2025ApJS..279...42G

Real-time processing of ultra-wideband signals requires dividing them into several subbands by using a channelizer filter such as a polyphase filter bank. Original channelizers, like the critically sampling polyphase filter bank (CSPFB), face issues such as subband edge attenuation and spectral aliasing. An overlapped polyphase filter bank (OPFB) addresses these challenges, but in existing OPFB structures, the overlap factor is typically an integer, and the output data rate is usually at least twice that of a CSPFB. In practice, some OPFBs could have an overlap factor of less than 2, but their implementation structures are more complex. In this context, this paper proposes a novel OPFB structure based on dual polyphase decomposition (DPD-OPFB). It can be implemented in FPGAs by using existing IP cores, such as the SMAC finite impulse response filters. This makes it more suitable for FPGA implementation in high-speed broadband signal processing compared to original OPFB structures, especially when the overlap factor is not an integer. The algorithm has been implemented using an FPGA and validated through pulsar astronomical observation experiments. Experimental results demonstrate that the DPD-OPFB effectively mitigates issues in original channelizers, thereby preserving the original data more comprehensively. It significantly simplifies the FPGA implementation of the OPFB with wideband output subbands.