As Electronic Warfare (EW) relevant technologies develops, more complex and solid defense algorithm is required to keep the military secrets and message in safe. In the wireless communication system, Low Probability of Interception (LPI) communication algorithms, based on Code Division Multiple Access (CDMA), had been introduced to keep the message safe. Frequency Hopping CDMA (FH-CDMA) and Direct Sequence CDMA (DS-CDMA) are the representative methods of the LPI communication algorithm. FH-CDMA is widely used; however, the system has an expensive hardware configuration requirement due to the implementation complexity. In addition, several studies presented improved LPI communication algorithms with a physical-layer level processing; however, they have limitations of unable to well-hide the original signal. To address these problems, we propose an algorithm, called noise communication, which generates Gaussian distributed random numbers that added to In-phase and Quadrature (I/Q) input data independently to hide the original signal. Through Matlab simulation, we compare the conventional DS-CDMA and the proposed algorithm applied DS-CDMA and evaluate the effect of the algorithm with different configurations of PN sequence length and the effective level of the algorithm. Through the evaluations, we observe the algorithm can provide Signal-to-Noise Ratio (SNR) gain than the conventional DS-CDMA and SNR gain has an inverse relationship with system stability. Moreover, we exploit a Software Defined Radios (SDR) hardware to implement the algorithm for the demonstration of a noise-shaped Radio Frequency (RF) signal.