Achieving Resilience of Heterogeneous Networks Through Predictive, Formal Analysis

Citation

Qin, Z., Denker, G., Talcott, C., & Venkatasubramanian, N. (2013, April). Achieving resilience of heterogeneous networks through predictive, formal analysis. In Proceedings of the 2nd ACM international conference on High confidence networked systems (pp. 85-92).

Abstract

Rapid development and wide deployment of wireless technologies in recent years have brought an increasing number and variety of services that are accessible directly from mobile terminals via multiple network access technologies (e.g, Ethernet, WiFi, Bluetooth, LTE, etc). A particular traffic flow may go through different kinds of networks, which greatly increases the end-to-end connectivity opportunities. However, the disadvantage of multinetworks is that a failure or change in one network type may affect many traffic flows. Thus, the various networks in a multinetwork cannot be managed in isolation. Rather we need methodologies that analyze the effects of changes in these dynamic and heterogeneous network environments in unison. Traditional network analysis approaches only focus on static network attributes and do not fully consider the impact of failures on quality of services (QoS) across flows. In this paper, we design and implement a “what-if” analysis methodology using formal methods. Our methodology analyzes the impact of failures and changes in heterogeneous networks on QoS of flows. The results of the formal analysis can guide network administrators in their decisions to proactively adapt network configurations to achieve mission or application objectives. We illustrate our methodology with the help of use cases such as incorporating additional nodes in a network or reconfiguring the network due to failure. We compare our results with conventional network configuration approaches and show how our formal methodology provides more effective decision support than conventional network configuration approaches and that it scales better than simulation approaches.


Read more from SRI