In this study, AWGN generator is implemented on FPGA by using Box-Muller method. Firstly, AWGN generator is designed with floating point in C++ language. Then fixed-point model is developed in C++. The means and stardard deviations of the designs with fixed point and floating point are tested and it is proved that designs have normal distribution. After this step, the fixed-point design is modelled in Verilog HDL. The design in Verilog HDL is matched with C++ model and then the Verilog HDL model is implemented on the Xilinx Virtex-7 FPGA. The design implemented on FPGA has 110 MHz clock frequency, and 7.04 Gbps data speed. The AWGN generator designed in C++ has 39 Mbps data speed and it is confirmed that the AWGN generator implemented on FGPA is working 200 times faster than C++ model.