<p>The use of directional modulation is one of the prominent and practical solutions for ensuring physical layer security in modern telecommunications systems. In this method, the message signal is modulated by an array of antennas and transmitted in a specific direction toward the legitimate receiver, in such a way that the signal is degraded and not receivable correctly by eavesdroppers in other directions. By employing directional modulation based on a random frequency diverse array, secure communication can be achieved in two-dimensional space, including angle and distance, for the legitimate receiver. The secrecy performance of directional modulation based on a random frequency diverse array is considerably dependent on the information available to the transmitter side about the receiver’s location. In wireless cellular networks where the receiver’s location may change at any moment, there will be inherent errors in estimating the receiver’s location. The occurrence of errors in estimating the angle and distance between the legitimate receiver and the transmitter leads to a significant reduction in the system’s secrecy rate. In this article, a new solution is proposed to increase the robustness against errors in estimating the legitimate receiver’s location by an optimization process based on the minimum mean square error criterion. This solution leads to the improvement of the physical layer security by employing random frequency diverse array directional modulation in the presence of estimation errors. Simulation results indicate an enhancement in the secrecy rate performance of the physical layer in wireless networks.</p>