Fast-growing urban demand drives increase of production at a global scale. A full understanding of how carbon footprint is driven by socioeconomic factors in local, domestic and international economies is essential. Herein, we develop a cross-boundary carbon tracking approach based on input-output analysis, network control analysis and structural decomposition analysis. Using Beijing as a case study, we quantify both urban embodied and controlled carbon footprints over 1985-2012, and look into how they are impacted by socio-economic factors in local, domestic and foreign regions. We find that the carbon controlled by urban economy from inside accounts for 60% of the total footprint over 1985-2000, while this proportion decreased to 45% in 2012 due to externalization of production supply chains. Carbon intensity and urban consumption strongly compete with each other and together determine the variation trend of the city's consumption-based and controlled carbon footprint. Compared to a consumption-based perspective, this control approach reveals a higher impact of production structure transition on urban carbon footprint, and clearly tracks how carbon emissions are increasingly manipulated by other regions. The local-production-related carbon footprint have decreased by 15-22% over 2000-2012, while meanwhile that from domestic and foreign imports has increased dramatically by 700-960%. Network control approach is able to unveil drivers of carbon emission that are actually regulated by a city as a consequence of its interactions with the rest of global economy.