Online low-sidelobe waveform generator for noise radars based on the graph theory

A common and well-known method of signal processing in noise radars is measuring the cross-correlation between transmitted and received signals. However, it creates a lot of undesirable sidelobes which can mask weak echoes of far targets. There are many methods of masking effect removal based on signal processing at the receiver side. In this study, a method of waveform generation based on the graph theory is presented and its ability to reduce masking effect will be compared with that of radars using purely random waveforms. The method tries to design a graph consisting of nodes each corresponds to a random subsequence. After generating the graph offline, output sequence is produced online by moving from one node to another based on the probability of each edge. The subsequences are designed in a way that the correlation sidelobe levels generated by two subsequences of two connected nodes are reduced. In addition, a combination of the graph-based method and random step frequency modulation is presented to guarantee randomness of the resultant waveforms. The waveform randomness is measured and compared with purely random waveforms. It is shown, by means of extensive computer simulations, the proposed waveforms can produce smaller correlation sidelobes, while preserving randomness characteristics.