Simplified technique for bounding error statistics for DFB receivers

A new metric, termed `recovery distance¿¿, is introduced for decision feedback receivsers, and is shown to provide the basis for a simple technique for obtaining bounds on a wide range of error statistics. The error propagation process is modelled using stochastic state transition diagrams with merging of states having the same recovery distance, providing the simplification required to render tractable the derivation of probability generating functions for error random variables of interest. The new technique is illustrated by deriving bounds on error statistics for various signal formats of interest for practical applications.