Adaptive algorithm for ternary filtering

Abstract

Despite their major advantage of hardware simplicity, ternary filters have limited usability in practice owing to their unresolved problem of adaptivity. This challenging problem is tackled by introducing an adaptive ternary LMS-like algorithm. Performance assessment using a sinusoidal input distorted by additive white Gaussian noise showed that the proposed algorithm is comparable to the traditional multi-bit Wiener LMS algorithm. This approach is expected to open the door for ternary systems to be ready for replacing multi-bit signal processing systems.