טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentMeirav Almogi-Nadler
SubjectBoost Phase Identification of Theatre Ballistic Missiles
Using Radar Measurements
DepartmentDepartment of Aerospace Engineering
Supervisors Full Professor Ben-Asher Yoseph
Full Professor Oshman Yaakov


Abstract

This research is concerned with the problem of early identification of theatre ballistic missiles (TBMs) during boost phase using RADAR measurements.  Several identification algorithms, with scenarios where two TBMs are present, are developed, based on the use of Wald’s sequential probability ratio test (SPRT), assuming various scenarios.  The use of the SPRT allows meeting specified false alarm and missed detection probabilities, while minimizing identification time.  When the missiles’ dynamic models and launch initial conditions (location and time) are completely known, The SPRT works directly with the raw measurements (target range and two line of sight angles), which are statistically independent. In other scenarios, including the case where the launch location and time are unknown (and need to be estimated), the measurements are statistically dependent, which requires the use of an extended Kalman filter (EKF) along with the SPRT.  The SPRT is then driven by innovations process realizations, which are computed by two EKFs (one for each TBM).

A Monte-Carlo simulation study is used to demonstrate the performance of the proposed algorithms.  Scud-B and Scud-C ballistic missiles are used to simulate the TBMs for the numerical example.  In scenarios where the launch parameters are completely known, the observed false alarm and missed detection probabilities are identical to the pre-specified values.  In the more practical cases, where the launch parameters are either partially or completely unknown, the identification times and the observed error probabilities are shown to depend on proper filter tuning.  The simulation study demonstrates the method’s viability, with mean identification time of about 20 sec and error probabilities smaller than 0.01.