In the last decades, the uniform circular array antenna has been used extensively in wireless and satellite communications, especially in 5G. However, it suffers from high side-lobe interference. This paper proposes Back-propagation neural network (BPNN) algorithm to reduce the side-lobe levels. Via the gradient descent algorithm to adjust the weight of each neural network neuron. The result achieved excellent side-lobe suppression. The Back-propagation compares with other adaptive side-lobe suppression techniques (LMS) algorithms and (MVDR) to validate the algorithm’s efficiency. The method has a good learning impact and forms a deep narrow notch at main lobe interference.