An Optimal Deep Learning CNN Based Classification Approach for Improved Antenna Capacity for MIMO Internet of Things
Abstract
A technique of effective deep learning (DL) based Convolutional neural network is presented for the end-to-end multi-input multiple-output based Internet of Things (IoT) seamless communication system. The main intention of the proposed system is to improve the capacity parameters of the antenna using effective chicken swarm optimization and DLCNN approaches. This approach is mainly based on the selection of transmit antenna scheme (AS-antenna selection) which is in the condition of correlated channel. In contradiction of the conservative DL schemes like single-label multi-class classification, the usage of multi-label concept is preferred in this presented approach DLCNN-based MIMO transmit AS IoT scheme that might decrease the training labels length significantly in case of antenna selection. The optimization is made with the use of Chicken swarm secrecy probability optimization. The classification is made using Cooperative Antenna selection Neural Network classifier. Moreover, applying concept of DL might considerably enhance the accuracy of prediction in the trained model of DLCNN in the large-scale correlated channel conditions of MIMO by means of less data on training. The simulation outcomes show that the presented DLCNN-dependent scheme of AS might be competent of attaining performance of near-optimal capacity in the real time, and a performance is insensitive comparatively to the imperfect CSI effects.