The potential of using RAIM to detect and exclude multipath effects and NLOS receptions is mentioned in many studies. However, it is difficult to be practically used in the urban canyon due to the lack of satellites. As coming of the multi-GNSS era, the number of satellite increases dramatically. But still, the advance RAIM handles the measurements of different constellations separately. This paper applies the DGNSS technique, which provides the correction including the time-offsets between GPS and other constellations. By the help of the DGNSS correction, all the measurements can be processed together in the RAIM FDE. L1-norm minimization is previous proven to be the most effective RAIM FDE based on simulation results in many studies. However, its evaluation using real data collected in urban canyons is yet to release. This paper, hence, evaluates the RAIM FDE based on the L1-norm minimization in the middle and deep urban canyons. The results indicate the L1-norm minimization can achieve preferable positioning performance when the number of healthy satellite is sufficient.