Capacity penalty due to ideal zero-forcing decision-feedback equalization

The capacity C_ZF of an idealized zero-forcing decision-feedback equalization (ZF-DFE) system is compared to the capacity C of the underlying Gaussian-noise channel. It is found that, for strictly band-limited channels, the capacity penalty gets vanishingly small as the signal power increases. For nonstrictly band-limited channels, the capacity difference C - C_ZF grows without bound as the signal power increases, and the asymptotic ratio C _ZFC is as low as 93.6% for some channels. It is concluded that ZF-DFE on nonstrictly band-limited channels is a suboptimal, information-lossy equalization technique, even in the limit of infinite signal power