Hybrid optimization‐based deep neuro‐fuzzy network for designing m‐user multiple‐input multiple‐output interference channel

Abstract

SummaryMultiple‐input multiple‐outputs (MIMOs) have eminent quality in maximizing the throughput of wireless communication models. In MIMO, the antenna arrays can be utilized for fulfilling the needs of 5G by utilizing various spatial signatures of users. Even though 5G communication is imminent, there exist issues, like network interference that arise due to reused frequency spectrum resources. This delving presents an optimized deep model for suppressing interference occurring in the Rayleigh channel in the multiple‐user MIMO (MU‐MIMO) model. Here, an MU‐MIMO model is employed with correlated interference wherein there exist various users around the base station (BS) with several antennas at the transmitter and receiver. Here, a deep neuro‐fuzzy network (DNFN) is used to upgrade the performance of detectors underneath correlated interference. Here, the model comprises zero forcing‐maximum likelihood detection (ZF‐MLD) that assists to generate an initial estimate of broadcasted signals in a particular time slot. The DNFN is used to capture latent correlation among several symbols. Here, the DNFN training is performed using developed autoregressive Henry gas spider monkey optimization (RHGSMO), which is the combination of conditional autoregressive value at risk (CAViaR), Henry gas solubility optimization (HGSO), and spider monkey optimization (SMO). With the lowest symbol error rate (SER), bit error rate (BER), and signal to interference and noise ratio (SINR), the suggested RHGSMO‐based DNFN performed better than existing approaches.