1 within the small-scale period, the power-failure notification message generated by itself and the power-failure notification message received are merged, and power-failure information to broadcast the power-failure notification message merged is generated. Next, in a random transmission standby time within the large-scale period, in a power-failure notification period after the small-scale period, power-failure information to unicast the power-failure notification message received in the small-scale period to another slave unit 1 or the master unit 21 is generated, and the power-failure information generated is output to the mesh network control unit 12.
[0014] The measurement information processing unit 10 is connected to the primary power source, the measurement device and the mesh network control unit 12. The measurement information processing unit 10 adds the power consumption measured by the measurement device 7 during a normal operation of the slave unit 1 to a measured value message, to generate a measured value message including the power consumption,

and outputs the measured value message generated to the mesh network control unit 12.
Further, besides the measured value message, the measurement information processing unit 10 processes a measured value acquisition message to be transmitted at the time when a measured value of power consumption is acquired, from the master unit 21 to the slave unit 1, a measured value response message to be transmitted at the time when the slave unit 1 responses to the master unit 21 in response to the message, a control message to be transmitted to control measurement of power consumption of the slave unit 1 from the master unit 21, and a response message to be transmitted at the time when the slave unit 1 responses to the master unit 21 in response to the control message.
Note that the explanation will be provided by taking such messages to transmit or receive a measured value of power consumption of the slave unit 1 in a normal operation to or from the master unit 21 are collectively called a normal message. [0015] The message processing unit 11 is a component integrating the power-failure information processing unit 9 and the measurement information processing unit 10, to generate a power-failure notification message and a measured value message including power consumption, and outputs the messages generated to the mesh network control unit 12.
[0016] The mesh network control unit 12 is connected to the power-failure information processing unit 9, the measurement information processing unit 10 and the wireless media access control unit 13. The mesh network control unit 12 generates a frame by adding mesh information to designate a path of the wireless mesh network managed by itself to the power failure information input from the power-failure information processing unit 9, and outputs the frame generated to the wireless media access control unit 13. Further, a frame generated by adding mesh information

managed by itself to the measured value message including the power consumption input from the measurement information processing unit 10 is output to the wireless media access control unit 13.
[0017] The wireless media access control unit 13 is connected to the mesh network control unit 12 and the wireless communication control unit 14. The wireless media access control unit 13 designates broadcasting or unicasting by adding media access control (MAC) information to the frame whereto the mesh information input from the mesh network control unit 12 has been added, and outputs the frame to the wireless communication control unit 14.
[0018] The wireless communication control unit 14 is connected to the wireless media access control unit 13. The wireless communication control unit 14 broadcasts or unicasts the frame input from the wireless media access control unit 13 to a peripheral terminal (another slave unit 1 or the master unit 21). [0019] Next, the master unit 21 will be described.
Fig. 4 is a diagram of a hardware configuration of a communication device (master unit 21) according to the first embodiment of the present invention.
In Fig. 4, the master unit 21 is composed of a central operation unit 22 to perform calculation processing, a memory 23 to store programs and data, a transceiver 24 and an antenna 25 to perform wireless transmission, a secondary power source 26 to activate the master unit 21 temporarily when power supply from the primary power source to the master unit 21 is interrupted, and a WAN interface 27 whereby the master unit 21 communicates with the server 300 via the WAN 200.
[0020] Fig. 5 is a diagram of a function configuration of the master unit 21 according to the first embodiment of the present invention.
In Fig. 5, the master unit 21 is composed of a power-failure information

processing unit 28, a measurement information processing unit 29, a mesh network control unit 31, a wide-area network communication control unit 32, a wireless media access control unit 33 and a wireless communication control unit 34. Further, the power-failure information processing unit 28 and the measurement information processing unit 29 are collectively called a message processing unit 30.
Then, the processes of the wireless communication control unit 34 and the wireless media access control unit 33 are performed by the transceiver 24 of Fig. 4. The processes performed by the mesh network control unit 31, the measurement information processing unit 29 and the power-failure information processing unit 28 are executed by the central operation unit 22 by reading the programs stored in the memory 23 of Fig. 4. The processes of the wide-area network communication control unit 32 are performed by the WAN interface 27 of Fig. 4. [0021] Next, operations of the slave unit 1 will be described.
Fig. 6 is a flowchart illustrating operations of the power-failure detection unit 8 of the slave unit 1 according to the first embodiment of the present invention.
When a power failure occurs in the wireless mesh network A 100-a, the power-failure detection unit 8 of the slave unit 1 detects that power supply from the primary power source stops by detecting voltage reduction of the primary power source supplied to the slave unit 1 itself. When a power failure is detected, the secondary power source 6 of Fig. 2 is activated, and a power-failure judgment timer is started (SI).
Then, it is judged whether the power-failure judgment timer is timed out (whether a time period set has passed) (S2).
[0022] When the power-failure judgment timer is timed out, it is judged that power fails, and an instruction to process a power-failure notification is output to the power-failure information processing unit 9.

The power-failure notification process is a process to generate a power-failure notification message and to transmit power-failure information to a peripheral slave unit 1 or the main unit 21 by the power-failure information processing unit 9, by outputting information on power-failure detection from the power-failure detection unit 8 to the power-failure information processing unit 9.
When the power-failure judgment timer is not timed out, it is judged whether power supply from the primary power source is restored (S3).
When power supply from the primary power source is restored, the power-failure judgment timer is cleared (S4), and the procedure is transited to a normal operation.
When the primary power source is not restored, the procedure is returned to S2, to make a judgment on whether the power-failure judgment timer is timed out. [0023] Fig. 7 is a flowchart illustrating the inspection process S20 according to the present embodiment.
When the instruction of the power-failure notification process is input from the power-failure detection unit 8, the power-failure information processing unit 9 of the slave unit 1 generates a power-failure notification message to the master unit 21 (S21), and divides a power-failure notification period set beforehand into a small-scale period set beforehand, and a large-scale period. The method to divide will be described later.
Then, a timer (transmission standby timer) to measure a random transmission standby time within the small-scale period in the power-failure notification period illustrated Fig. 8 is started (S22).
[0024] Then, it is judged whether an unreceived power-failure notification message is received by broadcasting from another slave unit 1 (S23).
When an unreceived power-failure notification message is received, a

power-failure notification message generated by itself is merged with the power-failure notification message received (S24).
When an unreceived power-failure notification message is not received in S23, or after receiving and merging the unreceived power-failure notification message in S24, it is judged whether the transmission standby timer is timed out (S25). [0025] When the transmission standby timer is timed out, power-failure information including an instruction to transmit a power-failure notification message by broadcasting is generated. Then, the power-failure information generated is output to the mesh network control unit 12, and the power-failure notification message is broadcasted to another slave unit 1 or the master unit 21 via the wireless media access control unit 13 and the wireless communication control unit 14 (S26).
When the transmission standby timer is not timed out, the procedure is returned to the process of S23, and it is judged whether the unreceived power-failure notification message is received by broadcasting from the other slave unit 1.
The slave unit 1 which has detected a power failure broadcasts a power-failure notification message in the small-scale period without judging whether peripheral slave units 1 have detected power failure.
[0026] Further, after the power-failure notification message is broadcasted, by taking a finish time of the small-scale period as a starting point, a random transmission standby timer is started in a large-scale period (S27).
Then, it is judged whether the transmission standby timer is timed out (S28), and when the transmission standby timer is timed out, power-failure information including an instruction to transmit a power-failure notification message by unicasting is generated. Then, the power failure information generated is output to the mesh network control unit 12, and the power-failure notification message is unicasted to

another slave unit 1 or the master unit 21 via the wireless media access control unit 13 and the wireless communication control unit 14 (S29). When the transmission standby timer is not timed out, the procedure is returned to judgment on whether the transmission standby timer in S28 is timed out or not.
[0027] Fig. 8 is an explanatory diagram of a time period for the slave unit 1 to report a power failure according to the first embodiment of the present invention.
When information of power-failure detection 44 to indicate that a power failure has been detected is input from the power-failure detection unit 8, the power-failure information processing unit 9 of the slave unit 1 divides a power-failure notification period 41 set beforehand into a small-scale period 42 and a large-scale period 43.
The small-scale period 42 is a period from the power-failure detection 44 to a time limit (small-scale time notification time limit) 45 to report a single small-scale power failure, which is assumed to be set beforehand. Further, the large-scale period 43 is a period of time from the small-scale time notification time limit 45 to a large-scale time notification time limit 46, and a period of time to report a large-scale power failure being a power failure detected by a plurality of communication terminals. The large-scale period 43 is calculated by: large-scale period 43 = power-failure notification period 41 - small-scale period 42.
For example, when the power-failure notification period 41 is set beforehand to be 200 seconds, and when a small-scale power failure occurs, if the small-scale period 42 within which it is necessary to notify the server 300 via the master unit 21 by multi-hop is 20 seconds, the large-scale period 43 is 180 seconds. Therefore, when a power failure is detected, the slave unit 1 broadcasts a power-failure notification message in a random transmission standby time within 20 seconds, and further unicasts the power-failure notification message 20 seconds later to 180 seconds later.

[0028] Next, an explanation will be provided of an operation when a power failure is not detected by the power-failure detection unit 8 of the slave unit 1, to receive a power-failure notification message from another slave unit 1, and to broadcast the power-failure notification message to another slave unit 1 or the master unit 21.
Fig. 9 is a flowchart illustrating an operation of the power-failure information processing unit 9 of the slave unit 1, according to the first embodiment of the present invention.
The power-failure information processing unit 9 of the slave unit 1 judges whether the power-failure notification message received from another slave unit 1 is a message that has not been received yet (S31).
When the power-failure notification message is the message that has not been received yet in S31, a timer (transfer standby timer) to measure a random transfer standby time is started (S32).
When the power-failure notification message is not the message that has not been received yet in S31, the transfer process is completed.
During the period between when the transfer standby timer is started in S31 and when the transfer standby timer is timed out, it is judged whether an unreceived power-failure notification message has been received from another slave unit 1 by broadcasting (S33).
[0029] When a power-failure notification message is received from the other slave unit 1 in S33, the power-failure notification message received from the other slave unit lis merged into a power-failure notification message waiting to be transferred (S34).
When a power-failure notification message is not received in S33, or after the power-failure notification messages are merged in S34, it is judged whether the transfer standby timer is timed out (S35).

When the transfer standby timer is timed out, a power-failure notification message is broadcasted to peripheral terminals via the mesh network control unit 12, the wireless media access control unit 13 and the wireless communication control unit 14, to another slave unit 1 or the master unit 21 (S36).
When the transfer standby timer is not timed out, the procedure is returned to the process of S33.
When a slave unit 1 that has not detected a power failure receives the power-failure notification message from another slave unit 1 by broadcasting, the power-failure notification message is broadcasted without judging whether a peripheral slave unit 1 has detected a power failure.
Then, when a power-failure notification message is transmitted from the other slave unit 1 by unicasting, transmission destination information is judged, and a power-failure notification message is transmitted to the next slave unit 1 by unicasting. [0030] Next, an explanation will be provided of a state wherein a power-failure notification message is broadcasted by the slave unit 1 to reach the master unit 21, by using a diagram.
Fig. 10 is a diagram illustrating a state wherein the slave unit 1 makes a power-failure notification message reach the master unit 21 by broadcasting, according to the first embodiment of the present invention.
In the wireless mesh network A 100-a, it is assumed that a slave unit 1-i reports power-failure detection to the master unit A21-a.
The slave unit 1-i which has detected a power failure broadcasts a power-failure notification message to peripheral slave units (slave units 1-e, 1-h and 1-j of Fig. 10) by S21 through S26 of Fig. 7. Then, the peripheral slave units (slave units 1-e, 1-h and 1-j) that have received the power-failure notification message transfer the

power-failure notification message by broadcasting by S31 through S36 of Fig. 9. The arrows illustrate a state where the power-failure notification message is transmitted. [0031] The slave unit 1-e whereto the power-failure notification message is transferred transfers the power-failure notification message to peripheral slave units (slave units 1-b, 1-d, 1-h and 1-i) by broadcasting. The slave unit 1-b transfers the power-failure notification message to peripheral slave units (slave units 1-a, 1-d and 1-e) by broadcasting, and transfers the power-failure notification message to the peripheral master unit A21-a and peripheral slave units (slave units 1-b and 1-c) by broadcasting. [0032] Further, a slave unit 1-j that has received the power-failure notification message transfers the power-failure notification message to peripheral slave units (slave units 1 -f, 1-k, 1-m and 1-n) by broadcasting. The slave unit 1-f transfers the power-failure notification message to peripheral slave units (slave units 1-c, 1-k and 1-j) by broadcasting. The slave unit 1-c transfers the power-failure notification message to peripheral slave units (slave units 1 -a and 1 -f) by broadcasting, and the slave unit 1 -a transfers the power-failure notification message to the peripheral master unit A21-a and peripheral slave units (slave units 1-b and 1-c) by broadcasting. [0033] Next, an explanation will be provided of a state wherein a power-failure notification message is unicasted by the slave unit 1 to reach the master unit 21, by using a diagram.
Fig. 11 is a diagram illustrating a state wherein the slave unit 1 makes a power-failure notification message reach the master unit 21 by unicasting, according to the first embodiment of the present invention.
The slave unit 1-i makes a power-failure notification message reach the master unit A21-a by unicasting via other slave units 1-e, 1-b and 1-a, by S27 through S29 of Fig. 7. When the power-failure notification message is received, the slave units 1-e,

1-b and 1-a judge transmission destination information, and unicasts the power-failure notification message to a slave unit 1 being the next transmission destination. [0034] Fig. 12 is a diagram illustrating a timing for the slave unit 1 to transmit a power-failure notification message in a power-failure notification period, according to the first embodiment of the present invention.
When a power failure is detected (power-failure detection 44), the slave unit 1 sets a random transmission standby timer in the small-scale period 42 until the small-scale time notification time limit 45. Then, when the transmission standby timer is timed out, the power-failure notification message is broadcasted to another slave unit 1. When the small-scale period 42 is completed, a random transmission standby timer is set in the large-scale period 43 until the large-scale time notification time limit 46. Then, when the transmission standby timer is timed out, the power-failure notification message is unicasted.
[0035] Fig. 13 is a flowchart illustrating an operation of the power-failure information processing unit 28 of the master unit 21 according to the first embodiment of the present invention.
The master unit 21 that has received the power-failure notification message judges whether the power-failure notification message received from a lower slave unit 1 is an unreceived power-failure notification message (S41).
At this time, the power-failure information processing unit 28 of the master unit 21 receives a power-failure notification message from the lower slave unit 1 by broadcasting, in a first period of the power-failure notification period divided in the lower slave unit 1. Further, the power-failure notification message is received from the lower slave unit 1 in a second period of the power-failure notification period divided in the lower slave unit 1. Irrespective of whether the power-failure notification message

is received by broadcasting or the power-failure notification message is received by unicasting, the master unit 21 judges whether the power-failure notification message received is an unreceived power-failure notification message, and when the power-failure notification message is an unreceived power-failure notification message, the master unit 21 transmits the unreceived power-failure notification message to the server 300 via the wide-area network communication control unit 32 (S42).
When the received power-failure notification message is not an unreceived power-failure notification message, the power-failure notification message is discarded (S43), and the procedure is completed.
Therefore, the server 300 only receives an unreceived power-failure notification message from the master unit 21 via the wide-area network. [0036] Fig. 14 is a diagram illustrating a state wherein a plurality of slave units 1 report a power failure by broadcasting, according to the first embodiment of the present invention.
In the wireless mesh network A 100-a, when the slave units 1-a through 1-n detect a power failure, a power-failure notification message is broadcasted in the small-scale period by the processes of S21 through S26 of Fig. 7 and the processes of S31 through S3 6 of Fig. 9, and notifies the master unit 21 of the power failure.
At this time, when the slave unit 1 detects a power failure by itself, and when a power-failure notification message is received from another slave unit 1 by broadcasting, the slave unit 1 broadcasts the power-failure notification message in the small-scale period without judging whether a peripheral slave unit 1 has detected the power failure. [0037] Fig. 15 is a diagram illustrating a state wherein a plurality of slave units 1 report a power failure by unicasting, according to the first embodiment of the present invention.

In the wireless mesh network A 100-a, when the slave units 1-a through 1-n detect a power failure, a power-failure notification message is unicasted in the large-scale period by the processes of S27 through S29 of Fig. 7, and notifies the master unit 21 of the power failure.
[0038] Fig. 16 is a diagram illustrating a timing for a plurality of slave units 1 to report a power failure notification time period, according to the first embodiment of the present invention.
When the plurality of slave units detect a power failure (power-failure detection 44), each of the plurality of slave units 1 sets a random transmission standby timer in the small-scale period 42 which has been set beforehand. Then, when the random transmission standby timer is timed out, each of the slave units 1 broadcasts a power-failure notification message to another peripheral slave unit 1 or the master unit 21. After that, when the small-scale period 42 ends, a slave unit 1 that has detected the power failure sets a random transmission standby timer in the large-scale period . Then, when the random transmission standby timer is timed out, the power-failure notification message is unicasted to a slave unit 1 indicated in the transmission destination information or the master unit 21, from the slave unit 1 that has detected the power failure.
The arrows of the small-scale period 42 illustrate states where each of a plurality of slave units 1 broadcasts a power-failure notification message each at a random timing. Further, the arrows of the large-scale period 43 illustrate a state wherein each of a plurality of slave units 1 unicasts a power-failure notification message each at a random timing.
[0039] As described above, since a power-failure notification period is divided into a small-scale period and a large-scale period, a power-failure notification message is

broadcasted in the small-scale period, and the power-failure notification message is unicasted in the large-scale period, without judging whether power supply to a peripheral slave unit 1 is stopped, when power supply to a single slave unit is stopped, it is possible to transmit a power-failure notification message to the server 300 within the small-scale notification period.
[0040] Further, since the power-failure notification message is broadcasted, and the plurality of slave units transmit the power-failure notification message to the master unit 21 by transferring the power-failure notification message, redundancy of communication paths is increased, and even when collision of normal messages transmitted from a slave unit 1 where a power failure has not occurred occurs, the possibility of the power-failure notification message to reach the master unit 21 is increased.
[0041 ] Furthermore, when a power failure is detected by a plurality of slave units 1, there is a possibility that when power-failure notification messages are broadcasted, collision between power-failure notification messages may occur. However, since it is broadcasting that is performed, retransmission does not occur in the wireless media access control unit 13; hence, it is possible to prevent congestion in the wireless mesh network 100 by a retransmission process.
[0042] Further, the slave unit 1 that has received a power-failure notification message by broadcasting waits for transfer in a random transfer standby time, and merges the power-failure notification message with another power-failure notification message that is received in the meantime; hence, it is possible to reduce the transmission frequency of power-failure notification messages.
[0043] Further, even when a plurality of slave units 1 detect a power failure, and a power-failure notification message transmitted in the small-scale period does not reach

unicasting in the large-scale period irrespective of whether the power-failure notification message has reached the master unit 21 in the small-scale period; hence, it is possible to transmit the power-failure notification message to the master unit 21 in the power-failure notification period until the large-scale notification period. [0044] Furthermore, the power-failure notification period is divided into the small-scale period and the large-scale period, a power-failure notification message is broadcasted in the small-scale period without judging whether peripheral slave units 1 have detected a power failure, and the power-failure notification message is unicasted again in the large-scale period. Therefore, since it is unnecessary to transmit a confirmation communication message to confirm whether the power-failure notification message transmitted in the small-scale period has reached the master unit 21, it is possible to reduce communication loads.
[0045] Further, it has been described that the slave unit 1 that has detected a power failure waits for transmission at a random time for transmitting a power-failure notification message in the small-scale period; however, it is also applicable to weight the random transmission standby time by the number of hops from the master unit 21. In this case, it may be applicable to perform transmission of a power-failure notification message from a slave unit 1 whereof the number of hops is large, and propagate the power-failure notification message of the slave unit 1 whereof the number of hops is large while being merged, to the master unit 21. [0046] Second Embodiment
In the first embodiment, it has been described the example wherein by dividing a power-failure notification period into a small-scale period and a large-scale period, and by broadcasting a power-failure notification message in the small-scale period, and

unicasting the power-failure notification message in the large-scale period by a slave unit 1 that has detected a power failure, a notification time limit is met in accordance with a scale of a power failure; meanwhile in a second embodiment, an example is described wherein transmission of a message other than a power-failure notification message is suspended when a power failure occurs, and an arrival rate of a power-failure notification message to the master unit 21 is increased.
As an example of messages other than the power-failure notification message, there is a normal message. The normal message is, as described above, a message for the slave unit 1 to transmit a measured value of power consumption to the master unit 21 by a normal operation, including a measured value message, a measured value acquisition message, a measured value response message, a control message and a response message.
[0047] A diagram of a function configuration and a diagram of a hardware configuration of the slave unit 1 are the same as those in the first embodiment.
Therefore, an operation of the slave unit 1 will be described as follows.
First, an explanation is provided of an operation in a case wherein the slave unit 1 itself has not detected a power failure, and the slave unit 1 receives a power-failure notification message from another slave unit 1.
Fig. 17 is a flowchart illustrating an operation in a case wherein a power-failure notification message is received from another slave unit 1, when the slave unit 1 itself has not detect a power failure, according to the first embodiment of the present invention.
When a power-failure notification message is received from another slave unit 1 in a case wherein the power-failure detection unit 8 of the slave unit 1 has not detected a power failure, the power-failure information processing unit 9 of the slave unit 1

judges whether the slave unit 1 itself has suspended transmission of a normal message (S51).
[0048] When the power-failure information processing unit 9 of the slave unit 1 has not been suspending transmission of its own normal message, the power-failure information processing unit 9 configures itself to suspend transmission of the normal message (S52). Then, a suspension timer to suspend transmission of the normal message is started (S53), and a power-failure notification process being a transmission process of a power-failure notification message is performed (S54). The power-failure notification process is the same as the process of Fig. 9.
In judgment of S51, when the slave unit 1 has been suspending transmission of its own normal message, the power-failure notification process of S54 is performed.
Then, it is judged whether the suspension timer to suspend transmission of the normal message set in S53 is timed out (S55).
[0049] When the suspension timer is not timed out, it is judged whether a power-failure notification message has been received from another slave unit 1 (S56).
When a power-failure notification message is received from another slave unit 1 in S56, the power-failure notification process of S54 is performed. The power-failure notification process is the same as the process in Fig. 9, wherein the power-failure notification message received from the other slave unit 1 is merged and broadcasted.
When a power-failure notification message has not been received from another slave unit 1 in S56, the procedure returns to S55, to judge whether the suspension timer is timed out.
When the suspension timer is timed out in S55, transmission suspension of the normal message is released (S57), and the process is completed.

[0050] Next, an explanation is provided of an operation in a case wherein the slave unit 1 itself has detected a power failure, and the slave unit 1 receives a power-failure notification message from another slave unit 1.
Fig. 18 is a flowchart illustrating an operation in a case wherein a power-failure notification message is received also from another slave unit 1, when the slave unit 1 itself detects a power failure, according to the first embodiment of the present invention.
When the power-failure detection unit 8 of the slave unit 1 detects a power failure, the power-failure information processing unit 9 configures itself to suspend transmission of a normal message (S61). Then, a suspension timer to transmit a normal message is started (S62), and a power-failure notification process being a transmission process of a power-failure notification message is performed (S63). The power-failure notification process is the same as that in Fig. 7, wherein a power-failure notification message is broadcasted to another slave unit 1 or the master unit 21 within the small-scale period, and the power-failure notification message is unicasted to another slave unit 1 or the master unit 21 within the large-scale period. [0051 ] Then, it is judged whether the suspension timer to suspend transmission of the normal message is timed out (S64), and when the suspension timer is not timed out, it is judged whether a power-failure notification message has been received (S65).
When the power-failure notification message has been received, the transmission process of the power-failure notification message of S63 is performed. The power-failure notification process is the same as that in Fig. 7.
When the power-failure notification message has not been received, it is judged whether the suspension timer to suspend transmission of the normal message of S64 is timed out.
When the suspension timer to suspend transmission of the normal message is

timed out in S64, transmission suspension of the power-failure message is removed (S66), and the process is completed.
[0052] As described above, in the wireless mesh network 100, when a power failure occurs, by suspending transmission of normal messages other than a power-failure message, it is possible to reduce a possibility of causing the power-failure notification message to disappear by collision of the normal messages other than the power-failure notification message and the power-failure notification message, and to increase an arrival rate of the power-failure notification message to reach the server 300. [0053] Third Embodiment
The second embodiment is aimed at improving the arrival rate of a power-failure notification message by suppressing transmission of normal messages at the time when a power failure occurs; meanwhile, in a third embodiment, it is explained an example wherein when a power-failure notification message is received from another slave unit 1, the power-failure notification message is transmitted to the server 300 by unicasting only in a large-scale period without transmitting the power-failure notification message in a small-scale period.
[0054] When a power-failure notification message is broadcasted in a small-scale period in a case wherein a plurality of slave units 1 detect a power failure, there is a possibility that power-failure notification messages collide with one another, and the power-failure notification messages disappear. Therefore, in a case wherein a slave unit 1 that has detected a power failure by itself receives a power-failure notification message from another slave unit 1, when it is possible to detect that there is no power supply to a plurality of slave units 1, the slave unit 1 transmits the power-failure notification message to the server 300 by unicasting only in a large-scale period without transmitting the power-failure notification message in the small-scale period. In this

manner, by transmitting a power-failure notification message by unicasting only in the large-scale period, there is a possibility that consumption of a secondary power source may be suppressed.
[0055] A diagram of a function configuration and a diagram of a hardware configuration of the slave unit 1 are the same as those of the first embodiment.
Therefore, the operation of the slave unit 1 will be described as follows.
First, the operation of the slave unit 1 when the slave unit 1 detects a power failure will be described.
Fig. 19 is a flowchart illustrating an operation to transmit a power-failure notification message only in a large-scale period, in a case wherein the slave unit 1 itself has detected a power failure, according to the first embodiment of the present invention. When the power-failure detection unit 8 of the slave unit 1 detects a power failure, the power-failure information processing unit 9 generates a power-failure notification message (S71), and starts a random transmission standby timer within a small-scale period (S72).
[0056] Then, before the transmission standby timer is timed out, the power-failure information processing unit 9 judges whether the power-failure notification message is received from another slave unit 1 by broadcasting (S73).
When it is not judged that the power-failure notification message is received from another slave unit 1 by broadcasting in S73, it is judged whether the transmission standby timer is timed out (S74).
When the transmission standby timer is not timed out, the procedure returns to the process of S73.
[0057] When the transmission standby timer is timed out, the power-failure information processing unit 9 broadcasts its own power-failure notification message

(S76), and performs processes of S77 and after S77.
When it is judged that the power-failure notification message is received from the other slave unit by broadcasting in S73, the transmission standby timer is suspended (S75).
Then, by taking a finish time of the small-scale period as a starting point, a random transmission standby timer is started within the large-scale period (S77). [0058] Next, it is judged whether the transmission standby timer is timed out (S78).
When the transmission standby timer is timed out, the power-failure information processing unit 9 unicasts its own power-failure notification message via the mesh network control unit 12, the wireless media access control unit 13 and the wireless communication control unit 14 (S79).
When the transmission standby timer is not timed out, the process of S78 is performed, and time-out of the transmission standby timer is waited. [0059] Next, an operation of the slave unit 1 in a case wherein the slave unit 1 itself has not detected a power failure will be described.
Fig. 20 is a flowchart illustrating an operation to transfer a power-failure notification message only in the large-scale period, in a case wherein the slave unit 1 itself has not detected a power failure, according to the first embodiment of the present invention.
When the slave unit 1 receives a power-failure notification message from another slave unit 1 by broadcasting in a case wherein the power-failure detection unit 8 of the slave unit 1 has not detected a power failure, the power-failure information processing unit 9 starts a random transfer standby timer (S81).
Next, it is judged whether power-failure notification messages are received from a plurality of slave units 1 by broadcasting (S82).

[0060] When power-failure notification messages are not received from a plurality of slave units 1 by broadcasting, it is judged whether the transfer standby timer is timed out (S83).
When the power-failure notification messages are received from the plurality of slave units 1 by broadcasting, the transfer standby timer is suspended (S84), and the process is completed.
Since the process is completed, the power-failure notification message is not transmitted to peripheral slave units 1 by broadcasting. Then, when the power-failure notification message is transmitted from another slave unit 1 by unicasting, transmission destination information is judged and the power-failure notification message is transmitted to the next slave unit 1 by unicasting.
[0061] Then, when the transfer standby timer is timed out in S83, the slave unit 1 that has received the power-failure notification message by broadcasting broadcasts the power-failure notification message to peripheral slave units , 1 via the mesh network control unit 12, the wireless media access control unit 13 and the wireless communication control unit 14 (S85).
Then, when the power-failure notification message is transmitted from another slave unit 1 by unicasting, the power-failure notification message is transmitted to the next slave unit 1 by unicasting.
[0062] As described above, by receiving power-failure notification messages from a plurality of slave units 1, it is possible to detect that a power failure occurs in a plurality of slave units 1. Therefore, transmission of a power-failure notification message is suspended in the small-scale period, and the power-failure notification message is transmitted only in the large-scale period; hence, it is possible to suppress consumption of the secondary power source 6.

[0063] Further, by receiving power-failure notification messages from a plurality of other slave units 1, it is possible to detect a power failure occurs in a plurality of slave units 1. Therefore, transmission of a plurality of power-failure notification messages are suspended in the small-scale period, and the power-failure notification messages are transmitted only in the large-scale period, and it is possible to suppress consumption of a secondary power source. Industrial Applicability
[0064] As described above, a communication device according to the present invention divides a power-failure notification period into a small-scale period and a large-scale period, and broadcasts a power-failure notification message in the small-scale period without judging whether power supply to peripheral slave units 1 is stopped. Then, the power-failure notification message is transmitted again in the large-scale period by unicasting after the small-scale period, and transmission of a confirmation communication message of the power-failure notification message transmitted in the small-scale period is unnecessary; therefore, it is possible to reduce communication loads. Reference Signs List
[0065] 1: communication device (slave unit); 2: central operation unit; 3: memory; 4: transceiver; 5: antenna; 6: secondary power source; 7: measurement device; 8: power-failure detection unit; 9: power-failure information processing unit; 10: measurement information processing unit; 11: message processing unit; 12: mesh network control unit; 13: wireless media access control unit; 14: wireless communication control unit; 21: communication device (master unit); 22: central operation unit; 23: memory; 24: transceiver; 25: antenna; 26: secondary power source; 27: WAN interface; 28: power-failure information processing unit; 29: measurement

information processing unit; 30: message processing unit; 31: mesh network control unit; 32: wide-area network communication control unit; 33: wireless media access control unit; 34: wireless communication control unit; 41: power-failure notification period; 42: small-scale period; 43: large-scale period; 44: power-failure detection; 45: small-scale time notification time limit; 46: large-scale time notification time limit; 100: wireless mesh network; 200: WAN; 300: server; 600: automatic meter reading system

WE CLAIM:
[Claim 1] A communication device to constitute a wireless multi-hop mesh
network, the communication device comprising:
a power-failure detection unit to detect a power failure of a power source that is supplied to the communication device;
a power-failure information processing unit to divide, when the power failure is detected by the power-failure detection unit, a power-failure notification period set into a first period and a second period, to generate and output power-failure information to transmit a power-failure notification message by broadcasting in the first period, and to generate and output power-failure information to transmit the power-failure notification message by unicasting in the second period; and
a wireless communication control unit to transmit a frame including the
power-failure information to another communication apparatus.
[Claim 2] The communication device as defined in claim 1, wherein
the power-failure information processing unit merges, when power-failure
information is received from another communication apparatus in the first period, a
power-failure notification message from the other communication device into the
power-failure notification message of itself.
[Claim 3] The communication device as defined in claim 2, wherein
the power-failure information processing unit merges, when the power-failure
notification message is received from the other communication apparatus in a random
transmission standby time in the first period, the power-failure notification message
received into the power-failure notification message of itself.
[Claim 4] The communication device as defined in claim 1, wherein

the power-failure information processing unit does not output, when a
power-failure notification message is received from another communication apparatus
in a random transmission standby time in the first period, the power-failure information
by broadcasting in the first period.
[Claim 5] The communication device as defined in claim 3 or claim 4, wherein
the power-failure information processing unit weights the random transmission standby time by using the number of hops of the communication device from a master unit.
[Claim 6] The communication device as defined in any one of claim 1 through
claim 5, wherein
the power-failure information processing unit suspends transmission of a
message other than the power-failure notification message.
[Claim 7] The communication device as defined in claim 1, wherein
the power-failure information processing unit generates and outputs, when a
power-failure notification message is received from another communication device by
broadcasting in a case wherein the power failure is not detected by the power-failure
detection unit, power-failure information to merge the power-failure notification
message into the power-failure notification message that has been previously received,
and broadcast the power-failure notification message merged.
[Claim 8] The communication device as defined in claim 7, wherein
the power-failure information processing unit suspends transfer of a message other than the power-failure notification message.
[Claim 9] A communication device to constitute a wireless multi-hop mesh
network, and to transmit a power-failure notification message that is received from a subordinate communication device to a server, the communication device comprising:

a power-failure information processing unit to receive, in a first period into which a power-failure notification period set is divided, power-failure information from the subordinate communication device by broadcasting, and to judge whether the power-failure information has not been received, and in a second period into which the power-failure notification period is divided, to receive power-failure information from a subordinate communication device by unicasting, and to judge whether the power-failure information has not been received, and to output the power-failure information that has not been received; and
a wide-area network communication control unit to transmit the power-failure
information that has not been received, to the server, via a wide-area network.
[Claim 10] A communication method of a communication device to constitute a
wireless multi-hop mesh network, the communication method comprising:
a power-failure detection step to detect a power failure of a power source that is supplied to the communication device;
a power-failure information processing step to divide, when the power failure is detected by the power-failure detection step, a power-failure notification period set into a first period and a second period, to generate and output power-failure information to transmit a power-failure notification message by broadcasting in the first period, and to generate and output power-failure information to transmit the power-failure notification message by unicasting in the second period; and
a wireless communication control unit to transmit a frame including the
power-failure information to another communication apparatus.
[Claim 11] A communication system configured by a server, and communication
devices that are connected to the server via a wide-area network, to constitute a wireless multi-hop mesh network, the communication system comprising:

a first communication device including:
a power-failure detection unit to detect a power failure of a power source that is supplied to the first communication device itself;
a power-failure information processing unit to divide, when the power failure is detected by the power-failure detection unit, a power-failure notification period set into a first period and a second period, to generate and output power-failure information to transmit a power-failure notification message by broadcasting in the first period, and to generate and output power-failure information to transmit the power-failure notification message by unicasting in the second period; and
a wireless communication control unit to transmit a frame including the power-failure information to another communication apparatus, and
a second communication device including:
a power-failure information processing unit to judge whether the power-failure information received from the first communication device subordinate thereto has not been received, and to output the power-failure information that has not been received; and
a wide-area network communication control unit to transmit the power-failure information that has not been received, to the server, via the wide-area network,
wherein the server receives the power-failure information from the second communication device.