Submarine Sensor Environmental Inference ---- 18-075

A new optimal sequential Bayesian algorithm is proposed that processes ambient ocean noise collected on the submarine’s passive sonar sensors into estimates of key aspects of the water sound speed profile (SSP), sea state (SS), and bottom loss (BL). Two modes of operation are proposed. In “tracking” mode, the software tracks changes in the environmental parameters and sends real-time updates to the submarine tactical decision aid (STDA). In “analysis” mode it reprocesses previously recorded element-level data stored on the BQH-9 to produce a through-the-sensor environmental “hindcast.” This provides an enhanced forward-deployed reconstruction capability. Architecturally, the software will run on the compute engine inside the BQH-9. The proposed inferencing capability is made possible by pivoting from the traditional beamforming-based environmental inversion methodology to an information theory based methodology. In Phase I, we intend to use the Cramér-Rao lower bounds and unclassified data to gain insight into which parameters can be estimated by the different submarine sensors. This will lend insight into the design of the environmental tracker.