Spectral shaping for faster-than-Nyquist signaling

Abstract

Spectral shaping is applied to Faster-Than-Nyquist (FTN) signaling to improve information rates. Under a transmit power constraint, the best shape is an inverse filter but for root raised cosine pulses this requires infinite power at the shaper input. Both input and output power constraints at the shaper are thus proposed as a solution to maximize the information rate while still guaranteeing practicality.