Exoplanets observed by the {it Kepler} telescope exhibit a bi-modal, radius distribution, which is known as the radius gap. We explore an origin of the radius gap, focusing on multi-planet systems. Our simple theoretical argument predicts that type I planetary migration produces different configurations of protoplanets with different masses and such different configurations can result in two distinguishable populations of small-sized multi-planet systems. We then perform an observational analysis to verify this prediction. In the analysis, multiple Kolmogorov-Smirnov tests are applied to the observed systems, using the statistical measures that are devised to systematically characterize the properties of multi-planet systems. We find with 99.5% confidence that the observed, small-sized multi-planet systems are divided into two distinct populations. The distinction likely originates from different spatial distributions of protoplanets, which are determined by type I migration and subsequently trigger giant impact. We also show that these distinct populations are separated around the radius gap when the gas surface density of protoplanetary disks is $sim 10^2$ g cm$^{-2}$ in the vicinity of the host stars. This work therefore emphasizes the importance of planetary migration and the inner disk properties.