Towards Low-Complexity Hardware Realization of Adaptive Filters

This paper presents a novel method to implement low-complexity and reconfigurable adaptive filters using processing in memory (PIM) technique. The proposed scheme includes an array of memristive devices, which are programmed by the filter coefficients to realize the desired frequency response of the filter. Additionally, the other fundamental operations of the adaptive filter, such as error calculation and error minimization, are implemented using the proposed memristive-based circuits. As a result, the proposed scheme provides flexibility and parallelism while reducing the computational complexity. Moreover, the idea proposed in this paper can be utilized to implement adaptive filters with arbitrary impulse responses and filter orders.