Random Deep Packet Inspection and Dynamic Bandwidth Reallocation Algorithms in Mobile Networks

  • 林 專勝

Student thesis: Doctoral Thesis


Along with the deployment of LTE network there are more and more mobile connect to the LTE network directly According to the Cisco's network report the bandwidth in mobile network is lack for serving so many subscribers especially in the peak hours The urgent priority of mobile operators are increasing the quality of service (QoS) of users According to the previous research there is much bandwidth occupied by the malicious application and botnets For solving this problem the 3GPP had introduced the Traffic Detection Function (TDF) for performing the deep packet inspection (DPI) However if the TDF performs DPI on every packets the cost is very high and the transmission performance is downgrade Therefore we had proposed a random deep packet inspection scheme for operators to find the balance between cost and performance By random deep packet inspection scheme the TDF can only perform DPI on some of packets and verify the application types of sessions The basic unit in the random deep packet inspection scheme is a session the policy control enforced function (PCEF) controls the bandwidth by passing or dropping sessions The random deep packet inspection can improve the performance of DPI in TDF After vacating the bandwidth the bandwidth allocation is another issue for operators How to allocate the bandwidth to users for getting the best QoS? The QoS of users is an import index for measuring the satisfaction of network services And if the deadline assurance service can be done before deadline the QoS will be improved Therefore the bandwidth allocation should consider the service time of deadline assurance services We also develop analytic models and simulations experiments to evaluate the performance of the proposed algorithms In the second part of this dissertation we proposed the dynamic bandwidth reallocation algorithms We introduced how to deploy multiple TCP (MPTCP) in the mobile networks MPTCP allows a mobile equipment connect to Internet by more than two network interfaces simultaneously The demand cellular bandwidth may decrease when a mobile equipment connect to cellular network and WiFi simultaneously Therefore the mobile operators can reallocate the idle bandwidth to other mobile equipment for improving the QoE The Dynamic Bandwidth Reallocation (DBR) algorithm we proposed will allocate the idle bandwidth to mobile equipment equally By the DBR algorithm the miss rate of deadline assurance service had improving than static algorithm And we proposed the QoS-Aware Dynamic Bandwidth Reallocation algorithms Earliest Deadline First (QEDF) allocation algorithm and Smallest download Size First (QSSF) allocation algorithm for improving the performance of DBR The QEDF and QSSF will calculate the demand bandwidth of devices first And then the QEDF will allocate the idle bandwidth to the session with least deadline period And the QSSF will allocate the idle bandwidth to the session with smallest download file size These two algorithms had better performance than the DBR algorithm Based on our study the mobile operators can achieve high QoS in data transmission by vacating the bandwidth from malicious applications and reallocating the bandwidth to real need mobile equipment
Date of Award2018 Aug 25
Original languageEnglish
SupervisorSok-Ian Sou (Supervisor)

Cite this