A Lightweight Detection Method for DoS Attacks in Wireless Sensor Networks Based on Random Forest

Abstract

The invention proposes a random forest-based wireless sensor network DoS attack lightweight detection method, which relates to the field of wireless sensor network information security. The method collects a set of monitoring data for wireless sensor network monitoring nodes at regular intervals and sends them to the base station, and generates detection features (A) and (B) at the base station using the attack characteristics of DoS attacks on sensing data, and sends them Normalized to (A*) and (B*). The training set is combined with the detection feature (A*) using the k-means++ algorithm to obtain normal data and noise data, and the normal data obtained at this stage is added to a certain amount of attack data to train the random forest. For each sample to be detected, combined with the detection feature (A*), first determine whether it is noise. If not, use the random forest model combined with the detection feature (B*) to detect whether a DoS attack occurs. The method is implemented in the base station, and the monitoring node does not need to make additional contributions to DoS attack detection, which saves computing, storage and energy of the node.

Description

technical field [0001] The invention belongs to the field of information security of wireless sensor networks, and relates to a light-weight detection method for DoS attacks of wireless sensor networks based on random forests. Background technique [0002] Wireless sensor network is a special mobile ad hoc network, which can be applied to areas where power supply is difficult, people cannot reach (such as polluted, environment cannot be destroyed or hostile areas) and some temporary occasions (such as natural disasters, fixed The communication network is destroyed), the system can collect real-time environmental information in the monitoring area, such as temperature, humidity, brightness and pressure, etc., and transmit the monitoring data to the base station in a wireless multi-hop manner. However, wireless sensor networks have many unique features, including: (1) network resources are limited, including energy, computing power, communication capabilities, and communicatio...

AI Technical Summary

Problems solved by technology

In addition, the detection method based on clustering will also consume the computing and storage capacity of nodes

Therefore, for sensor nodes with limited computing power, limited storage capacity, and energy constraints, detection methods such as intrusion detection characterized by network traffic will greatly reduce the service life of the network.

Images