Kalman prediction-based virtual network experimental platform for smart living

Abstract

Today, the increasing investigation of smart living opens new attack surfaces to intruders. Therefore, developing an emulation platform to test new technologies for current large-scale smart living scenarios is an emerging issue. However, existing network emulation systems suffer from coarse-grained virtual link emulation, including bandwidth, delay, and packet loss rate (PLR). Furthermore, a lack of virtual link emulation accuracy results in an unpredictable emulation fidelity. In order to address this issue, this paper presents Kalman prediction-based experimental platform based on multiple virtualization technologies for smart living to test new technologies and security threats. We introduce the core deployment, link-emulation, and emulator modules in the framework and propose Delay-PLR Kalman Prediction (DPKP) algorithm, which employs Kalman prediction theory based on the packet delay and loss measurement scheme. In particular, the DPKP takes the inherent network traffic errors of the substrate network into consideration, thus improving emulation fidelity. Experiments on the emulation of a BeiDou based satellite network scenario verify the ability of our emulation system to flexibly construct a target virtual network and set the expected bandwidth for each virtual link within 5% errors. In addition, the experimental results demonstrate that the proposed DPKP-based virtual link emulation scheme outperforms the traditional method by up to 21.54% and 78.65% in terms of delay and PLR, respectively.