Online Learning-based Trust Prediction for Reliable and Energy-efficient Transmission

Abstract

Industrial wireless communication networks (IWCNs) have been widely applied in data interaction between field sensors and edge computing units. Nevertheless, harsh industrial environments and malicious attacks cause data loss and delay, which makes it challenging to satisfy the reliability and timeliness requirements of IWCNs. Considering the limited communication energy budget, computation capacity, and multiple unreliable factors, traditional reliable transmission policies become less efficient for IWCNs. To handle these issues, in this paper, we introduce an online learning-based trust model and present a trust-delay aware energy-efficient transmission scheme (TDEETs) to reduce communication energy consumption while satisfying data reliability and control stability constraints. Firstly, a novel trust prediction mechanism based on online extreme learning machine (ELM) with a forgetting factor is proposed. Then, with the aid of low-complexity trust prediction, the optimal path selection strategy and retransmission policy are designed by online solving the optimization problem. Finally, numerical examples demonstrate the effectiveness of the proposed trust prediction mechanism and the transmission performance improvement using TDEETs.