Variability Index Constant False Alarm Rate (VI-CFAR) for Sonar Target Detection

The sonar background environment is not homogeneous at all times but is often filled with in-homogeneities like multiple targets and reverberation edges. Cell Averaging Constant False Alarm Rate (CA-CFAR) gives optimum performance in a homogeneous environment but its performance degrades in presence of in-homogeneities. Greatest Of CFAR (GO-CFAR) has been designed to control false alarms during reverberation edges while Smallest Of CFAR (SO-CFAR) works well in the presence of multiple targets. But none of these CFARs give optimum performance in both homogeneous and non-homogeneous environment together. Variability Index CFAR (Vl-CFAR) is a composite CFAR approach which adapts itself to CA-CFAR, GO-CFAR or SO-CFAR depending on whether the background environment is homogeneous, has reverberation edges or has multiple targets respectively. This adaptation is based on the values of Variability index (VI) and Mean Ratio (MR) calculated from the leading or lagging CFAR reference windows. Comparison of VI-CFAR with CA-CFAR is done based on simulations in this paper. This is then followed by performance evaluation of VI-CFAR and CA-CFAR with field data.