Unifying perspective on log-domain filter synthesis

A unifying framework is presented for the in-depth understanding of the seemingly unrelated state-space-based and translinear methods proposed for the synthesis of externally linear but internally nonlinear log-domain filters. The translinear methods exploit either the ‘dynamic translinear principle’ or the Bernoulli Cell dynamics. Light is shed in an insightful manner on key interconnections between Frey's nodal state-space-based synthesis relations, translinear loop-based synthesis conditions and the archetypal linear filter dynamics by considering the form of the state variables used in each case.