Performance Analysis of Pseudo-Random Radars Identification Using Hidden Markov Model

Nowadays, radar signals are automatically adjusted by computer software to maximize radar performance. These signals are no longer stable and different pulse repetition interval (PRI) values may be assigned for each radar. Therefore, new,
sophisticated (Electronic Surveillance), (Electronic Alert), and (Electronic Intelligence) signal processing algorithms are needed. This is the motivating factor behind the research on using HMM, for Radar System Identification (RSI). This paper, investigates the HMM recognition performance for identifying a pseudorandom PRI radar that randomly selects its PRI value from a number of fixed values. When pseudo-random sequences are considered, it becomes much more difficult to
find an optimal HMM parameters that describe the dynamic behavior of the pseudorandom PRI radar. It will be shown that sub-optimal HMM parameters can still provide good recognition performance. Artificial pseudo-random PRI radar pulses are used to show that applying HMMs can provide adequate signal identification that is far superior to conventional cross-correlation techniques. The simulation results show that, it is necessary to retrain the HMM with an error-corrupted version of the original training sequence to improve the model's robustness.