Adaptive Locally-Optimum Detection Signal Processor and Processing Methods

Reference#: P00288

Previous adaptive nonlinear techniques do not implement optimum signal detection transforms based on the full probability distribution of interference variables, and therefore do not suppress as broad a range of interference types as effectively as does the invention described and claimed herein. Further, the invention also does not need the multiple inputs found in, e.g., directional antenna combining or reference interference subtraction.

The present invention relates generally to a processor and processing methods which provide adaptive locally-optimum detection. Local means that the interference is much stronger than the signal of interest, i.e., the sum of the interference plus signal of interest is near or local to the interference alone. Optimum refers to the best nonlinearity of a given class, e.g., zero-memory amplitude transformations, to suppress the interference.

The ability to adapt is crucial, since the interference statistics/characteristics determine what processing is appropriate (as the examples above illustrated), and those characteristics cannot always be anticipated. If a signal processor fails to adapt, it may easily aggravate, rather than suppress, the impact of the interference.

Hitherto, nonlinear processing has been implemented with only very limited adaptation, such as varying the clipping or hole-punching threshold, A.sub.t. The invention described and claimed herein permits applying a much broader class of nonlinear processes than just hole punchers and limiter/cancellers by implementing a general zero-memory amplitude nonlinearity, ##EQU1## Additionally, the invention implements nonlinearities which have memory and which act on phase components of the input signal

As noted above, linear signal processing methods usually use an iterative approach which can be slow to adapt and may adapt inappropriately. The invention, on the other hand, adapts without iteration by analyzing the interference environment and computing the optimum nonlinearity at each sample time. This method is free of the convergence and stability problems of iterative adaptation.

Patent Status: U.S. patent(s) 5,018,088 issued.

Mr. E. Chalfin
Phone: (443) 778-7473