WSAN SIMULTANEOUS FAILURES RECOVERY METHOD

摘要

The WSAN simultaneous failures recovery method ranks each node based on the number of hops to a pre-designated root node in the network. The method identifies some nodes as cluster heads based on the number of their children in the recovery tree. The method assigns a recovery weight and a nearby cluster node to each node. Nearby cluster nodes serve as gateways to other nodes that belong to that cluster. The recovery weight is used to decide which node is better to move in order to achieve lower recovery cost. The recovery method uses the same on-going set of actors to restore connectivity. Simulation results have demonstrated that the recovery method can achieve low recovery cost per failed node in small and large networks. The results have also shown that clustering leads to lower recovery cost if the sub-network needs to re-establish links with the rest of the network.

主权项

1. In a wireless sensor-actor network (WSAN) having a topology defined by a plurality of connected sensor nodes and actor nodes, a computer-implemented WSAN simultaneous failures recovery method, comprising the steps of:
searching the nodes in a breadth first search (BFS) manner, starting at a root node; assigning a rank to the nodes during the BFS of the nodes; creating a recovery tree based on the rank assigned to the nodes, the recovery tree having parent nodes, each of the parent nodes having a set of children nodes farther down the tree; defining roles of the nodes based on levels of the nodes in the recovery tree; assigning a cluster weight and a recovery weight to each of the nodes, the recovery weight being based on a number of hops to the node's closest leaf node in the recovery tree, the recovery weight being used to determine which nodes are to move to restore connectivity, the cluster weight being based on the number of children nodes down in the recovery tree until a cluster is encountered, the cluster weight being used to determine which of the nodes shall be form a new cluster head; checking the clustering weight against a predefined cluster size, the checking being done by each of the nodes, the node identifying itself as a cluster head and sending cluster identification to its neighbours when the clustering weight exceeds the predefined cluster size; monitoring the network for the occurrence of a node failure; and upon determination of a failed node, each of the nodes in the set of children nodes of the failed node F in the recovery tree recursively transmitting its assigned recovery weight and an ID of its parent to the other nodes in the set of children nodes, each of the nodes in the set saving the transmitted assigned recovery weight and the ID, each of the nodes in the set ceasing transmission of the assigned recovery weight and ceasing saving received recovery weights when the saved assigned weight is lower than the node's own assigned weight, thereby leaving the other children nodes of the failed node with lower weight to recover; moving all of the nodes to positions that are at most rank/2 levels away from the failed node, one of the nodes becoming a recovering node taking the place of the failed node in the recovery tree.