Performance of Dot-product preprocessing for Track-Before-Detect tracking of noise objects

Track-Before-Detect (TBD) algorithms are applied for the tracking of signals below the noise floor. The noise object is the signal that has noise samples only. The processing of such signal using Spatio-Temporal TBD is not possible directly. The preprocessing technique based on the window approach and dot-product calculations emphasis the differences between global and local empirical distributions. The Monte Carlo tests are applied for the analysis of performance for two smoothing coefficients, different width of the window of analysis and different size of the object.