Implementation and Verification of an SDR-Based NB-IoT gNB with Two-Stage Band-Edge Filter for GEO-Relayed NTN Scenario

In designing Narrowband Internet of Things (NB-IoT) base station (gNB) for the 3GPP non-terrestrial networks (NTN) via a GEO relay, it is required to strictly suppress the gNB out-of-band emission and interference. Here we propose a sharp band-edge filter design in terms of two-stage interpolated finite-impulse response (IFIR) structure, and its prototype filter coefficients are optimized to give a good tradeoff between the band-edge attenuation and error vector magnitude (EVM) performance. Next, we present an efficient combined software/hardware implementation of the IFIR filter on a gNB using AD9361-based software defined radio (SDR). In so doing, the 1^st-stage filter is effectively realized in single-instruction multiple-data software code running in real time, while the 2^nd-stage masking filter can be directly mapped to the AD9361 programmable FIR filter. Finally, the proposed gNB filtering design was not only verified through measurements to reveal an excellent band-edge suppression of over 50 dB, but was practically adopted in the world's first NB-IoT NTN experimental trial via Inmarsat satellite.