An Autonomous Method for Aggregating and Repairing Paths in Sensor Networks Inoue, Shinji; Masuda, Noriyuki; Kakuda, Yoshiaki; Eighth International Symposium on Autonomous Decentralized Systems, ISADS '07. 報告：蔡建邦

Outline Introduction Sensor Network Model An Autonomous Aggregating and Repairing Paths Method for Sensor Networks Simulation Experiments Conclusion

Introduction (1/2) A wireless sensor network consists of hundreds or thousands battery-powered sensor nodes generally. If a battery of a sensor node which has to relay sensing data has run out, data sensed by another sensor node X would not be delivered to a sink node even if a battery of sensor node X is still alive.

Introduction (2/2) Some techniques for reducing battery power: Data aggregation Battery-aware routing MAC-layer protocols No methods to repair a path from a sensor node to a sink node have been proposed in case of the path being invalid caused by battery power shortage of a node on the path.

Sensor Network Model (1/1) In the model, there exist three kinds of nodes: sensor node sensing function wireless communication function relay node only a wireless communication function sink node wireless communication function gathering sensing data function sending sensing data to an end user

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (1/12) This method consists of two procedures : The first procedure is for setting initial paths where the initial paths are aggregated. Another procedure is for repairing a broken path by searching a substitution path for the broken path.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (2/12) Sensor nodes send environmental data periodically toward the sink node. Every node knows the maximal number of hops between a sink node and a sensor node. ABCD Sink lower nodechild nodeparent nodeupper node

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (3/12) node 23 is a parent node of node 20 Cmsg C20A to node 23 and Cmsg C20B to node 32. Dead means that node’s battery power is lower than a lower bound value Close means that a path between a node and a sink node is invalid. Two Cmsg’s C11B and C17A are delivered from node 28 to node 20.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (4/12) Step 1: Flooding of PRmsg Flooding of path request message (PRmsg) is caused by a sink node. Payload of a PRmsg includes information about hop count and battery power.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (5/12) Step 2: Relaying of Cmsg toward a sink node Case 1 – Procedure for a sensor node Chooses k nodes in transmission range of the sensor node, and delivers a candidate message (Cmsg) with an ID to each chosen node. Case 2 – Procedure for a relay node Hop count (less is better) The number of receiving Cmsg’s (much is better) Battery power (much is better).

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (6/12) Step 3: Reserving paths from each sensor node to a sink node In order to reserve paths to a sink node from each sensor node, a sink node sends reserve messages (RSRVmsg’s) which includes chosen Cmsg’s ID’s, toward each sensor node along the route.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (7/12) Step 1: Alerting dead alerting When battery power of a node becomes lower than predetermined lower bound value close alerting When a node can detect that any upper nodes of the node is dead

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (8/12) Step 2a: Procedure for a node which asks for repairing a path When a node receives a Dmsg or a CLSmsg from a parent node of the node, the node broadcasts a RRmsg If the node receives no RRAmsg for the waiting time, the node performs a close alerting. The node receives RRAmsg’s, the node chooses one of the RRAmsg’s, and sends a repair reserve message (RRSRVmsg) to the new parent node. Also the node broadcasts an RRDmsg in transmission range of the node in order to inform that it has found the new parent node to upper nodes on the previous path before switching.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (9/12) Step 2b: Procedure for a node which is asked for repairing a path Suppose that node X receives an RRmsg from node A. Case 1: Status of node X is Dead or Close: Node X performs nothing. Case 2: Status of node X is Alive and node A is a parent node of node X: Node X performs nothing.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (10/12) Case 3: Status of node X is Alive and node X is reserved for a path from some sensor nodes to a sink node: Node X sends a RRAmsg to node A. Current battery power A parent node of node X The number of paths which are aggregated at node X. Case 4: Status of node X is Alive and node X is not reserved for a path from any sensor nodes to a sink node: That is, node X does not have its parent node. In order to search a path to a sink node from node X, node X broadcasts an RRmsg in transmission range of node X to find the new parent node.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (11/12) Step 3: Procedure for a node whose status is Close When a close node receives an RRDmsg, the close node changes status field’s value of the close node in transmission range table into Alive, switches the close node’s parent node to a node which sent an RRDmsg, and broadcasts an RRDmsg.

An Autonomous Aggregating and Repairing Paths Method for Sensor Networks (12/12) Sink C A S1S1 D B F E S2S2 Dmsg RRmsgCLSmsg RRmsg RRAmsg RRDmsg RRSRVmsg RRmsg RRDmsg RRSRVmsg RRAmsg RRDmsg RRSRVmsg

Simulation Experiments (1/4)

Simulation Experiments (2/4) The proposed method expands lifetime of the first model 33%

Simulation Experiments (3/4) The proposed method expands lifetime of the first model 63%

Simulation Experiments (4/4) small overhead

Conclusion (1/1) Different from the previous methods, the proposed method for repairing paths are a new technique to expand lifetime of sensor networks in the sense that it is autonomously performed based on local information. Results of the experiments show that the proposed method expands lifetime of sensor networks with small overhead. In future works, we will extend our research results to adaptive role changing between sensor nodes and relay nodes.