<p>In the present paper, a channel equalizer of low complexity is developed using the Adaptive Neuro-Fuzzy Inference System (ANFIS) to suit the software-defined radio (SDR) system and performs equalization in a nonlinear and dispersive channel environment. The equalizer proposed is based on the hybrid learning property of ANFIS to approximate nonlinear channel performance at the same time being interpretable and low in a computational cost. The study is performed through systematic analysis of the influence of the complexity of fuzzy rule on equalization performance, behavior of convergence and cost of computation. The simulations of evaluations in a non-minimum phase multipath channel indicate that a well-tuned ANFIS architecture, in particular, the 9-rule model exhibits a better performance with respect to that of conventional linear equalizers, including zero-forcing (ZF) and minimum mean square error (MMSE) in moderate and high signal-to-noise ratio (SNR) regimes. The appropriateness of the proposed approach in resource-constrained SDR platform is confirmed by quantitative complexity analysis such as training time and multiply, accumulate, and operations (MAC). The design decisions are explicitly aimed to be feasible on real time despite the fact the results are derived by use of offline simulating. The suggested ANFIS equalizer, therefore, provides a viable and explainable alternative to computationally expensive deep learning-based equalizers to provide adaptive channel equalization in SDR systems.</p>