Abstract
Underwater distributed sensor systems require synchronization to support network connectivity and achieve the system objectives. Synchronization between transmitter and receiver allows transmitted packets to be identified at the receiver, associated with a given source, and permits the data payload contained in them to be successfully demodulated and decoded. This is particularly critical for mobile assets that are spatially dispersed and communicate intermittently, with transmit schedules that are sensor and mission dependent. Examples include oceanographic data collection, pollution monitoring, offshore exploration, military applications, and surveillance. This thesis is concerned with synchronization under an extreme low source power level constraint for scenarios in which an undersea system must not dominate the acoustic landscape and drown out the senses of other natural users of the undersea acoustic environment. Examples of natural users include diverse cetaceans that depend on acoustic signaling for communication, ranging, and foraging. For this reason, low source level underwater acoustic systems are an enabling technology that permits the natural environment to remain unchanged for other users of the undersea space. Establishing transmission packet level synchronization at sufficiently low source levels is challenging due to both the ambient acoustic noise floor and the multi-path ocean acoustic environment, which is complicated by refractive effects associated with the stratified sound speed. Explored here is the performance of detection schemes for multi-path environments at extremely low received signal-to-noise ratios based on broadband multi-path arrivals, combining with limited side information regarding the delay spread. These detectors rely on broadband spread spectrum waveforms with good autocorrelation properties that permit hypothesis testing from reception on a single acoustic hydrophone. This thesis develops the detectors for this scenario and presents probability of detection relative to probability of false alarm for a wide range of time-bandwidth products through a receiver operation characteristic (ROC) curve of the broadband signal transmission. Performance of the detection system is compared relative to increased side information regarding the multi-path profile, so that the observed effect of channel uncertainty can be quantified in terms of a ROC curve. The thesis implements these matched filters and post-process detectors and demonstrates them through Monte Carlo simulation. The slight degradation in ROC performance that attends the small degradation from ideal broadband signal autocorrelation is discussed. Results are presented on two ocean acoustic environments based on acoustic ray tracing: the first is the shallow water, lossy bottom, iso-velocity environment; and the second is a refractive sound speed profile that admits an acoustic duct. This thesis presents reasonable synchronization at in-band SNR levels as low as -16 dB in either of these environments. At -14 dB in-band SNR, false alarm rates on average of one per three month period can be achieved at a detection probability of .60. With additional information regarding the delay spread of the arrivals, the detector improves to an average false alarm rate of once every five months at a near perfect detection probability.