Development of interference mitigation techniques based artificial neural network for IoT network

Abstract

Nowadays, many sectors and systems have applied the Internet of things (IoT) concept in their works such as operating the real-time service, increasing the performance of the machine to machine (M2M) communication, and avoiding human error. Therefore, the co-channel interference problem is taking into account because more devices are sharing the same communication channel. Also, the rapid increase of IoT devices is the major reason why the interference problem becomes serious. Moreover, overlapping and sharing the same channel can cause the corrupted signal which led to the retransmission of the signal. In this paper, the multi-layer perceptron (MLP) which is an artificial intelligence neural network (ANN) model applied to reduce the adjacent channel interference. These proposed interference mitigation techniques (IMTs) investigate the IoT network at 2.4 GHz. The results are compared between the interference mitigation based on MLP and the traditional minimum mean square error (MMSE) approach. In the MLP model, the fast Fourier transform (FFT) data and the amplitude data are used for the model's input. Moreover, the IoT network topology with the effect of path loss and small-scale fading is generated to make a more realistic system. According to the results, the performance of both IMTs with the MLP model can perform better than the MMSE filter. In addition, the accuracy of our proposed technique is up to 80% in all investigated scenarios.