Sensor placement

GT_20Allison and Peter had a meeting with Cerasela Tanasescu and Marc Demange from RMIT University to investigate the potential collaboration opportunities between the RISER project and the GEO-SAFE project. One challenging research area that drew the interest of both project participants is the sensor placement problem. RISER is keen to develop a sensor placement strategy for the RISERnet networks that is maximizes information gained at the lowest communication cost. Communication cost can be defined as the expected number of message transmissions in the networ. In this meeting the team discussed how this problem should be initially formulated. After formulating the problem, the optimization expertise of Cerasela and Marc will help us to obtain an optimal or near-optimal sensor placement plan.

To begin this research, Peter tried an existing sensor placement algorithm (Krause et. al. 2006) which approximately maximizes the mutual information of observing a Gaussian random field. The figure below illustrates the placement of 20 sensors in Olinda. Peter is investigating if other information, such as slope and the fire danger map generated by IBM, can be fused with the mutual information to given the total degree of information captured. Together, the different information sources could be fused to yield to optimal sensor placement.

Tagged with:
Posted in Activites