Research On User Behavior Based On Traffic Log And Its Application In DHCP

With the decrease of acquisition cost of intelligent terminal device,the improvement of network bandwidth and speed,and the expansion of wireless network coverage capacity,the scale of terminal device in wireless network is increasing explosively,and the pressure of network DHCP server is also increasing.Due to the scale limitation of network IP address pool,the IP addresses that can be allocated to user terminals are limited and increasingly scarce.How to reduce the load of DHCP server as much as possible while ensuring that the available space of IP address pool can meet the sudden increase of terminal devices is important for improving user experience,avoiding IP address exhaustion and saving server resources.In the campus wireless network environment,due to the fixed users of the network,user's online habits and interests will not change greatly in a short period of time.Therefore,it is of great significance to analyze and study user's historical online behavior and dynamically determine the DHCP lease time in the network according to user's session time characteristics.This thesis will focus on the optimization of DHCP lease time in campus network environment.Based on the collected historical user behavior data,this thesis proposes the simulation calculation method of IP address usage and DHCP load,establishes the relationship model of lease time and DHCP load as well as the relationship model of lease time and IP address usage,and proposes the user behavior driven DHCP lease time optimization algorithm according to the established model,so as to satisfy the intelligent terminal with good mobility and transient access mode under the limited IP address resources.The main work of this thesis is as follows:1)This thesis investigates the overall situation of campus network,focuses on the distribution of wireless network access devices,and analyzes the interaction process between user terminal devices and campus network,the authentication mechanism of campus network,the topological structure of campus network and the collection process of relevant data.2)This thesis preprocesses AP log,network traffic and web authentication record,extracts user mobile behavior attribute and user access behavior attribute,and proposes a conditional reclassification algorithm based on classification regression tree algorithm to distinguish user's identity in campus network.Experiments show that the algorithm is effective in accuracy,precision and recall.3)Based on the previous research,the user spatio-temporal attribute,user access attribute and user session duration attribute are mined respectively,and the online user behavior knowledge map is established accordingly.This thesis analyzes the characteristics of user session length from three aspects: user identity,terminal type and building area,and shows the different characteristics of user session length distribution in a graphic way.4)Combined with the network DHCP data,we propose a simulation method based on the change of user status to calculate the IP address usage and DHCP load in the network.On this basis,we model the relationship between the lease time and IP address usage,and between the lease time and DHCP load respectively.At last,we propose a user behavior driven algorithm to optimize the DHCP lease time.According to the historical network data,we dynamically set the network lease time.Experiments show that our algorithm has higher efficiency and stability than the existing DHCP optimization strategy.