Ground-level ozone (O3) is a primary air pollutant with potential adverse impacts on human health and ecosystems. Aiming to detect O3 concentration and develop efficient O3 sensing materials, sensing behavior of heterogenous cation (Fe3+, Sn4+ and Sb5+) doped In2O3 nanostructures was investigated. The incorporation of these cations modulated the electronic structure of semiconductor oxides, affecting the density of chemisorbed oxygen species and reactive sites. From O3 sensing results, Fe3+ doped In2O3 based sensors featuring saturated resistance curves in O3 gas, demonstrated fast sensing speed and qualified detection threshold (20 ppb). In contrast, Sn4+ and Sb5+ doped counterparts exhibited non-saturated sensing curves, resulting in slower response/recovery speed. As a proof-of-concept, these optimized sensors were integrated as the sensor array. Coupled to the image recognition technique, this sensor array could successfully discriminate O3 and NOx. That is, through the tailored combination of material modulation and sensor array, this study paves a novel approach for highly sensitive and selective O3 detection.