Backup Placement in WSNs in the Network Management Distributed Setting

Sensor nodes are inherently a cheap piece of hardware – due to the common need to use many of them over a large area – and usually contain a small amount of RAM and flash memory, which are insufficient in case of high degree of data sampling. An overloaded sensor can harm the data integrity, or even completely reject incoming messages. The problem gets even worse when data should be received from many nodes, as missing data becomes a more common phenomenon as deployed WSNs grow in scale. As a solution, we consider the Backup Placement problem in networks in the network management distributed setting: Given a network graph G = (V,E), the goal of each vertex v ∈ V is selecting a neighbor, such that the maximum number of vertices in V that select the same vertex is minimized. In cases of an overflow, our Distributed Adaptive Clustering algorithm (D-ACR) reconfigures the network, by adaptively and hierarchically re-clustering parts of it, based on the rate of incoming data packages in order to minimize the energy-consumption, and prevent premature death of nodes. In the network management setting of a grid, we proved that a backup placement of load at most O(1) can be maintained within O(log
²
n) update time, and an O(n log
²
n) energy consumption using the constructed ACR tree.