Determining Optimal Zone Radius of Zone Routing Protocol Based on Deep Recurrent Neural Networks in the Next Generation Wireless Backhaul Networks

Abstract

Next-generation wireless networks are becoming more popular and rely on reliable backhaul networks to work properly. Wireless backhaul networks also adopt various innovative technologies to improve capacity and provide more flexible deployments to meet networks' high-quality requirements. One of the essential innovations to maintain the wireless backhaul performance is combining the existing routing protocol technology and the deep learning concept. The concept of deep learning is gaining traction as a powerful way to add intelligence to wireless networks with complex topologies and radio environments. This is because conventional routing protocols do not learn from their previous experiences with various network anomalies. This paper proposed a predictive model of zone radius value using the deep recurrent neural network variant, namely the long short-term memory recurrent neural network (LSTM-RNN) algorithm. Determination of zone radius value conducted by measuring the whole of nodes routing zone using various network performance as input parameters such as Routing Overhead, Energy Consumption, Throughput, and User Usage. Performance measurements such as mean square error (MSE), error distribution histogram, training state, regression, correlation, and time series response are gauged and compared for static and mobile node environments. Results showed that the proposed algorithm can accurately predict zone radius for both environments. However, the accuracy of the proposed algorithm is higher when implemented in a static node environment.

Computational of Concrete Slump Model Based on H2O Deep Learning framework and Bagging to reduce Effects of Noise and Overfitting

“…H2O Deep Learning framework (H2O) will be used to create the model [21] without convolutional layers and maxpooling layers of CNN or RNN. H2O works better than RNN for transactional data because RNN is strong in sequential or time series data [22]. To improve performance of the model, ensemble approaches like Random Forests and Gradient Boosting Regression Trees [23], and Bagging [18], Smoothed Bootstrap Resampling [26], [27] could be used to reduce the negative effect of inherent noise [28] and overfitting.…”