The number of antennas on a site increases due to a simultaneous deployment of multi-band and multi-mode radios to combat extremely growing data demand in the network. The correct values of physical parameters of antennas, including azimuth, height and tilt, are essential to optimize the radio frequency (RF) network automatically. It is seen that poor results in RF network optimization are mainly due to incorrect azimuth. The proposed algorithm can estimate the azimuth of an antenna in the field using passive monitoring data from the user equipment. It has been developed to identify the correct value of azimuth without doing the field audit, which can significantly reduce the time for optimization and operational expenditure (OPEX) as well. The field trial reveals that the estimated azimuth value closely matches within ±12°range in comparison to the actual value in the field. Moreover, field results show that the same algorithm is equally applicable for urban and rural morphologies as well. It can also be automated to sanctify the physical site database with proper azimuth values at large level without introducing any kind of human error.