Classification of very wide bandwidth acoustic signals

Active sonar systems are being developed to transmit very wide bandwidth signals. This has the potential for improved classification of the returned echo. However, the receiver processing requirements imposed by these very wide bandwidth transmitted signals are increased significantly. A new method based on a modified sequential loglikelihood ratio classifier has been developed that can classify the returned target echo from very wide bandwidth transmitted signals with only a modest increase in the receiver processing requirements. This represents an important classifier implementation for sonar systems because of the reduced requirements and therefore reduced costs. The paper discusses the processing technology that enables the classification of target echoes from very wide bandwidth transmitted signals. The classification performance results for a matched filter preprocessor and a higher-order cumulant preprocessor are compared for several types of acoustic objects