GLIDE: A Game Theory and Data-Driven Mimicking Linkage Intrusion Detection for Edge Computing Networks

Abstract

The real-time and high-continuity requirements of the edge computing network gain more and more attention because of its active defence problem, that is, a data-driven complex problem. Due to the dual constraints of the hybrid feature of edge computing networks and the uncertainty of new attack features, implementing active defence measures such as detection, evasion, trap, and control is essential for the security protection of edge computing networks with high real-time and continuity requirements. The basic idea of safe active defence is to make the defence gain more significant than the attack loss. To encounter the new attacks with uncertain features introduced by the ubiquitous transmission network in the edge computing network, this paper investigates the attack behaviour and presents an attack-defence mechanism based on game theory. Based on the idea of dynamic intrusion detection, we utilize the game theory in the field of edge computing network and suggest a data-driven mimicry intrusion detection game model-based technique called GLIDE. The game income of participants and utility computing methods under different deployment strategies are analysed in detail. According to the proof analysis of the Nash equilibrium condition in the model, the contradictory dynamic game relationship is described. Therefore, the optimal deployment strategy of the multiredundancy edge computing terminal intrusion detection service in the edge computing network is obtained by solving the game balance point. The detection probability of the edge computing network for network attacks is improved, and the cost of intrusion detection of the edge computing network is reduced.

Benchmark-Based Reference Model for Evaluating Botnet Detection Tools Driven by Traffic-Flow Analytics

“…Li et al [ 36 ] proposed a data-based mimetic intrusion detection game model called GLIDE as a defense against intrusion attacks for edge computing networks, combining a strategy of multi-redundancy voting algorithms to optimize their intrusion detection rate at edge computing network terminals and game theory. The model was designed to establish a measure to detect and eliminate intrusion attacks in complex environments for edge computing networks because they are characterized by having ambiguous interconnections.…”

Computing Unloading Strategy of Massive Internet of Things Devices Based on Game Theory in Mobile Edge Computing

“…In this case, the base station will encourage CRs to offload their tasks to the nearby CPs through energy-efficient Mathematical Problems in Engineering D2D links. Of course, users can also choose to perform their tasks locally [32]. e available spectrum in this single-cell system is divided into N orthogonal subchannels, and the bandwidth of each subchannel is w. Each D2D link will be assigned an OFDMA subchannel.…”

“…(1) Local Computing. f l r is used to represent the computing power of r (such as the amount of data the device can process per second), and different CRs have different computing power [32]. According to the task's parameter set, the execution time of the r local computing task can be expressed as…”

An Enhanced Cascading Model for E-Commerce Consumer Credit Default Prediction

“…Experiments prove that the classification performance is better than other classification algorithms. Qianmu, Jun, & Shunmei, (2020) constructs WT models based on different forms of decision trees by using the idea of integration. It is found that WT integration models overcome the shortcomings of each sub-model and realize effective early warning of customer default.…”