In this talk, we consider a distributed storage problem in largescale sensor networks, where coding can be utilized. Assume such a network has $n$ sensors, and among them $k<0$.
We propose two new distributed networked storage algorithms, which reduce the previous assumptions about the network topology or geographic locations of the sensing nodes. We show how this problem can be solved by using the random walks and Fountain codes, giving two algorithms to deploy the codes in a decentralized way, and estimating the number of transmissions required to disseminate the data. Also, we demonstrate how each node in the network can estimate the total number of source nodes $k$ and storage nodes $n$.
