Optimal Backpressure Data Transmission Using Deep Learning

Abstract

In Mobile Ad Hoc Networks (MANETs) Wireless devices are designed to work in co-operative manner. These devices are connected via wireless link to communicate with each other. The MANET has emerged as an important aspect in the wireless domain. This feature is due to the due to its simplicity and self-reconfigurable. In such networks, each mobile node is not only acting as a host but also act as a router which performs the packet forwarding process to other mobile nodes in the network. It is occurred due to the lack of precise congestion indication and hidden terminal problems. In order to handle the congestion issues, the explicit notification is generated as the process of backpressure transmission. The solution must have capability to handle bad channel condition and connectivity failures in unicast transmission. Joint congestion control scheme with scheduling algorithm is improved for dynamic wireless network by changing scheduling scheme with adaptation model. Our end-to-end delay and throughput bounds are in simple and closed forms, and they explicitly quantify the trade-off between throughput and delay of every flow. Furthermore, the per-flow end-to-end delay bound increases linearly with the number of hops that the flow passes through, which is order-optimal with respect to the number of hops. In the proposed system, the optimal congestion control and flow control model is being developed using the deep belief network in deep learning.