Robust Noncoherent Detection.
JOHNS HOPKINS UNIV BALTIMORE MD DEPT OF ELECTRICAL ENGINEERING
Pagination or Media Count:
In this report the problem of robust detection of non-coherent signals in noise is investigated under the assumption that the noise distribution is unknown, but a member of a known class of distribution functions. This problem is divided into three main categories partially-coherent signals, unknown frequency and phase, and completely random phase. In the partially-coherent problem, two general methods are introduced for the design of robust detectors which guarantee a non-trivial lower bound on the probability of detection and a non-trivial upper bound on the probability of a false alarm. Two receivers are designed for the special case in which the distributions of the noise inphase and quadrature phase components are members of a class of p-point distributions. Simulation results for finite sample size are also given for different distributions. The M-detector method is extended to give a general solution for the second problem.
- Statistics and Probability