الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 61 |  |  |
| | | from | 61 | | |
Abstract
Wireless sensor networks WSNs are composed of large number of distributed nodes that coordinate together to perform a common task. Cut-vertex nodes are most likely nodes to fail due to their high rate of energy consumption, and their failures cause the network to split into multiple disjoint segments. WSN can autonomously recover such failures scenario through relocation of some nodes. However such a recovery process remains challenging since tolerating the failed nodes individually is not globally optimal solution and may cause resource conflict.
This thesis presents a centralized Recovery approach that reforms a topology with Increased Robustness against recurrent failure (RIR). RIR tolerates the failure of multiple connectivity-critical nodes through repositioning of non-critical healthy nodes. RIR can handle multiple simultaneous failures of either collocated or scattered nodes. The approach favors substituting a failed node with one with the highest residual energy in order to sustain the network connectivity for the longest time possible.
RIR models the recovery process as a Minimum Cost Flow (MCF) problem to determine the best set of node relocations for repairing the network topology while minimizing the motion overhead of the recovery process. The performance of RIR has been validated through simulation experiments and compared to a competing solution in the literature. The simulation results have confirmed the effectiveness of RIR and demonstrated that it ensures higher residual energy for nodes at critical positions and imposes less travel overhead on the relocated nodes as compared with the competing scheme.