This paper presents SpotFi, an indoor localization system that leverages existing WiFi infrastructure to achieve decimeter-level accuracy without requiring any hardware or firmware changes. SpotFi incorporates two key technical contributions: super-resolution algorithms for accurate angle of arrival (AoA) estimation with only three antennas, and novel filtering and estimation techniques to identify the direct path between the target and access point (AP). The system achieves a median accuracy of 40 cm in multipath-rich indoor environments and is robust to obstacles and multipath. SpotFi's architecture involves collecting CSI and RSSI measurements from APs, estimating AoA and ToF of multipath components, identifying the direct path, and calculating the target's location using these estimates. The system is evaluated using Intel 5300 WiFi cards and demonstrates superior performance compared to state-of-the-art systems like ArrayTrack and Ubicarse.This paper presents SpotFi, an indoor localization system that leverages existing WiFi infrastructure to achieve decimeter-level accuracy without requiring any hardware or firmware changes. SpotFi incorporates two key technical contributions: super-resolution algorithms for accurate angle of arrival (AoA) estimation with only three antennas, and novel filtering and estimation techniques to identify the direct path between the target and access point (AP). The system achieves a median accuracy of 40 cm in multipath-rich indoor environments and is robust to obstacles and multipath. SpotFi's architecture involves collecting CSI and RSSI measurements from APs, estimating AoA and ToF of multipath components, identifying the direct path, and calculating the target's location using these estimates. The system is evaluated using Intel 5300 WiFi cards and demonstrates superior performance compared to state-of-the-art systems like ArrayTrack and Ubicarse.