Multipath Time Delay Estimation and Registration for a Moving Contact with Interference
Vu, Liet T.
MetadataShow full item record
The objective of this work is to develop effective algorithms for Time Difference of Arrival Estimation and Time Delay path pair association for multiple moving sources in the presence of other interfering sources in a multipath environment using pairs of omnidirectional receivers. The work will focus on the problem of two different sources moving on straight line paths past two receivers. Received signals are simulated based on the assumptions that the source signals are stationary, uncorrelated, zero mean, Gaussian random signals. For simplicity two signal paths are considered in this work; direct and surface bounce paths. Interpolation is used to create the time companding effect at the received signal. Ambient noise at the receiver's filter is also simulated and added to the received signal. An expression is developed for the expected value of the time cumulative average correlation of the two received signals for a single moving source in a multipath environment. This expression is further developed to enhance the multiple moving sources case. The resulting equation is a function of spectral densities of the two source signals. Reasonable simplifying assumptions are used to produce a more manageable expression for the expected value of the time average cross-correlation. The Time Difference of Arrival (TDOA) estimation algorithm is developed using the deskewed short-time correlator technique. The desirable property of this technique is "correlate first, then compensate" which has the advantage of significant computational reduction compared to other techniques . An alternative algorithm, which uses the Select-Correlate-Sum technique , is also proposed. The alternative algorithm is similar to the proposed technique where the main difference is in the sequence of execution. Simulated results indicate both algorithms provide similar performance where the proposed algorithm has the advantage of computational time reduction and simplicity of implementation. The TDOA estimates registration algorithm uses the relationships among the TDOA estimates obtained from the autocorrelation and cross-correlation to classify the TDOA estimates as belonging to one source or the other. Furthermore, the ascending order in time of the TDOA estimates are used to register the TDOA estimates to the pairs of paths that produce each TDOA estimate.