High-Performance Low-Cost DFE using IFIR Filters

In many communication applications, the feedback section in decision feedback equalizers (DFE) must be implemented with a long FIR filter. To reduce computational complexity, we propose to use a cost-effective class of FIR filters termed interpolated FIR (IFIR) filters. In specific, we study three scenarios, namely, i) the approximation of a long FIR filter by an IFIR one using the minimax criterion, ii) the statistical optimization of the IFIR filter assuming a stationary environment, and iii) the adaptive optimization of the IFIR filter assuming a non-stationary environment. Our simulation results show that the IFIR implementation can provide a significant reduction in complexity with a minimal sacrifice in BER performance.