A Low-Latency and Programmable Band-Pass Filter

Band-Pass filters play a pivotal role in signal processing across domains like wireless communication and biomedical analysis. The traditional analog band-pass filters suffer from limited frequency selectivity and a large number of resonators. On the other hand, digital implementation of band-pass filters results in high complexity especially for high-order filters. This paper presents a new approach for design and implementation of low-latency and reconfigurable band-pass filters by utilization the processing-in-memory (PIM) technique. In this framework, the filter coefficients are programmed into memristive devices, which perform the filtering operation. As a result, high-order bans-pass filters with arbitrary filter impulse responses can be realized while the computational complexity is reduced significantly.