Allison and Peter had a meeting with Anna and Laura from IBM Research Australia on 3rd February to discuss the potential extensions to collaborative sensor placement work (see above). They discussed possible solutions to the problem of optimizing sensor placement through the fusion of multiple data layers, including the maps of wildfire community impact. Two particular questions being discussed include how the performance of a senor network should be quantified; and what is the theoretical bound of the performance of a given sensor placement. IBM is planning to allocate some staff to work on this problem with Peter.
Following this meeting, Peter, Anna and Laura had a meeting with Marc and Cerasela from the RMIT University on 5th February. Marc and Cerasela have expertise of graph optimization, which can be an important part of formulating and solving our sensor placement problem. Cerasela proposed an idea to consider the sensor placement problem as a covering problem. More explicitly, locations of sensor nodes are determined by a connected graph which covers a topological map. The graph is found by optimizing the number of sensors and robustness of the network communication. The optimization is constrained by that every segment of the map must be covered by at least one sensor node. Peter will work on this problem under this general frame.
