Nowadays, the increasing demand for wireless communication services has led to a significant rise in the density of wireless devices. By using the sparse nature of wireless signals, the sparse techniques detect the signals and interference cancellation (IC) efficiently. The existing Deep Belief Network (DBN) is employed for accurate signal detection, but it has high dimensionality of wireless signals due to the presence of noise and interference. Hence, this research proposes an efficient Transformer-based Deep Neural Network (T-DNN) for signal detection and IC. Initially, the data is gathered from Real-World Wireless Communication dataset, a complete source of Radio Frequency (RF) signals. Then, the collected signals are fed into preprocessing for noise reduction and IC by using Butterworth Frequency Filter (BFF) and Least Mean Square (LMS), respectively. After that, the features are extracted by using the Synchrosqueezing Transform (SST) and then, the proposed T-DNN is introduced for signal detection. From the results, the proposed T-DNN achieved outstanding results in accuracy for Root Mean Square Propagation (RMSProp) and Adam Optimizer (Adam) with 99.94% and 99.96%, respectively, when compared to the existing Bidirectional Long Short-Term (Bi-LSTM).